How Much You Know Versus How Well I Know You

8/14/2019 How Much You Know Versus How Well I Know You

1/22

Strategic Management JournalStrat. Mgmt. J., 26: 7596 (2005)

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/smj.453

HOW MUCH YOU KNOW VERSUS HOW WELL

I KNOW YOU: SELECTING A SUPPLIER FOR A

TECHNICALLY INNOVATIVE COMPONENT

GLENN HOETKER*

College of Business, University of Illinois at Urbana Champaign, Champaign, Illinois,U.S.A

How do firms select a supplier for an innovative component? Three literatures speak to thisquestion. Transaction cost economics focuses on the value of internalization, the literature oninter-firm relationships on the value of past relationships, and the firm capabilities literatureon the value of superior capabilities. Choosing a supplier means choosing a bundle of thesecharacteristicsinternal vs. external, amount of prior transactions, and capabilitiesbut nostudy has integrated all three characteristics, making it impossible to understand the trade-offsinvolved either theoretically or managerially. I propose and test a model integrating all three

factors, allowing us to understand the trade-offs involved. Ifind that when uncertainty is low, thedecision is made primarily on the basis of differences in technical capabilities. As uncertaintyincreases, prior relationships and a supplier being internal take on greater positive significancerelative to the importance of technical capabilities. At extreme levels of uncertainty, the valueof internal supply relationships becomes very high and past relationships lose their significance.While adherents of each literature have criticized the others for what they omit, this model movesbeyond this mutual recrimination by incorporating the key concerns of each literature, settingthe stage for future research that draws upon the strength of each. The model provides guidance

for any situation where a firm must chose a partner under uncertainty. Lastly, it addresses thestrategic question of how companies should organize in the long run to access the capabilitiesneeded for competitive success. Copyright 2004 John Wiley & Sons, Ltd.

INTRODUCTION

How do firms select a supplier for an innova-tive component? This question is critical becausefirms often seek to improve a product by incor-porating innovative components, i.e., componentsthat do not yet exist in their desired form. For

example, a laptop computer manufacturer can

Keywords: buyersupplier relationships; capabilities;transaction cost economics; technology management*Correspondence to: Glenn Hoetker, College of Business, Uni-versity of Illinois at UrbanaChampaign, 350 Wohlers Hall,1206 South Sixth Street, Champaign, IL 61820, U.S.A. E-mail: [email protected]

attempt to improve its computer by incorporat-ing a larger, higher-resolution display than cur-rently exists. However, it cannot buy the inno-vative display off the shelf. Developing a newdisplay will require a period of intense interactionbetween buyer and supplier. Thus, selecting theright supplierpossibly an internal supplieris

an important strategic decision.Three literatures speak to the issue of supplierchoice. Transaction cost economics (Williamson,1975) focuses on the effect of some character-istic of the transaction, e.g., uncertainty, on theassociated governance costs. It then asks how oneaspect of the buyersupplier relationship, internalvs. external, magnifies or diminishes that effect.A significant strand of the literature on inter-firm

Copyright 2004 John Wiley & Sons, Ltd. Received 6 November 2002

Final revision received 19 March 2004


2/22

76 G. Hoetker

relationships (e.g., Gulati, 1995; Uzzi, 1996) asksthe same question, but looks at a different aspect of

the buyersupplier relationship: prior transactions.The literature on firm capabilities (Barney, 1991;Winter, 1987) points to the value of a supplierstechnical capabilities.

Choosing a supplier means choosing a bundle ofthese characteristics: internal vs. external, amountof prior transactions, and technical capabilities.Of course, the supplier with the greatest technicalcapabilities may not be the supplier with which thebuyer has the strongest relationship, but no studyhas integrated all three characteristics into a sin-gle model. This makes it impossible to understand

the trade-offs involved from either a theoretical ormanagerial viewpoint. Transaction cost economicsand the literature on inter-firm relationships informus about the strength of the buyersupplier rela-tionship, but not how it is balanced against tech-nical capabilities. The reverse is true of the firmcapabilities literature. Also, neither transaction costeconomics nor the inter-firm relationship literaturecan help us understand when a long-term relation-ship outweighs a hierarchical relationship with asupplier.

This paper proposes and tests a model that inte-grates all three factors in a single framework. Thisallows me to make three primary contributions.

First, the model provides a more complete,nuanced, and accurate understanding of how firmsresolve the inevitable trade-offs between internal-ization, past relationships, and capabilities in a waythat no one of the literatures can do in isolation.Doing so helps resolve some of the intransigenttensions between these three literatures.

From the viewpoint of transaction cost eco-nomics, much of the literature on firm capabili-ties suffers from a lack of attention to the poten-tial for opