creating and managing a high-performance knowledge-sharing network: the toyota case

Strategic Management JournalStrat. Mgmt. J.,21: 345–367 (2000)

CREATING AND MANAGING A HIGH-PERFORMANCE KNOWLEDGE-SHARING NETWORK:THE TOYOTA CASE

JEFFREY H. DYER1* and KENTARO NOBEOKA2

1Marriott School of Management, Brigham Young University, Provo, Utah, U.S.A.2Research Institute for Economics and Business Administration, Kobe University,Kobe, Japan

Previous research suggests that knowledge diffusion occurs more quickly within Toyota’sproduction network than in competing automaker networks. In this paper we examine the ‘blackbox’ of knowledge sharing within Toyota’s network and demonstrate that Toyota’s ability toeffectively create and managenetwork-level knowledge-sharing processesat least partiallyexplains the relative productivity advantages enjoyed by Toyota and its suppliers. We provideevidence that suppliers do learn more quickly after participating in Toyota’s knowledge-sharingnetwork. Toyota’s network has solved three fundamental dilemmas with regard to knowledgesharing by devising methods to (1) motivate members to participate and openly share valuableknowledge (while preventing undesirable spillovers to competitors), (2) prevent free riders, and(3) reduce the costs associated with finding and accessing different types of valuable knowledge.Toyota has done this by creating a strongnetwork identity with rules for participation andentry into the network. Most importantly,production knowledge is viewed as the property ofthe network.Toyota’s highly interconnected, strong tie network has established a variety ofinstitutionalized routines that facilitate multidirectional knowledge flows among suppliers. Ourstudy suggests that the notion of a dynamic learning capability that creates competitiveadvantage needs to be extended beyond firm boundaries. Indeed, if the network can create astrong identity and coordinating rules, then it will be superior to a firm as an organizationalform at creating and recombining knowledge due to the diversity of knowledge that resideswithin a network.Copyright 2000 John Wiley & Sons, Ltd.

The ideas behind the Toyota Production Systemhave basically diffused and are understood byour competitors. But the know-how regardinghow to implement it in specific factories andcontexts has not. I believe that Toyota Groupcompanies are better at implementing the ongoingkaizenactivities associated with the Toyota Pro-duction system . . . I think we are better at learn-ing. (Michio Tanaka, General Manager of Inter-national Purchasing, Toyota Corp., June 10,1996)

Recently both executives and academics have

Key words: knowledge management; learning;networks*Correspondence to: Professor J. Dyer, Marriott School ofManagement, Brigham Young University, 790 Tanner Build-ing, Provo, UT 84602, U.S.A.

CCC 0143–2095/2000/030345–23 $17.50Copyright 2000 John Wiley & Sons, Ltd.

identified organizational learning as perhapsthekey factor in achieving sustainable competitiveadvantage. Indeed, the academic literature onorganizational learning as a source of competitiveadvantage has been expanding in unprecedentedfashion (Cohen and Levinthal, 1990; Teece, Pis-ano, and Shuen, 1997; Kogut and Zander, 1992;Spender, 1996; Grant, 1996). For example, Teeceet al. (1997) have proposed a ‘dynamic capabili-ties’ approach to firm-level advantage suggestingthat a firm’s ability to continually learn, adapt,and upgrade its capabilities is key to competitivesuccess. Other scholars have argued for a ‘knowl-edge-based view of the firm’ suggesting that thekey role of the firm is in creating, storing, andapplying knowledge (Kogut and Zander, 1992;Conner and Prahalad, 1996; Grant, 1996) rather

346 J. H. Dyer and K. Nobeoka

than simply reducing transaction costs (Coase,1937; Williamson, 1985).

Although the focus of the organizational learn-ing literature is on the individual firm, there isincreasing evidence which suggests that a ‘net-work’ of firms may be a critical, but less under-stood, unit of analysis for understanding firm-level learning (Powell, Koput, and Smith-Doerr,1996; Dyer and Singh, 1998). Various scholarshave recognized thatinter-organizational learningis critical to competitive success, noting thatorganizations learn by collaborating with otherfirms as well as by observing and importing theirpractices (March and Simon, 1958: 188; Powellet al., 1996; Levinson and Asahi, 1996). Forexample, von Hippel (1988) found that a firm’scustomers and suppliers were its primary sourcesof innovative ideas. He argues that a productionnetwork with superior knowledge transfer mecha-nisms among users, suppliers, and manufacturerswill be able to ‘out-innovate’ networks with lesseffective knowledge-sharing routines.

In a similar vein, Powellet al. (1996) foundthat in the biotechnology industry the locus ofinnovation was the network, not the individualfirm. Patents were typically filed by a large num-ber of individuals working for different organi-zations, including biotech firms, pharmaceuticalcompanies, and universities. Powellet al. (1996)argue that biotech firms who are unable to create(or position themselves in) ‘learning networks’are at a competitive disadvantage. Although Pow-ell et al. (1996) make the case for the ‘network’as the appropriate unit of analysis for learningand innovation in biotechnology, they do notaddress the specific routines involved in the inter-organizational learning process.

In fact, although there has been considerabletheoretical discussion on organizational and net-work learning, there is little empirical researchto date. As Grant (1996: 384) recently observed,‘detailed study of the operation of organizational[learning] routines is limited. Further progress iscritically dependent upon closer observation ofthe processes through which tacit knowledge istransferred.’ How exactly are learning networkscreated? What are the structures and processesthat allow for effective interorganizational learn-ing? How do networks solve problems inherentin knowledge sharing, such as free rider problemsand preventing undesirable spillovers? To answerthese questions, we must empirically examine

Copyright 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J.,21: 345–367 (2000)

those networks that have demonstrated a parti-cular ability at knowledge transfers.

The automotive industry offers an interestingopportunity to examine interorganizational learn-ing. Automobiles are developed and manufacturedby OEMs and their supplier networks, who pro-duce as much as 70 percent of the value of avehicle. Consequently, the cost and quality of avehicle are a function of the productivity of anetwork of firms working in collaboration.Research to date suggests that Japanese automo-tive networks, and Toyota’s in particular, havebeen superior at transferring productivity-enhancing knowledge throughout the network(Nishiguchi, 1994; Lieberman, 1994; Liebermanand Asaba, 1997). For example, Lieberman andAsaba (1997) examined the diffusion of leanproduction practices as measured by labor produc-tivity improvements and inventory reductions byautomakers and their suppliers from 1965 to1990. They found that in Japan labor productivity(as measured by value added per employee)increased steadily and consistently for both auto-makers and suppliers throughout the time period(see Figure 1; we have added data on Toyota).In contrast, the productivity of U.S. automakersand suppliers was stagnant until the mid-1980swhen U.S. automaker productivity began toincrease. These productivity increases began whenJapanese automakers began establishing trans-plants in the United States (U.S. figures includetransplants) and when U.S. automakers were seri-ously attempting to imitate ‘lean’ production prac-tices. However, these productivity improvementsdid not spillover to U.S. suppliers, whose produc-tivity remained stagnant until roughly 1990.1 Whyis this the case?

We submit that the answer has to do withthe fact that Toyota and other leading Japaneseautomakers (notably Honda) have developedbilateral and multilateral knowledge-sharing rou-tines with suppliers that result in superior inter-organizational or network-level learning. Toyota,in particular, is widely recognized as a leaderin continuous learning and improvement. Thereare a number of reasons to examine Toyota’spractices in greater detail. First, Toyota is the

1Lieberman and Asaba’s analysis on inventory reductions overtime mirror the labor productivity data. Moreover, Sako(1997) replicated Lieberman’s study in the United Kingdomwith virtually identical results.

Knowledge-Sharing Network 347

Figure 1. Japanese and US labor productivity (1968-1992)

largest Japanese company and is regularly votedby Japanese executives as the best-managed andthe most respected Japanese company. Second,the most rapid diffusion of lean production tech-niques (e.g., kanban, inventory reduction) hasoccurred within Toyota and its suppliers. Lieber-man et al. (1997) found a significant positivecorrelation between membership in Toyota’ssupplier association and supplier productivity.In summary, Toyota’s ‘network’ appears to behighly effective at facilitating interfirm knowl-edge transfers and may be a model for thefuture. However, at present the collaborativeprocess used by Toyota to facilitate these trans-fers is somewhat of a black box.

This exploratory case study has two primaryobjectives. The first objective is to examinein detail the institutionalized knowledge-sharingroutines developed by Toyota and its suppliers.We also examine how Toyota has addressed thefundamental dilemmas associated with knowl-edge transfers in a network setting (e.g., freerider problems). Second, since Toyota is earlyin the process of creating a learning networkwith U.S. suppliers, we have the opportunity toexamine how a firm establishes a new knowl-edge-sharing ‘network.’ Thus, we examine thecreation and evolution of Toyota’s U.S. sup-plier network.


THEORETICAL BACKGROUND

Previous research suggests that organizations thatare effective at ‘learning’ have developed routinesthat allow them to effectively develop, store,and apply new knowledge on a systematic basis(Nelson and Winter, 1982; Levitt and March,1988; Cohen and Levinthal, 1990; Nonaka, 1994).Nelson and Winter (1982) were among the firstto argue that organizational ‘routines’ are theessence of the firm and that organizational learn-ing would be expected to occur when firmsdevelop ‘adaptation routines’ that allow the firmto modify existing routines based upon newknowledge. Other scholars have also viewedorganizational learning as ‘routine based and his-tory dependent’ (Levitt and March, 1988: 319).Following Grant (1996) we define a learningroutine asa regular pattern of interactions amongindividuals that permits the transfer, recombi-nation, or creation of specialized knowledge. Col-lectively, these routines may be viewed as acapability at managing knowledge flows in inter-firm networks (Lorenzoni and Lipparini, 1999).Of course, organizational learning may also behaphazard in the sense that knowledge is notgenerated by routines but rather is based on‘luck’ or haphazard events. In this study weare interested in studying ‘routine-based’ learning


rather than ‘haphazard’ learning. Our objective isto examine learning that is facilitated throughinterorganizational routines that arepurposefullydesigned to facilitate knowledge transfersacrossorganizational boundaries.

We are interested in the ‘production network’as the unit of analysis rather than the individualfirm (although these two levels of analysis areclearly related). Consequently, the routines ofinterest are those collectively developed withinToyota’s production network that are designedto facilitate knowledge transfers among networkmembers. Since our focal interest is knowledgesharing, it is useful to define what we mean byknowledge. Most scholars divide knowledge intotwo types: (1)explicit knowledge or information,and (2) tacit knowledge or know-how(Kogutand Zander, 1992; Grant, 1996; Ryle, 1984).Information is defined as easily codifiable knowl-edge that can be transmitted ‘without loss ofintegrity once the syntactical rules required fordeciphering it are known. Information includesfacts, axiomatic propositions, and symbols’(Kogut and Zander, 1992: 386). By comparison,know-how involves knowledge that is tacit,‘sticky,’ complex, and difficult to codify (Nelsonand Winter, 1982; Kogut and Zander, 1992; Szul-anski, 1996). The properties of know-how suggestthat, compared to information, know-how is morelikely to result in advantages that are sustainable(Nonaka, 1994). Our research elaborates on howthe Toyota network is designed to facilitate thesharing of tacit know-how.

Previous research suggests a number ofdilemmas associated with knowledge sharing(collaboration) in a network setting. The firstdilemma is how to motivate self-interested net-work members to participate in the network andto openly sharevaluable knowledge with othernetwork members (Wood and Gray, 1991). Thenatural tendency of individual firms is to protectknow-how viewed as proprietary to prevent unde-sirable knowledge spillovers. Consequently, manyfirms (especially those with proprietary know-how) will be reluctant to participate in interfirmknowledge-sharing activities. This is the essenceof the knowledge-based view of the firm asargued by Kogut and Zander (1992), who claimthat firms exist because they are better than mar-kets at transferring, recombining, and creatingknowledge. Thus, a key challenge for a knowl-edge-sharing network is to motivate members to


participate and contribute knowledge to the col-lective good. The knowledge that ismost likelyto be valuable to other firms in the network isoften exactly the kind of knowledge that individ-ual firms want to keep proprietary.

The second major dilemma is the ‘collectiveaction’ or ‘free rider’ problem. The free riderproblem is often discussed in collective actiontheories that examine the challenges associatedwith achieving collaboration toward commongoals among self-interested individuals, groups, ororganizations (Sandler, 1992; Marwell and Oliver,1993). Successful collaboration may produce ‘col-lective’ or ‘public’ goods (e.g., knowledge) thatare accessible to all members of the network.However, the creation of a public good (e.g.,useful knowledge) has the potential for ‘free rid-ers,’ members who enjoy the benefits of thecollective good without contributing to its estab-lishment and/or maintenance. In the case of aknowledge-sharing network, a firm may willinglyparticipate in network knowledge-sharing activi-ties to acquire the desired knowledge, and thenexit the network or refuse to contribute its knowl-edge.

The third dilemma has to do with how tomaximize the efficiency of knowledge transfersamong a large group of individual members. Itis possible for a network to create conditionsthat solve the first two dilemmas (members aremotivated to participate and ‘rules’ are in placeto prevent free riding), but the network may stillbe inefficient at knowledge sharing because ithas not created the necessary pathways amongmembers to facilitate efficient knowledge flows.By efficiency, we mean the speed and ease withwhich network members can find and access valu-able knowledge within the network. Some typesof knowledge (e.g., explicit) may be easily codi-fied and transferred in a large group setting (e.g.,through meetings), whereas other types of knowl-edge (e.g., tacit) require intense interaction andare likely to be successfully transferred only ina small group setting at the specific locationwhere the knowledge is used. If the network onlyconvenes large group meetings to share infor-mation it will likely be inefficient at transferringtacit knowledge among members. Thus, the net-work will likely need to create multilateral tiesamong members (and a variety of processes fortransferring knowledge) to reduce search costsand the speed and ease with which both explicit


and tacit knowledge is transferred amongmembers.

In summary, a successful knowledge-sharingnetwork must devise methods to (1) motivatemembers to participate and openly share valuableknowledge (while preventing undesirable spillov-ers to competitors), (2) prevent free riders, and(3) reduce the costs associated with finding andaccessing different types of valuable knowledge.

RESEARCH METHODOLOGY

In our exploratory multimethod case study the‘network’ consisted of Toyota and its group offirst-tier suppliers who collaborate to develop andmanufacture a motor vehicle. The automaker,Toyota, is the ‘core’ or nodal firm in the networkbecause: (1) Toyota is the only firm with directties to every other firm in the network, and (2)Toyota is the ‘convener’ of the network (to useGray’s, 1989, term). As the nodal firm, Toyotanot only has direct ties with each supplier, butalso has some economic interdependence witheach supplier. Thus, Toyota also has the most togain from developing learning routines thatincrease the efficiency of the entire productionnetwork.

Interview data

The primary objective of this study was to (1)identify the institutionalized knowledge-sharingroutinesdeveloped by Toyota and its network offirst-tier suppliers, and (2) explore how Toyota’snetwork attempts to resolve the knowledge-sharing dilemmas described in the previous sec-tion. Consequently, it was necessary to identifythe organizational units (divisions) within Toyotathat interact with suppliers and are involved in‘supplier development’ activities. This was doneby interviewing Toyota’s purchasing general man-agers in both Japan and the United States. Thesesenior purchasing executives identified the variousToyota divisions that had significant interactionswith suppliers. These divisions included: (1) Pur-chasing Division, (2) Operations ManagementConsulting Division (OMCD), (3) QualityAssurance Division (QAD), (4) LogisticsAdministration Division (LAD), (5) Manufactur-ing Operations Division (MOD), and (6) DesignEngineering Division (DED). Executives from


each of these divisions were interviewed to iden-tify the formalized routines designed to facilitateknowledge sharing with, and among, suppliers.As is appropriate in qualitative research, theo-retical sampling was used (Glaser and Strauss,1967; Strauss and Corbin, 1990). Our researchteam interviewed more than 30 Toyota executivestotaling more than 100 hours in interviews. Toverify the knowledge-sharing activities identifiedby Toyota, as well as explore the supplier-to-supplier routines not identified by Toyota, wealso interviewed senior executives at 10 of Toy-ota’s first-tier suppliers in Japan and 11 Toyotasuppliers in the United States (see Table 1 for alist of Toyota executives and suppliersinterviewed).

As we conducted interviews with both Toyotaand supplier executives, we focused our attentionon identifying and understanding the bilateral andmultilateral knowledge transfer routines amongToyota and its suppliers. We also explored thechallenges associated with the creation and main-tenance of those routines.

Archival and survey data

To corroborate the data from the interviews, wegathered archival and survey data to serve asnonreactive measures of changes in practice orperformance (Webb, 1981). For example, whenToyota executives and suppliers claimed that Toy-ota’s consultants were effective at knowledgetransfers to suppliers, we sought for data thatwould document changes in supplier performanceafter having received visits from Toyota’s con-sultants. We also conducted a survey of all 97suppliers in Toyota’s U.S. supplier associationand 50 of their largest first-tier Japanese suppliersto have them report on their knowledge-sharingactivities with Toyota. We received responsesfrom 39 U.S. suppliers and checked for responsebias (and found none) in a follow-up surveywhich added nine suppliers. Thus, we receivedsurveys from 48 U.S. suppliers (49.5% responserate). The Japanese survey, conducted with thesupport of the Japanese Auto ManufacturersAssociation (JAMA), produced 38 responses (a76% response rate). Because of the high responserate to our surveys in both the U.S. and Japanesesamples, we feel comfortable that our findings arerepresentative of the entire population of Toyotasuppliers. Whenever possible, we tried to verify


Table 1. Totota executives and suppliers interviewed

Toyota executives interviewed

Name Position

1. Koichiro Noguchi Gen. Mgr.–Int. Purchasing2. Michio Tanaka Gen. Mgr.–Int. Purchasing3. Junzo Matsumoto Mgr.–Planning & Admin. Int. Purchasing Div.4. Kenji Sato Corp. Mgr.–Purchasing5. Nobuhiko Suzuki Mgr.–Cost Reduction Int. Purchasing Div.6. Keiichi Tamura Asst. Mgr.–Engine, Power Train, & Electronics Parts &

Body Purchasing7. Tsuyoshi Kuriyamoto Mgr.–Metal, Interior & Chemicals, Parts & Body

Purchasing8. Bart D. Heller N. Amer. Project Group Int. Purchasing Div.9. Hiroshi Kawaguchi Mgr.–Administration Purchasing Planning

10. Ichiro Yamada Asst. Mgr.–Planning & Administration Group Int.Purchasing Div.

11. Yukihito Takemura Asst. Mgr.–Equipment Design Dept. Body12. Suguya Fukusato Proj. Mgr.–Int. Affairs Technical Admin. Div.13. Takaaki Matsumoto Gen. Mgr.–Public Affairs14. Noriyuki Yokouchi V.P.–Purchasing15. Chris Nielson Asst. Gen. Mgr.–Purchasing Planning16. Motoo Usui Coord.–Purchasing17. Lance Lewis Mgr.–Tech. Support Purchasing Planning18. F. E. (Gene) Tabor Gen. Mgr.–Parts and Components Purchasing19. Masami (Max) Suzuki Coordinator–Purchasing20. Tom Fitzgibbons Asst. Project Mgr.–Tech. Support–Purchasing21. Hajime Ohba General Mgr., TSSC22. Lesa Nichols Asst. Mgr.–Research & Training, TSSC23. Cindy Kuhlman-Voss Specialist–Research & Training, TSSC24. Koji Kondo Proj.Mgr.–Bus. Planning Proj. Planning/Mgmt.25. James R. Olson V.P.–External Affairs26. Kenji Miura Manager, OMCD27. Kazunori Hayashi Project GM, Int. Purchasing28. Toshihiro Sugai Gen. Mgr., Quality Control29. Ryoichi Hibio Manager, Purchasing30. Akihiko Morikawa, Manager, Purchasing

Toyota suppliers interviewed

Japan United States

1. Nippondenso Co. 1. Johnson Controls, Inc.2. Kojima Press Co. 2. Continental Metal Specialty, Inc.3. NSK Ltd. 3. Summit Polymers, Inc.4. Asahi Glass Co. 4. Amtex5. Aishin Seiki Co. 5. Aisin U.S.A. Mfg., Inc.6. Araco Corporation 6. Quality Safety Systems Co.7. Tokai Rika Co. 7. Lucas Body Systems NA–Ettrick8. Toyota Automatic Loom Works 8. Tokai Rika U.S.A., Inc.9. Toyota Auto Body Co. 9. Takata Inc.

10. Central Motor Wheel Co. 10. McKechnie Vehicle Components11. Grand Haven (GHSP)



the interview data through archival or surveydata.2

The processes involved in the constant com-parative method were used as internal checks onthe validity of the data. The constant comparativemethod is described well by Browning, Beyer,and Shetler (1995: 121):

As the research proceeds and new data are col-lected, they are constantly being compared toprior data in terms of categories and hypotheses.When new data yield new or inconsistent infor-mation, conceptual categories and the emergingtheory are modified to take them into account.This process is repeated until theoretical satu-ration is reached: until no new categories areemerging and no new information inconsistentwith the categories and tentative hypotheses isbeing generated. (Glaser and Strauss, 1967;Strauss and Corbin, 1990)

The multimethod approach to data collectionpermitted considerable within-method andbetween-methods triangulation. We compared thedata from interviews with the archival data. Thearchival and survey data could be juxtaposed tothe interview data to check for potential system-atic biases in the interview accounts. Discrep-ancies between interview data and archival datadiscovered in the course of the research raisedquestions which guided further data collectionand analysis. Data collection was concluded whena level of saturation was reached (Glaser andStrauss, 1967; Burgelman, 1994). This study issubject to the general limitations of generaliz-ability associated with field research which arewell documented (see Eisenhardt, 1989).

OVERCOMING THE KNOWLEDGE-SHARING DILEMMAS: TOYOTA’SSOLUTIONS

Our exploratory study of Toyota’s network sug-gests a number of important variables that influ-ence the extent to which knowledge is accumu-lated and transferred within a vertical network.In particular, we found that Toyota was able tosolve the dilemmas inherent in a knowledge-sharing network. First, to encourage suppliers to

2Nonreactive or unobtrusive data are historical, archival datawhich are not influenced by the perceptions or biases of theindividuals providing, or gathering, the data (Webb, 1981).


participate and openly share knowledge, Toyotahas heavily subsidized the network (with knowl-edge and resources) during the early stages offormation to ensure that suppliers realize substan-tial benefits from participation (we will discussthis in greater detail when we examine howToyota created its U.S. network). Suppliers aremotivated to participate because they quicklylearn that participating in the collective learningprocesses is vastly superior to trying to isolatetheir proprietary knowledge. Toyota has alsointroduced a number of network-level knowledge-sharing processes that have, over time, helpedcreate a strong ‘identity’ for the network. Pre-vious research on collaboration suggests that theeffectiveness of collaboration increases whenstakeholders ‘have a shared purpose’ (Wood andGray, 1991; Gray, 1989). Toyota has effectivelymotivated suppliers to participate in the networkby creating a set of conditions that help suppliersto strongly identify with the network.

Second, to address the free rider problem Toy-ota has established network rules/norms that pre-vent suppliers from accessing Toyota’s knowledgeunless they first explicitly agree to openly shareknowledge with other network members. Thisagreement is taken seriously because Toyota hasthe ability to impose economic sanctions (e.g.,withdraw business) on a supplier that violates therule. Finally, to ensure that the network isefficient at tacit knowledge transfers, Toyota hascreated a highly interconnected, strong tie net-work with a variety of processes that facilitateknowledge transfers. The network has multiplepathways among members (effectively elimi-nating most structural holes). Toyota’s strong tienetwork is well suited for the diffusion(exploitation) of Toyota’s production know-how(e.g., the Toyota production system) as well asthe existing know-how that resides within itssuppliers. In the following sections we examinehow Toyota has overcome the fundamentaldilemmas associated with creating an effectiveknowledge-sharing network.

Creating a network ‘identity’ throughnetwork-level knowledge-sharing routines

Kogut and Zander (1992: 383) argue that ‘whatfirms do better than markets is the sharing andtransfer of the knowledge of individuals andgroups within an organization.’ They further


argue that ‘a firm is distinct from a marketbecause coordination, communication, and learn-ing are situated not only physically in locality,but also mentally in an identity . . . This sharedidentity does not only lower the costs of com-munication, but establishes explicit and tacit rulesof coordination’ (Kogut and Zander, 1996: 502–503). Thus, knowledge is most effectively gener-ated, combined, and transferred by individualswho ‘identify’ with a larger collective. Creatingan ‘identity’ for a collective (e.g., firm, network)means that the individual members feel a sharedsense of purpose with the collective. The ‘iden-tity’ of the firm is defined by the organizationalboundaries which dictate who is (and who is not)a member of the organization, by shared goalsand values, and by patterns of interaction amongindividuals that give rise to a common languageand common frameworks for action (MacDuffieand Helper, 1997).

While Kogut and Zander’s arguments are appli-cable to firms, we found that in Toyota’s casethey are equally applicable to networks. If anetwork can create a shared identity among mem-bers, then that shared identity can lower thecosts of sharing knowledge within the network.Furthermore, the diversity of knowledge thatresides within a network is much greater thanthat which resides in a single firm. Consequently,if the network can get its members to ‘cooperatein a social community’ it will create learningopportunities far superior to firms that do notreside within such a network.

Toyota’s network is effective at knowledgesharing, in part, because a strong network ‘iden-tity’ has emerged and the network has establishedrules (network norms) that support coordination,communication, and learning. In Japan, Toyota’snetwork is known as the ‘Toyota Group’ andToyota openly promotes a philosophy within theToyota Group called ‘coexistence and co-prosperity’ (Kyoson kyoeiin Japanese). Toyotahas also promoted this philosophy as a core valueheld by Toyota in a manual (called ‘Selling toToyota’) that they created to help U.S. supplierslearn how to best work with Toyota. StatedKoichiro Noguchi, Toyota’s head of internationalpurchasing, ‘We must have the best suppliers inthe industry. We rely on them for our success.Quality cars require quality suppliers’ (August,1993). This sentiment was echoed by a Toyotasupplier executive in Japan (from Kojima Press)


who stated that ‘Toyota truly believes inkyosonkyoei. It’s not just a slogan. They are truebelievers.’ This executive unknowingly acknowl-edged that his company had internalized thephilosophy when he commented that, ‘As a Toy-ota supplier, we know that our success is tiedto Toyota’s.’

Toyota promotes the philosophy ofkyosonkyoei and creates a shared network identity bydeveloping network-levelknowledge acquisition,storage, and diffusion processes. The mostimportant of these network-level processes are:(1) the supplier association (a network-levelforum for creating a shared social community,inculcating network norms, and sharing [mostlyexplicit] knowledge), (2) Toyota’s operationsmanagement consulting division (a network-levelunit given accountability for knowledge acqui-sition, storage, and diffusion within the network),(3) voluntary small group learning teams(jishuken), or a sub-network forum for knowledgesharing that creates strong ties and a shared com-munity among small groups of suppliers, and (4)interfirm employee transfers (some job rotationsoccur at the network level). These four key proc-esses create an ‘identity’ for the network andalso facilitate knowledge transfers among networkmembers.3 We examine each of these processesin greater detail.

Supplier association

Toyota’s supplier association (kyohokai) in Japanwas established in 1943 to promote ‘mutualfriendship’ and the ‘exchange of technical infor-mation’ between Toyota and its parts suppliers.In 1996 Toyota’skyohokaihad three stated ‘pur-poses’: (1) information exchange between mem-ber companies and Toyota, (2) mutual develop-ment and training among member companies, and(3) socializing events (Internal Toyota Document,1996). To achieve these purposes, Toyota’s

3We view the relationship between identity and collectivelearning processes as analogous to the relationship betweeninterfirm (or interpersonal) trust and information sharing.Some trust (minimum threshold) is necessary for one partyto share confidential information with another party. Thus,some trust is a required condition for information sharing.However, as one party begins to share information with theother party, then trust increases. This begins a relationshipcharacterized by mutual causality. In similar fashion, webelieve that identity is both a cause, and a consequence, ofcollective learning processes.


kyohokai is divided into three regions: Tokaikyohokai (150 members) for the Tokai region(Aichi prefecture where Toyota City resides),Kanto kyohokai (65 members) for the Tokyoregion, and Kansaikyohokai(29 members). Toy-ota has created three separate regional associ-ations because it believes that suppliers mustbe in close geographic proximity to achieve theassociation’s objectives.

Toyota’s kyohokaihas general meetings everyother month (e.g., general assembly, top man-agement meetings) that are designed to allow forhigh-level communication within the networkwith regard to production plans, policies, markettrends, etc. Thus, these meetings primarily facili-tate the sharing of explicit knowledge amongmembers. More frequent interaction occurs withinthe association’s topic committees (cost, quality,safety, etc.).

The ‘topic’ committees on cost, quality, safety,and general affairs are designed to facilitateknowledge sharing on topics that are critical toall members in the network.4 The PR-Sports com-mittee (which primarily sets up golf events andbaseball activities) is designed to facilitate socialinteraction among members. To illustrate the rolethat the topic committees play in facilitating net-work identity and learning, we examine the qual-ity committee activities in greater detail.

The quality committee engages in a numberof activities designed to increase the amount ofknowledge that resides in the network with regardto product quality. The ‘regular committee’ picksa theme for the year (e.g., the 1994 theme was‘Eliminating supplier design defects’) and meetssix times each year to share knowledge withregard to that particular theme. These themes areselected by suppliers (with Toyota’s input) inareas believed to be important and relevant to alarge number of members in the network. Inaddition to the ‘regular committee’ meetings, thequality committee also sponsors ‘basic quality train-ing,’ ‘excellent plant tours,’ and an annual ‘qualitymanagement conference.’ The training course,which is provided by the association for both Toy-ota and its suppliers’ engineers, offers 12 days of

4The general affairs committee focuses on activities of currentinterest to a broad set of members. For example, during 1994(a recession year) the committee’s ‘theme’ was improvingthe productivity of white-collar workers. This is of particularimportance during a recession because Japanese firms cannoteasily lay off white-collar workers.


quality training to approximately 100 engineers eachyear. This training provides a basic, commonknowledge base (builds absorptive capacity) amongnetwork members which allows them to moreefficiently transfer more complex quality knowledgethrough plant tours, conferences, etc.

The excellent plant tours allow network mem-bers to visit ‘best-practice’ plants both inside andoutside the automotive industry so that memberscan see first hand the processes used by firms thatachieve high quality. The quality managementconference is held once each year and offersToyota suppliers the opportunity to learn fromlectures delivered by experienced Toyota directorsand senior managers, as well as six successfulsupplier cases of quality improvement: two plantmanagers’ cases, two supervisors’ cases, and twoquality circles’ cases. The six supplier cases areselected from case write-ups submitted to a com-mittee panel by member companies. In 1994,146 out of 150 Tokai Kyohokai members (98%)submitted cases to the conference.

In summary, the supplier association’s primaryobjective is to develop ties among members andtransfer explicit knowledge through multilateralknowledge transfers. The committees are designedto facilitate the transfer of both explicit and tacitknowledge. The supplier association is animportant vehicle for creating an ‘identity’ for theToyota production network or ‘Toyota Group.’Membership in the association makes membersfeel like they are part of a larger collective.Toyota’s attempt to replicatekyohokai in theUnited States began in 1989 when it organizedthe Bluegrass Automotive Manufacturers Associ-ation (BAMA).

Consulting teams/problem-solving teams

Toyota’s Operations Management ConsultingDivision (OMCD) was established in the mid-1960s by Taiichi Ohno to help solve operationalproblems both at Toyota and at suppliers. OMCDis the organizational unit within Toyota assignedthe responsibility to acquire, store, and diffusevaluable production knowledge that resides withinToyota’s production network. OMCD currentlyconsists of six senior and highly experiencedexecutives (each with responsibility for two Toy-ota plants and approximately 10 suppliers) andabout 50 consultants. OMCD facilitates knowl-edge sharing by providing direct ‘on-site’ assis-


tance to suppliers. This typically involves sendinga team of consultants to the supplier for a periodof time ranging from one day to many months,depending on the nature of the problem. Thisassistance is ‘free’ to suppliers who are notcharged for the consultants’ time. Thus, this is anetwork-level resource that is accessible to allmembers. Our 1993 survey of 38 first-tier Toyotasuppliers in Japan revealed that all suppliers inour sample had, at one time or another, beenvisited by Toyota personnel who assisted inimproving the suppliers’ operations. On average,suppliers reported receiving 4.2 visits per yearand these visits lasted an average of 3.1 days.

The U.S. version of OMCD is called the Toy-ota Supplier Support Center (TSSC). TSSC wasestablished in 1992 with the objective of ‘assist-ing North American suppliers to implement theirown version of TPS’ (Internal Toyota document,1995). TSSC’s general manager, Hajime Ohba(formerly of OMCD) heads a staff that had grownto 20 consultants by 1997. Since 1992 TSSC hasreceived approximately 100 requests for assis-tance and has entered into 53 consultationprojects.

To date, Toyota has found that know-howtransfers with regard to TPS are extremely diffi-cult and time consuming because the knowledgeis mostly tacit. Although the goal is to achievesuccess in 6 months, no project has been com-pleted in less than 8 months and most last atleast 18 months. States Mr. Ohba, ‘It takes a

Figure 2. TSSC project results (31 suppliers)


very long time and tremendous commitment toimplement the Toyota Production System. Inmany cases it takes a total cultural and organi-zational change. Many U.S. firms have man-agement systems that contradict where you needto go.’ Consequently, some of TSSC’s consultingprojects can be quite time and resource intensive.According to an executive at Grand Haven,‘TSSC has been providing assistance to our plantfor almost three years. One consultant practicallylived at our plant for a year and became part ofthe plant family. In fact, we even threw him ababy shower when he had his first baby.’

By November of 1996, TSSC had engaged in31 projects with suppliers with impressive results.On average, TSSC had assisted suppliers inachieving an average inventory reduction of 75percent and an average increase in productivityor output per labor hour of 124 percent (seeFigure 2). This compares to an average inventoryreduction of 8 percent, and an increase in outputper worker of 6 percent during the 2-year periodbefore TSSC began working with the supplier.Thus, the rate of improvement has increased dra-matically, suggesting that the supplier has beenlearning faster since working with TSSC. Thesedata provide evidence that TSSC’s knowledgetransfer processes substantially improve supplierperformance.

Toyota also has a process of forming problem-solving teams designed to bring knowledge tobear in solving emergent problemswithin the


network. For example, a supplier may experiencequality problems where the root cause is noteasily determined. In this case, OMCD or Toy-ota’s Quality Assurance Division (QAD) will setup a problem-solving team (with various Toyotadivisions and possibly even other Toyotasuppliers) to collectively bring their knowledgeto bear to ‘fix’ the quality problems. When estab-lished, the team defines the root cause(s) of thesupplier’s quality problems and hands over theproblem-solving process to an appropriatedivision(s) within Toyota. For example, once theproblem-solving team has defined the root of thesupplier’s quality problem as being in productdesign, Toyota’s Design Engineering Division(which has already been involved in the problem-solving team) will be asked to take the lead inworking more closely with the supplier toimplement solutions to improve quality. In somecases, Toyota may determine that the relevantknowledge resides within a competitor of thesupplier. In this case, Toyota will attempt toorchestrate a supplier-to-supplier knowledgetransfer.

We did not find this type of supplier-to-supplierknowledge sharing in the United States. However,eight of 10 U.S. suppliers we interviewed reportedthat Toyota orchestrated trips to Japan duringwhich they visited Japanese suppliers producingthe same component. As a supplier executive atLucas Body Systems stated, ‘Toyota told us towork on cutting our changeover time from 2hours to 30 minutes. I told them it was impos-sible. Then they sent me to visit a Japanesesupplier in our same business that had changeovertimes of 15 minutes. I never would have believedit if I hadn’t seen it with my own eyes. My bossstill doesn’t believe it.’

Voluntary learning teams (jishuken/PDA coregroups)

OMCD facilitates ties and knowledge sharingamong suppliers in a way that is quite unique.In 1977 OMCD organized a group of roughly55–60 of its key suppliers (providing over 80percent of its parts in value) into ‘voluntary studygroups’ (Jishukenkyu-kai or jishuken) for the pur-pose of assisting each other with productivityand quality improvements. Each supplier groupconsists of roughly five to eight suppliers, someof whom use similar production processes (e.g.,


stamping, welding, painting). Body suppliers areplaced in one of two groups (Group I, II) andparts suppliers are placed in one of seven groups(Groups A–G). Toyota groups suppliers togetherbased upon: (1) geographic proximity, (2) com-petition (direct competitors arenot in the samegroup),5 and (3) experience with Toyota (eachgroup has at least one affiliated Toyota supplier,such as Denso or Aisin Seiki, that are expectedto take a leadership role). Groups are usuallyreorganized every 3 years by Toyota to put some‘stimulus’ into the activity and maintain diversityof ideas. Each year the suppliers meet togetherwith the responsible OMCD manager to deter-mine a ‘theme’ (project) for the year. The basicidea is to help each other increase productivityin areas of common interest. Supplier executivesthat participate injishukenactivities are typicallyplant managers, assistant plant managers, and/orsection managers.

After a theme is decided, the group establishesa schedule to visit each supplier’s plant to jointlydevelop suggestions for improvement. The groupspends 4 months (with four phases) focused oneach supplier acting as consultants to the plant.The phases are as follows: (1) preliminary inspec-tion, (2) diagnosis and experimentation, (3) pres-entation, (4) follow-up/evaluation. A member ofOMCD also visits frequently (i.e., every week ortwo) to give advice and monitor progress. Thisallows Toyota to bring its expertise to bear tohelp solve supplier problems. It also allows Toy-ota to ‘learn’ what is being learned by suppliers.This adds to Toyota’s stock of knowledge andallows OMCD to keep abreast of new ideas andapplications of the Toyota Production System.This is valuable to both Toyota and the networkbecause OMCD can transfer this knowledge toToyota’s internal operations or to other suppliers.At the end of the year Toyota organizes a meeting(conference) where all of thejishuken groupsmeet together to share what they have learnedfrom the year’s activities.

In 1994 Toyota established its Plant Develop-ment Activity (PDA) core groups, which was anattempt to replicate thejishuken concept in theUnited States. Three groups were formed (using

5Only a small number (perhaps one to five) of suppliers inthe network are viewed as direct competitors. Thus, mostsuppliers in the network are in a noncompetitive situationwhen it comes to knowledge sharing.


the same criteria as in Japan) with 11 suppliersplaced in each group. The theme for the firstyear was ‘quality improvement’ because, as Toy-ota’s Chris Nielsen noted, ‘everyone agrees thatthey can improve quality’. Each core group mem-ber was asked to select a ‘demonstration line’within a plant as a place to experiment whenimplementing the concepts. A schedule wasdeveloped to meet for one day each month at asupplier plant, during which Toyota personnelfrom the technical support group would demon-strate some key concepts at the demonstrationline, and the group members would discuss waysto improve the line. The group would visit thesame supplier for 3 months and then rotate to anew supplier.

After the first year, many other supplierswanted to participate, so Toyota added 15 sup-pliers and another PDA core group in 1995(roughly 12 suppliers per group). However, someof the groups were experiencing difficultiesbecause of markedly different skills and knowl-edge of TPS methods among suppliers. StatedToyota technical support specialist Tom Fitzgib-bons, ‘we tried high skill and low skill supplierstogether but sometimes it didn’t work wellbecause we had to keep stopping to explain basicconcepts to the new suppliers. The more experi-enced suppliers were getting frustrated’. Conse-quently, Toyota reorganized the groups in 1996into four groups where ‘skill level’ was alsoconsidered. The ‘orange’ group included supplierswith strong TPS skills, the ‘blue’ and ‘green’groups with mid-level skills, and the ‘purple’group consisted of less experienced members.

To be considered for participation in a PDAcore group the supplier must be a member ofBAMA for at least one year. Toyota claims thatthis requirement is necessary to assure assimi-lation into BAMA, top management commitment,and familiarity with the basic concepts of TPS.U.S. plant managers report that they believe thePDA core group activities have been extremelyvaluable. Stated the plant manager of Amtex,

We get blinded just like everyone else. Whenyou bring a whole new set of eyes into yourplant you learn a lot. We’ve made quite a fewimprovements. In fact, after the (PDA) coregroup visits to our plant, we made more than 70changes to the manufacturing cell. One suppliereven showed us a new way to start up ourequipment that was better on the machinery andsaved on energy costs.


In fact, all 10 U.S. suppliers we interviewedclaimed that the PDA core group activities weremore valuable to suppliers than BAMA. TheAmtex plant manager went on to say,

I think BAMA is valuable. But the PDA coregroup activities are even more valuable to usthan BAMA. If I had to choose, I would defi-nitely choose to be involved in the core groupsover BAMA. We learn more that is useful inour daily operations.

This comment suggests that the tacit knowledgeacquired through PDA core groups is perceivedto be more valuable than the mostly explicitknowledge acquired through BAMA. A key rea-son that the PDA core group activities are partic-ularly effective at tacit knowledge transfers isthat they involve learning that is context specific(‘hands on’ and ‘on site’). For example, the plantmanager from Kojima Press (a Japanese supplierof spoilers and body parts) described how theyhad acquired tacit knowledge through a jishukenproject: ‘The jishuken projects are very helpful.Last year we were able to reduce our paintcosts by 30 percent. This was possible due to asuggestion to lower the pressure on the paintsprayer and adjust the spray trajectory, therebywasting less paint’. Furthermore, as the plantmanager at Continental Metal Specialty noted,‘We find more things that are useful visitingother suppliers’ plants versus Toyota’s plants.Suppliers’ operations are more similar to ours’.

Like the supplier association, the learningteams help create an identity for the larger collec-tive. As a supplier executive at McKechnieVehicle components put it, ‘We’re a member ofthe “blue group” and we have developed goodrelationships with the suppliers in ourgroup . . . So we do what we can to help otherblue group members’.

Interfirm employee transfers

The practice of interfirm employee transfers(shukko) in Japan is by now well known(Cusumano, 1985; Lincoln, Gerlach, and Takah-ashi, 1992; Gerlach, 1992; Dyer and Ouchi,1993). Some previous studies suggest thatimportant reasons forshukko include helpinglarge assemblers maintain control of suppliersand the opportunity to shed unwanted employees(Lincoln et al., 1992; Gerlach, 1992). However,


our interviews suggest that, at least in Toyota’scase, shukko is also an important mechanismfor creating a network identity and transferringknowledge from Toyota to suppliers. In our sur-vey of 38 of Toyota’s first-tier suppliers, wefound that 11 percent of the suppliers’ directors(yakuin) were former employees (the figure was23 percent for Toyota’s ‘affiliated’ suppliers orsuppliers in which Toyota owned some stock).Overall, Toyota transfers approximately 120–130individuals per year to other firms in the valuechain, most of whom go to suppliers. Some ofthese transfers are permanent in nature (usuallyat the director level) but others are temporary.For example, when we visited Kojima Press, wefound that the assistant plant manager was aToyota engineer on leave from Toyota for a 2-to 3-year assignment. Stated the Toyota engineer,‘I am here to apply what I’ve learned at Toyotato help the plant manager run a more efficientplant. Also, by working at the supplier I canunderstand the supplier’s perspective and whatproblems they experience’. At another supplier,we interviewed a ‘transferee’ who had been sentto the supplier to help it set up operations andaccounting systems in the United States. Thisparticular individual had worked in the automak-er’s U.S. operations and therefore had a knowl-edge of U.S. accounting systems which he wasable to transfer to the supplier. In some casesthe supplier had a need for particular skills orknowledge that members of its workforce did notpossess. Consequently, it would make a requestto the automaker (usually through purchasing tothe personnel department) for someone with parti-cular skills. The automaker would search withinits organization and then offer someone to thesupplier organization. Suppliers claimed that theyhad the right to refuse the person offered. How-ever, given Toyota’s importance as a customer,it is questionable as to whether this ‘right’ isever exercised. Regardless, these transfers arean important routine which fulfills a knowledgetransfer function. The transferred individuals notonly bring particular technical knowledge withthem, but more importantly they bring with thema knowledge of Toyota’s personnel, systems, andtechnology. Thus, they know where particularexpertise resides within Toyota as well as whoto contact to access that expertise. Furthermore,the fact that individuals can be transferred acrossfirm boundaries indicates that the ‘unit of analy-


sis’ for a job rotation is not the individual firm,but the network. Thus, this practice further createsan identity for the network. To date, these trans-fers only occur with Japanese suppliers in Japan.

In summary, these four network-level knowl-edge-sharing processes (supplier association, con-sulting teams, learning teams, and employeetransfers) each play an important role in creatinga ‘shared purpose’ among suppliers and helpingthem believe they are part of a larger collective.These network-level learning processes facilitatefrequent face-to-face interactions among suppliersand this creates a social community at the net-work level. Moreover, Toyota (the nodal firm)uses network-level processes to encourage andeducate members to believe that they are part ofan interdependent economic community. As thesupplier increasingly identifies with the network,it engages in knowledge-sharing activities quiteunconsciously—without any sort of cost–benefitcalculus. To illustrate, when an executive atMcKechnie Vehicle Components was asked whyhe allowed other Toyota suppliers to visit hisplant and learn from his operations, he responded,‘Because they are members of our supplierassociation, we do what we can to help themout. We help each other because it makes us allbetter off’. This statement reflects the sentimentthat ‘what’s good for the network is good forme, and what’s good for me is good for thenetwork’. Thus, the creation of a strong networkidentity lowers the costs of participation by mem-bers (they are less likely to engage in a cost–benefit calculus) and increases the value of par-ticipation (members are more likely to contributevaluable know-how).

Network ‘rules’ for knowledge protection andvalue appropriation

Creating a network identity was a critical firststep to solving the dilemma of motivating mem-bers to participate and contribute useful knowl-edge to the network. However, Toyota has alsoestablished some ‘rules’ within the network thatprevent members from both (a) protecting orhiding valuable knowledge, and (b) free riding.Toyota solves these dilemmas by simply elimi-nating the notion that there is ‘proprietary knowl-edge’ within certain knowledge domains (e.g.,production-related knowledge). By openly sharingall of its production know-how, Toyota has cre-


ated a norm (rule) within the network that verylittle of the knowledge that a firm possesses isproprietary (with the exception of certain productdesigns/technology). Production processes aresimply not viewed as proprietary and Toyotaaccepts that some valuable knowledge will spillover to benefit competitors. Thus, any production-related knowledge that Toyota or a supplier pos-sesses (cost, quality, inventory management, etc.)is viewed as accessible to virtually any memberof the network (withperhapsthe exception of adirect competitor) because it is, in effect, theproperty of the network.

Toyota creates a norm of reciprocal knowledgesharing within the production network by provid-ing free assistance to suppliers and allowing sup-pliers full access to Toyota’s operations and stockof knowledge. According to Koichiro Noguchi,head of international purchasing, ‘Suppliers canvisit any Toyota facility with the possible excep-tion of the new model design room. We hidenothing.’ But suppliers are told theymust bewilling to open their plants to other networkmembers if they choose to receive Toyota con-sulting assistance and/or participate injishuken/PDA core groups. States TSSC consul-tant Lesa Nichols, ‘That’s one of our requirementsbecause if we take the time and effort to transferthe know-how, we need to be able to use thesuppliers’ operations as a vehicle to help othersuppliers.’ This requirement allows Toyota to:(1) develop some ‘showcase suppliers’ that havesuccessfully implemented various elements of theTPS; this provides a valuable learning laboratoryfor other suppliers, and (2) get suppliers to opentheir operations to one another.6 This requirementalso effectively minimizes the free rider problembecause the ‘price of entry’ into the network isa willingness to open up your operations forinspection. As a supplier executive at SummitPolymers stated, ‘They gave us a gift [TPS];how can we not open our plant and share whatwe’ve learned with other Toyota suppliers.’ Thus,Toyota’s willingness to freely share its valuableknowledge with other network members acts asa ‘starting mechanism’ (see Gouldner, 1960) forreciprocity, or more specifically for reciprocalknowledge sharing of proprietary knowledge

6Suppliers can designate certain other areas of their plants(where Toyota has not provided assistance) as off-limits tovisits in order to protect proprietary knowledge.


within the network.7 In effect, Toyota states that‘We will help you, but in return, you must agreeto help the network.’ The rule can be stated asfollows: The price of entry into the network is alimited ability to protect proprietary productionknowledge. Intellectual property rights reside atthe network, rather than firm, level.

To date, Toyota claims that no suppliers havereceived Toyota’s help and then refused to opentheir operations to other suppliers. When askedwhat would happen if a supplier was unwillingto open their operations to other firms in thenetwork, one Toyota executive responded, ‘thatwould be a serious breach of faith. Such behaviorwould make it difficult to give that supplieradditional business.’ Suppliers also seem to beaware of the possibility that Toyota could imposeeconomic sanctions by withdrawing business.Stated an executive at Johnson Controls, ‘Toyotahas a certain way of working with suppliers withspecific expectations. Suppliers that understandand are effective at meeting those expectationswill be successful. Suppliers that don’t have ahard time . . . You don’t want to tell Toyota youwill do something and then not do it.’

In addition to a rule that delineates propertyrights, the network has also established a rulethat defines the timing and distribution of savingsthat result from knowledge transfers. Sinceknowledge transfers are both costly for thesource, and for the recipient, of knowledge(Szulanski, 1996), the network needs some wayto distribute the value realized from knowledgetransfer activities. For example, Toyota makes aconsiderable investment in OMCD and TSSC andyet their assistance is ‘free’ to suppliers. If so,how does Toyota benefit from its investment?Are suppliers allowed to appropriate all of thegains? We found that when Toyota’s consultantswork with suppliers to transfer knowledge thatresults in lower costs (and higher supplierprofits), it does not ask for an immediate pricedecrease. In fact, Toyota has intentionally sepa-rated OMCD and TSSC from the purchasingdivision so that suppliers can work with the

7As we shall describe later, Toyota heavily subsidizes thenetwork (with knowledge and financial resources) in the earlystages of its formation to get members to participate and toestablish norms around reciprocal knowledge sharing. How-ever, as the nodal firm, over the long term Toyota receivesthe greatest benefits from a production network that is activelyengaged in knowledge-sharing activities.


consultants without fearing that purchasing willask for a price decrease after the consultation.According to TSSC head Hajime Ohba, ‘Our jobis to help suppliers improve, not to worry aboutwho gets the additional profits. If the supplierbecomes more productive, over time we willcapture some of those benefits.’ Suppliers widelyrecognized that, at least in the short run (e.g., 1–2 years) they could appropriate 100 percent ofthe value from the knowledge transfers. Thisrule increased their incentive to participate in thenetwork’s knowledge-sharing activities. Stated asupplier executive at Amtex, ‘Toyota helps youwithout any immediate expectations. The fact thatthrough BAMA we have made process improve-ments that have saved us money and Toyota didnot directly go after those savings speaks vol-umes. Of course, over time we must give backsome of the savings in price cuts at the annualprice review.’ Thus, the tacit rule that hasemerged around value appropriation can besummarized as follows:The recipient of knowl-edge may appropriate 100 percent of the savingsin the short run, but over time will be expectedto share a proportion of those savings with thenetwork (e.g., through price cuts to Toyota).

The importance of this rule can be demon-strated by comparing the response of U.S. sup-pliers to receiving Toyota consultants vs. receiv-ing General Motors consultants. Most suppliersclaimed that they would willingly open theirfactories to Toyota consultants for two reasons.First, they believed that the Toyota consultantswere experienced and possessed knowledge thatcould be truly valuable to their plants. Second,they trusted Toyota to offer the assistance withoutdemanding an immediate price cut in return forcost improvements. In contrast, many GeneralMotors suppliers indicated that they would prefernot to have General Motors’ consultants (calledPICOS teams) come to their plants. GM’s ‘con-sultants’ have had a very different reception—largely because there is a different rule aboutvalue appropriation. As described by one sup-plier executive,

We don’t want to have a PICOS team pokingaround our plant. They will just find the ‘lowhanging fruit’—the stuff that’s relatively easy tosee and fix. We all have things in our plants thatwe know need to be fixed. They’ll just come in,see it, and ask for a price decrease. We’d preferto find it ourselves and keep all of the savings.


Stated another supplier executive,

We allowed a PICOS team into our plant. Theystayed for about a week and did make a fewuseful suggestions. Then we received a letterdemanding a price decrease based upon the pro-jected ‘savings’ from their suggestions.

Thus, even though suppliers may be able tolearn something from GM, they are reluctant toaccept assistance or share any informationbecause GM’s value-appropriate rule reduces thesupplier’s incentive to participate. The result isthat GM may try to engage in knowledge transferactivities with suppliers, but suppliers are notproperly motivated to participate.

Creating multiple knowledge-sharingprocesses and sub-networks in the largernetwork

The third dilemma has to do with how to maxi-mize the efficiency of knowledge transfers amonga large group of individual members. To addressthis problem, Toyota has established a variety ofbilateral and multilateral processes, each designedto facilitate the sharing of different types ofknowledge (both explicit and tacit) within thenetwork (see Figure 3 for a summary of the keynetwork-level routines that facilitate the sharingof different types of knowledge within Toyota’sproduction network). Also, some processes aredesigned primarily for knowledge diffusion, whileother processes result in both knowledge creationand diffusion. (In this context, knowledge ‘cre-ation’ primarily refers to new, incremental appli-cations of the Toyota Production System ratherthan the creation of innovative, breakthroughtechnologies.) These various processes allow fora ‘matching’ of the type of knowledge with aprocess that is the most effective and efficient attransferring that particular type of knowledge. Forexample, the supplier association is a vehicle forefficiently disseminating explicit knowledge (e.g.,market trends, production voles) to all membersof the network. In contrast, thejishuken groupsare an effective vehicle for multilateral transfersof tacit knowledge (e.g., TPS methods, processes,philosophies) to members in the network. Bycreating sub-networks or ‘nested networks’ (e.g.,the jishuken groups, supplier associationcommittees) within the full network, individualmembers are able to develop embedded ties with


Figure 3. Toyota’s network-level knowledge-sharing processes

those other members that have particularly rel-evant knowledge. Thus, there are a variety ofprocesses through which members can choose toreceive, or transfer, different types of knowledge.As a supplier executive at Amtex jokinglyremarked, ‘Do you know what TMM (ToyotaMotor Manufacturing) stands for? Too ManyMeetings. You just can’t avoid communication.There are so many avenues, between BAMA, thePDA core groups, and personal visits.’ Stated asupplier executive at Continental Metal Specialty,‘Each program serves a different purpose. BAMAis great for helping us keep abreast of what’sgoing on in the industry and what Toyota expectsfrom us. But for really useful ideas on how toimprove your operations, the [PDA] core groupsare superior.’ The ‘many avenues’ for communi-cating creates a high degree of multiplicity(interconnectedness) among members in Toyota’snetwork, thereby giving individual members achoice of medium for communicating. Further, itis not necessary for these pathways to culminate,or even intersect, with Toyota.

In summary, Toyota’s network is able toefficiently transfer knowledge because there area variety of processes available to transfer bothexplicit and tacit knowledge in a multilateral orbilateral setting. If the network only met in largegroup (i.e., supplier association) meetings, itwould not be very efficient at transferring tacitknowledge. Similarly, if Toyota only providedconsultants (but not the association) to engage in


bilateral knowledge transfers, the network wouldbe inefficient at transferring explicit knowledge.Thus, to maximize the speed and ease with whichvarious types of knowledge are transferred, avariety of routines or pathways for knowledgeflows is required.

THE CREATION AND EVOLUTIONOF TOYOTA’S U.S. KNOWLEDGE-SHARING NETWORK

By examining the creation and evolution of Toy-ota’s network over time, we gain insights intohow a core firm (convener) may attempt to suc-cessfully create and manage a knowledge-sharingnetwork. In particular, our longitudinal study ofToyota’s relatively new U.S. supplier networkoffers the opportunity to examine how the net-work has evolved over time. Our exploratorystudy of Toyota suggests that three institutionalinnovations have played an important role inthe creation of the network and in facilitatinginterorganizational learning. These innovationswere: (1) the supplier association, (2) the knowl-edge transfer consultants (OMCD, TSSC), and(3) jishuken/PDA core groupsor small-grouplearning teams. In both Japan and the UnitedStates, these institutions were introduced to sup-pliers in the same order (see Figure 4). Wedescribe the evolution of the network as occurringin phases, from weak ties, to strong bilateral ties


Figure 4. Evolution of Toyota’s U.S. supplier network

with the convener (Toyota), to strong multilateralties among suppliers.

Phase 1: Developing weak ties

In 1988, when Toyota first began producing carsin Georgetown, Kentucky, its suppliers did notinteract with each other (and the suppliers’ tiesto Toyota were ‘weak’ in the sense that therelationships were new and the frequency andintensity of interactions was low). Knowledgesharing in the supplier network was virtuallynonexistent. The first step that Toyota took toinitiate knowledge sharing in the supplier networkwas to establish the supplier association in 1989.The supplier association facilitated the sharing ofexplicit knowledge (mostly unidirectional flowsfrom Toyota to suppliers) and the creation of‘weak’ social ties among suppliers. This was acritical first step in getting members of the net-work to talk to each other in a nonthreatening


setting. Stated Toyota’s Chris Nielsen, assistantgeneral manager for purchasing planning.

We really didn’t know if this would work in theU.S. Getting suppliers to talk to each other wasa key element of the program. Before BAMA, itwas not very natural for supplier executives totalk and share information. It was uncomfortable.Over the years that has changed significantly assuppliers have built relationships at senior levels.(Interview, November 17, 1997)

This message was echoed by an executive atLucas Body Systems,

Before BAMA, we really didn’t know or shareinformation with executives at other suppliers.And we just didn’t think about calling them upor visiting. It just didn’t happen. BAMA hashelped us to get to know each other and now itfeels a lot more comfortable calling up anothersupplier for information or even visiting theirplants. (Interview, April 18, 1997)

Thus, BAMA was the catalyst for creatingweak ties among network members and it createdan avenue for sharing explicit knowledge. Partici-pation was not threatening to suppliers becausethey were not required to share any sensitiveinformation or commit significant resources.

Phase 2: Developing strong ties with Toyota

Subsequent to establishing the supplier associ-ation, Toyota made its well-trained consultantsavailable (free of charge) to transfer valuabletacit know-how regarding TPS at the suppliers’facilities. The Toyota consultants were the cata-lysts for creating a norm of reciprocal knowledgesharing and a feeling of indebtedness and open-ness within the supplier network. Indeed, toreceive assistance from Toyota consultants thesupplier had to agree to open up its plant toother Toyota suppliers. As the vice president ofplanning for Summit Polymers, a supplier ofplastic interior parts, stated,

I couldn’t believe it but Toyota sent approxi-mately 2–4 consultantsevery day for a periodof 3–4 months as we attempted to implementToyota Production System concepts in a newplant. They gave us a valuable gift [the ToyotaProduction System]. Naturally we feel indebtedtowards Toyota and view them as a special cus-tomer; they sincerely want to help usimprove . . . How could we try to keep what


we’ve learned from other Toyota suppliers?(Interview, November 19, 1996)

Thus, Toyota’s consulting assistance createdmuch stronger direct ties with suppliers and wasthe catalyst for creating, in Gouldner’s (1960)terminology, ‘a norm of reciprocal obligation.’As more and more suppliers had an intensiveknowledge transfer experience with Toyota’s con-sultants (typically over a 1-to 2-year period),they became comfortable with knowledge transferactivities. They also experienced first hand thepotential economic benefits associated withknowledge sharing. Moreover, an increasing num-ber of suppliers were obligated to allow visitsfrom other suppliers in the network.

Phase 3: Developing strong ties amongsuppliers

Finally, after suppliers had developed social tiesand norms of reciprocal knowledge sharing withToyota, Toyota divided suppliers into small learn-ing teams. Toyota carefully organized the teamsto maximize the willingness and ability of sup-pliers to learn from each other (e.g., keepingdirect competitors apart; rotating group member-ship to maximize diversity of ideas, requiring aminimum level of absorptive capacity with regardto TPS, etc.). By doing so, Toyota created a setof ‘sub-networks’ within the full network. Thesesub-networks were designed to facilitate the cre-ation of strong ties among suppliers, which inturn has facilitated the sharing of tacit knowledgeamong members. Stated the plant manager ofMcKechnie Vehicle Components,

We’ve benefitted greatly by participating in the[PDA core] group. It is now very natural for usto share what we know with the other suppliersin our group. We know each other well and arecommitted to helping each other . . . I don’t knowthat I could have imagined this sort of activityfive years ago. We just didn’t interact withother suppliers.

Not only have strong ties developed throughthe formal ‘core group’ activities—strong tiesand avenues for knowledge sharing have alsodeveloped informally. Stated an executive atAmtex,

We don’t just visit ‘core group’ members. Infact, one of our most helpful visits was to Tower.


We heard that Tower had one of the best kanbansystems so we asked them if we could visit andsee what they do. We had a very productive visitand they are coming to visit us as well.

Thus, over time the network has becomeincreasingly interconnected with numerous redun-dant ties. As evidence, our sample of 48 U.S.suppliers reported that in the yearbefore partici-pating in BAMA they only visited, or were visitedby, other automotive suppliers an average of 5.1days during the year. However, after joiningBAMA and the PDA core groups, they reportedvisiting, or being visited by, other suppliers 33.8days per year (a 7× increase in face-to-face con-tact, and these numbers do not include theincreased face-to-face contact through BAMA).Thus, the structural holes in the network aredisappearing. The result is that the network reliesless on Toyota to direct and facilitate the knowl-edge-sharing activities. States Nielsen, ‘We wantour suppliers to help each other. That’s the wholeidea. We realize we simply don’t have theresources or information to help all our suppliersas much as we’d like to.’ However, while Toy-ota’s role as an arbitrager of knowledge hasdiminished, Toyota’s ‘control’ over the network-level knowledge-sharing processes (e.g., associ-ation, consultants) helps it maintain its leadershiprole (power and relevance) in the network. Forexample, the consulting division is a mechanismthat allows Toyota to simultaneously ‘learn’ fromthe network (thereby giving Toyota powerbecause Toyota possesses greater access to moreleading-edge knowledge than other members) andalso ‘monitor’ the network’s activities.

In summary, the network structure, motivationfor member participation, and types of knowledgeshared have evolved substantially within Toyota’ssupplier network. In the initiation phase the net-work structure was essentially a collection ofdyadic ties with the nodal firm (Toyota) as ahub heavily subsidizing network activities (seeFigure 5). Toyota’s subsidies came in two forms:(1) financial (money for meeting rooms, socialactivities, organizing and planning meetings); and(2) valuable knowledge (Toyota invested inTSSC and would send its consultants free ofcharge to participating members). It wasimportant for Toyota to subsidize network knowl-edge-sharing activities early on to motivate mem-bers to participate and to ensure that they realizedsufficient benefits from participation.


Figure 5. Evolution of a knowledge-sharing network

In the early stages of network formation, therewere weak ties among network members andthere were numerous structural holes (Burt,1992). Suppliers were motivated to participate inthe supplier association primarily to demonstratetheir commitment to Toyota in hopes that Toyotawould reward them with more business. The typeof knowledge exchanged was primarily explicitknowledge. During this initial phase the networkwas just beginning to develop an identity, butthere was a weak social community and suppliersdid not perceive a strong sense of shared purposewith other firms in the network.

However, Toyota gradually built strong bilat-eral relationships with suppliers through the one-to-one knowledge transfers (consultants) and thesupplier association activities. Suppliers began toreceive valuable knowledge at minimal cost.Consequently, suppliers increasingly participatedin the network not only to demonstrate theircommitment to Toyota, but also to receive theknowledge transfers from Toyota. The knowledgeexchanged was primarily explicit knowledge inthe large network (supplier association) setting,but tacit (and more valuable) knowledge wasbeginning to be transferred in the bilateral(consulting) setting. During this phase thenetwork began to take on a stronger identity.Suppliers began to feel an obligation to Toyotaand began to identify with the network’s socialcommunity.


The final phase in the evolutionary process wasto strengthen multilateral ties among membersand develop ‘sub-networks’ for knowledge shar-ing within the larger network. This wasaccomplished through the learning teams whichstrengthened multilateral ties, thereby facilitatingtacit knowledge sharing among suppliers. At thisstage of network evolution, the supplier’s moti-vations for participating in network activities are(a) a recognized need for rapid knowledge acqui-sition, and (b) reciprocity. Suppliers have nowexperienced the benefits of knowledge sharing.Suppliers also appreciate how important it is, asa Toyota supplier, to rapidly learn and makecontinuous improvements. Suppliers are kept ontheir toes by knowing that the rent-creating poten-tial of their past productivity improvements isdeclining rather fast. Further, they are in a ‘learn-ing race’ with the other Toyota supplier(s) thatproduce similar parts in the sense that the fastest-learning suppliers are more likely to get thebusiness for new models. This creates strongincentives for suppliers to learn and improve asquickly as possible. In addition, suppliers nowidentify with the network and have developedreciprocal obligations for sharing knowledge withother members of the network.

In this mature phase, there are multiple path-ways for transferring both explicit and tacitknowledge, with tacit knowledge being exchangedprimarily in the sub-networks and the one-to-one


transfers. Thus, the degree of tacit knowledgebeing transferred issubstantial whereas it wasalmost nonexistent in the initiation phase (seeFigure 5). Not surprisingly, suppliers benefit moreby participating in network activities because tacitknowledge is typically more valuable than explicitknowledge,. Although the network is becomingan institutionalized knowledge-sharing network, itis still less ‘mature’ than Toyota’s supplier net-work in Japan. U.S. suppliers are less dependenton Toyota than Japanese suppliers (where affili-ated suppliers sell roughly 60 percent of theiroutput to Toyota). As a result, it is easier forToyota’s Japanese suppliers to ‘identify’ with theToyota network and participate in knowledge-sharing activities.

Network-vs. organization-level learning

One question worth briefly addressing is whetheror not it is possible to distinguish betweenorganization-vs. network-level learning. Webelieve there is a difference. We view networklearning as (a) knowledge development andacquisition that is useful in aspecific networkcontext, or (b) knowledge (e.g., a best practice)that is developed or resides within the networkthat is discovered and documented/codified bya network-level knowledge storage mechanism,such as within OMCD or the supplier associ-ation. In the first case, new knowledge withregard to how best to coordinate with other firmsin the network during the product design phasewould be considered network-specific knowledge(and knowledge that is only valuable whenmultiple firms share the same knowledge). Inthe second case, knowledge acquired, stored,and diffused by the supplier association qualitycommittee (e.g., through its quality trainingprogram) would constitute network-level learn-ing. This activity stores knowledge at the net-work level, and the knowledge is then madewidely available for individual member firms touse in changing their individual firm practices.Thus, the changes (learning) that take place atthe individual firm level are due to participationin network-level learning activities. We vieworganizational learning as simply referring tochanges/adaptations in an individual firm’sprocesses/routines based on knowledge it gener-ates internally, or on knowledge that it hasacquired externally on its own.


CONCLUSION

Our study has a number of implications for boththeory and practice. First, our study offers empiri-cal support for the relational view (Dyer andSingh, 1998) which argues that the ‘network’is an important unit of analysis for explainingcompetitive advantage. Our study demonstratesthat a network can be more effective than a firmat the generation, transfer, and recombination ofknowledge. The primary reason that a network issuperior to a firm is that there is greater diversityof knowledge within a network than a firm. Kogut(2000) refers to this as greater ‘variety gener-ation’ and Gulati and Lawrence (1999) refer tothis as ‘differentiation.’ However, if the networkis to be effective at knowledge management, itmust create ‘coordinating principles’ that ‘supportcoordination among specialized firms’ (Kogut,2000). Toyota does this by effectively creatingand maintaining an ‘identity’ for the network aswell as an infrastructure that supports knowledgetransfers among suppliers. Toyota’s ability toeffectively create and manage network-levelknowledge-sharing processes at least partiallyexplains the relative productivity advantagesenjoyed by Toyota and its suppliers (Liebermanet al., 1997; Dyer, 1996). Our study providesevidence that suppliers do learn more quicklyafter participating in Toyota’s network.

Second, we provide insights into the ‘coordi-nating principles’ that facilitate cooperation(knowledge transfers) among members in the net-work. More specifically, Toyota’s network hassolved three fundamental dilemmas with regardto knowledge sharing in a ‘collective’ by findingways to: (1) motivate members to participate andopenly share knowledge, (2) prevent membersfrom free riding, and (3) efficiently transfer bothexplicit and (most importantly) tacit knowledge.Toyota has done this by creating a highly inter-connected, strong tie network—a network wheremembers strongly identify with the ‘corefirm’/network and where there are clear rules forparticipation in the network’s knowledge-sharingactivities. Perhaps most importantly, productionknowledge is viewed as the property of the net-work rather than the individual firm.

Third, our research supports Rowley, Behrens,and Krackhardt’s (2000) assertion that a highlyinterconnected, strong tie network is well suitedfor the diffusion (exploitation) of existing knowl-


edge rather than exploration for new knowledge(which is the strength of a ‘weak tie’ network).Toyota’s network is largely designed to exploitToyota’s production know-how (e.g., the ToyotaProduction System) as well as the existing diver-sity of knowledge that resides within its suppliers.Moreover, a highly interconnected, strong tie net-work is effective at the diffusion of tacit knowl-edge because (1) the redundant ties make it easierfor network members to locate potentially valu-able knowledge, and (2) strong ties produce thetrust (social capital) necessary to facilitate thetransfer of tacit knowledge.

Fourth, we acknowledge that there are risksto Toyota’s approach. First, there is the risk thatthe diversity of knowledge that resides in thenetwork will diminish over time. As firms inthe network become increasingly alike throughimitation, the network may be less effective atgenerating new knowledge. The creation of anidentity, or integration, at the network level doesnot come without a cost (Kogut, 2000; Gulatiand Lawrence, 1999). Second, as Afuah’s (2000)work suggests, the network may become soinwardly focused that it will be unable torespond to, or adopt, major technological inno-vations that occur in firms outside of the net-work. The present highly interconnected networkstructure seems ideally suited for diffusion ofexisting knowledge in a mature industry ratherthan generating new knowledge in a fast-paced,technologically dynamic industry. However,Toyota has adopted some mechanisms to main-tain knowledge diversity (e.g., rotating jishukenmembership, scanning for ‘best practices’ out-side of the industry through supplier associationcommittee activities). Further, the very processof interactions between firms creates some newknowledge, or at least new applications of exist-ing knowledge.

Fifth, our study provides insights into the evo-lution of a network, from a collection of dyadic‘weak’ ties to a weblike structure with strong,multilateral ties. The evolution of Toyota’s net-work suggests that it takes time to nurture therelationships and processes necessary to facilitateeffective learning. The evolution of Toyota’s net-work may explain why Anand and Khanna (2000)found that joint ventures designed for a learningpurpose (e.g., R&D alliances) have a ‘learningeffect’ whereas other types of alliances do not.Over time, firms develop the capabilities and


processes necessary to facilitate knowledge flowsacross firm borders (Kale, Singh and Perlmutter,2000). Also, unlike Hite and Hesterley (1999),who find that entrepreneurial networks evolvefrom strong ties to weak ties, we find the oppositein Toyota’s case. As a large firm with a goodreputation and a stock of technical, commercial,and social capital (Ahuja, 2000), Toyota has theluxury of selecting its ‘partners’ from among themost capable in the world. It then has the chal-lenge of creating strong ties with, and among,those partners.

We see a couple of important directions forfuture research. First, a comparative study of asample of different vertical networks with dif-fering degrees of success at knowledge sharingwould allow for tests of the ideas offered in thisstudy. Second, future research might examinewhy Toyota’s competitors have not yet beenable to imitate the institutionalized knowledge-sharing processes that reside within the Toyotaproduction network. Why aren’t other automak-ers learning from Toyota? This is an importantquestion given that our research demonstratesthat network-level knowledge-sharing routinesare a critical factor in explaining Toyota’s andits suppliers’ success. One possibility offered byGarcia-Pont and Nohria (1999) is that the historyof cooperation and competition in an industryguides alliance behavior. Perhaps other automak-ers, particularly U.S. automakers, are constrainedby their past history of adversarial relationshipswith suppliers. Whatever the reason, Toyota doesnot seem to be particularly concerned. As oneToyota executive observed, ‘We are not so con-cerned that our knowledge will spillover to com-petitors. Some of it will. But by the time itdoes, we will be somewhere else. We are amoving target.’ This executive is expressing theidea that Toyota’s advantage is sustainablebecause Toyota and its suppliers have a‘dynamic learning capability’ and learn at a fas-ter rate than competitors. Our study suggeststhat the notion of a dynamic learning capabilitythat creates competitive advantage needs to beextended beyond firm boundaries. Indeed, if thenetwork can create a strong identity and effectivecoordinating rules, then it may be superior to afirm as an organizational form at creating andrecombining knowledge owing to the greaterdiversity of knowledge that resides within anetwork.


ACKNOWLEDGEMENTS

The authors would like to thank Masue Suzukifor her able research assistance. We would alsolike to thank the following individuals at Toyotawho were especially helpful throughout theresearch project: Michio Tanaka (General Man-ager, International Purchasing); Nobuhiko Suzuki(Manager, International Purchasing); NoriyukiYokouchi (V.P. Purchasing); Chris Nielsen(Assistant General Manager, PurchasingPlanning), Hajime Ohba (General Manager,TSSC); and Lesa Nichols (Assistant Manager,TSSC). We would also like to thank MarvinLieberman for sharing his data on Japanese andU.S. automotive labor productivity. Finally, theauthors would like to thank the Huntsman Centerfor Global Competition and Innovation at theWharton School and the International MotorVehicle Program at MIT for generously support-ing this research.

REFERENCES

Afuah, A. (2000). ‘How much do your co-opetitors’capabilities matter in the face of technologicalchange?’, Strategic Management Journal, SpecialIssue,21, pp. 387–404.

Ahuja, G. (2000). ‘The duality of collaboration: Induce-ments and opportunities in the formation of interfirmlinkages’, Strategic Management Journal, SpecialIssue,21, pp. 317–343.

Anand, B. N. and T. Khanna (2000). ‘Do firms learn tocreate value? The case of alliances’Strategic Manage-ment Journal, Special Issue,21, pp. 295–315.

Browning, L., J. Beyer and J. Shetler (1995). ‘Buildingcooperation in a competitive industry: Sematech andthe semiconductor industry’,Academy of Manage-ment Journal, 38(1), pp. 113–151.

Burgelman, R. A. (1994). ‘Fading memories: A processtheory of strategic business exit in dynamic environ-ments’, Administrative Science Quarterly, 39(1),pp. 24–56.

Burt, R. (1992).Structural Holes: The Social Structureof Competition. Harvard University Press, Cam-bridge, MA.

Coase, R. H. (1937). ‘The nature of the firm’,Eco-nomica, 4, pp. 386–405.

Cohen, W. M. and D. A. Levinthal (1990). ‘Absorptivecapacity: A new perspective on learning and inno-vation’, Administrative Science Quarterly, 35,pp. 128–152.

Conner, K. and C. K. Prahalad (1996). ‘A resource-based theory of the firm: Knowledge versus oppor-tunism’, Organization Science, 7(5), pp. 477–501.

Cusumano, M. A. (1985).The Japanese Automobile


Industry: Technology and Management at Nissanand Toyota. Harvard University—Council on EastAsian Studies, Cambridge, MA.

Dyer, J. H. (1996). ‘Specialized supplier networks asa source of competitive advantage: Evidence fromthe auto industry’,Strategic Management Journal,17(4), pp. 271–292.

Dyer, J. H. and W. G. Ouchi (1993). ‘Japanese stylebusiness partnerships: Giving companies a competi-tive edge’, Sloan Management Review, 35(1),pp. 51–63.

Dyer, J. H. and H. Singh (1998). ‘The relational view:Cooperative strategy and sources of inter-organizational competitive advantage’,Academy ofManagement Review, 23(4), pp. 660–679.

Eisenhardt, K. (1989). ‘Building theories from caseresearch’, Academy of Management Review, 14,pp. 532–550.

Garcia-Pont, C. and N. Nohria (1999). ‘Local versusglobal mimetism: The dynamics of alliance forma-tion in the automobile industry’, paper presented atthe SMJ Special Issue Conference on StrategicNetworks.

Gerlach, M. L. (1992).Alliance Capitalism. Universityof California Press, Berkeley, CA.

Glaser, B. and A. Strauss (1967).The Discovery ofGrounded Theory, Aldine, Chicago, IL.

Gomes-Casseres, B. (1994). ‘Group versus group: Howalliance networks compete’,Harvard BusinessReview, 72(4), pp. 4–11.

Gouldner, A. W. (1960). ‘The norm of reciprocity: Apreliminary statement’, American SociologicalReview, 25(1), pp. 161–178.

Grant, R. (1996). ‘Prospering in dynamically-competitive environments: Organizational capabilityas knowledge integration,’Organization Science,7(4), pp. 375–387.

Gray, B. (1989). Collaborating: Finding CommonGround for Multiparty Problems. Jossey-Bass, SanFrancisco, CA.

Gulati, R. and P. Lawrence (1999). ‘Organizing verticalnetworks: A design perspective,’ paper presented atthe SMJ Special Issue Conference on StrategicNetworks.

Hite, J. and W. Hesterly (1999). ‘The influence of thefirm life cycle on the evolution of entrepreneurialfirm networks’, paper presented at theSMJ SpecialIssue Conference on Strategic Networks.

Kale, P., H. Singh and H. Perlmutter (2000). ‘Learningand protection of proprietary assets in strategicalliances: Building relational capital’Strategic Man-agement JournalSpecial Issue21, pp. 217–237.

Kogut, B. (2000). ‘The network as knowledge: Genera-tive rules and the emergence of structure’, StrategicManagement Journal,21, Special Issue, pp. 405–425.

Kogut, B. and U. Zander (1992). ‘Knowledge of thefirm, combinative capabilities, and the replication oftechnology’, Organization Science, 3(3), pp. 383–397.

Kogut, B. and U. Zander (1996). ‘What firms do?Coordination, identity, and learning,’OrganizationScience, 7(5), pp. 502–518.


Levinson, N. S. and M. Asahi (1996). ‘Cross-nationalalliances and interorganizational learning,’Organiza-tional Dynamics, 24, pp. 51–63.

Levitt, B. and J. G. March (1988). ‘Organizationallearning,’ Annual Review of Sociology, 14,pp. 319–340.

Lieberman, M. (1994). ‘The diffusion of “lean manu-facturing” in the Japanese and U.S. automotiveindustry’, presented at ‘New Imperatives for Manag-ing in Revolutionary Change’ Conference, Shizuoka,Japan, 28–30 August.

Lieberman, M. and S. Asaba (1997). ‘Inventoryreduction and productivity growth: A comparison ofJapanese and U. S. automotive sectors’,Managerialand Decision Sciences, 18, pp. 73–85.

Lieberman, M., M. Sako, K. Wada and L. Demeester(1997). ‘The productivity-enhancing impact of sup-pliers associations in the Japanese auto industry’,IMVP Sponsors Forum, Kyung Ju, Korea, 25 Sep-tember.

Lincoln, J. R., M. Gerlach and P. Takahashi (1992).‘Keiretsunetworks in the Japanese economy: A dyadanalysis of intercorporate ties’,American Sociologi-cal Review, 57, pp. 361–585.

Lorenzoni, G. and A. Lipparini (1999). ‘The leveragingof interfirm relationships as a distinctive organi-zational capability: A longitudinal study’,StrategicManagement Journal, 20(4), pp. 317–338.

MacDuffie, J. P. and S. Helper (1997). ‘Creating leansuppliers: Diffusing lean production throughout thesupply chain’, California Management Review, 39(4), pp. 118–134.

March, J. G. and H. A. Simon (1958).Organizations.Wiley, New York.

Marsden, P. V. (1990). ‘Network data and measure-ment’, Annual Review of Sociology, 16, pp. 435–463.

Marwell, G. and P. Oliver (1993).The Critical Massof Collective Action: A Micro-social Theory. Cam-bridge University Press, Cambridge, U.K.

Nelson, R. and S. Winter (1982).An EvolutionaryTheory of Economic Change. Harvard UniversityPress, Cambridge, MA.

Nishiguchi, T. (1994).Strategic Industrial Sourcing.Oxford University Press, New York.


Nonaka, I. (1994). ‘A dynamic theory of organizationalknowledge’,Organization Science, 5, pp. 14–37.

Powell, W. W. (1996). ‘Inter-organizational collabora-tion in the biotechnology industry’,Journal of Insti-tutional and Theoretical Economics, 152, pp. 197–225.

Powell, W. W., K. W. Koput and L. Smith-Doerr(1996). ‘Interorganizational collaboration and thelocus of innovation: Networks of learning in biotech-nology’, Administerative Science Quarterly, 41,pp. 116–145.

Rowley, T. D. Behrens and D. Krackhardt (2000).‘Redundant governance structures: An analysis ofstructural and relational embeddedness in the steeland semiconductor industries’,Strategic Manage-ment Journal, Special Issue,21, pp. 369–386.

Ryle, G. (1984).The Concept of Mind. University ofChicago Press, Chicago, pp. 29–34.

Sako, M. (1997). ‘Knowledge diffusion in the U.K.automotive industry’, IMVP Sponsors’ Forum,Kyung Ju, Korea, 25 September.

Sandler, T. (1992).Collective Action: Theory andApplications. University of Michigan Press, AnnArbor, MI.

Spender, J.-C. (1996). ‘Making knowledge the basis ofa dynamic theory of the firm’,Strategic ManagementJournal, Winter Special Issue,17, pp. 45–62.

Strauss, A. and J. Corbin (1990).Basics of QualitativeResearch. Sage, Newbury Park.

Szulanski, G. (1996). ‘Exploring internal stickiness:Impediments to the transfer of best practice withinthe firm’, Strategic Management Journal, WinterSpecial Issue,17, pp. 27–43.

Teece, D. J., G. Pisano and A. Shuen (1997) ‘Dynamiccapabilities and strategic management’,StrategicManagement Journal, 18(7), pp. 509–533.

von Hippel, E. (1988).The Sources of Innovation.Oxford University Press, New York.

Webb, E. J. (1981).Nonreactive Measures in the SocialSciences. Houghton Mifflin, Boston, MA.

Williamson, O. E. (1985).The Economic Institutionsof Capitalism. Free Press, New York.

Wood, D. J. and B. Gray (1991). ‘Toward a compre-hensive theory of collaboration’,Journal of AppliedBehavioral Science, 27(2), pp. 139–162.

creating and managing a high-performance knowledge-sharing network: the toyota case

Documents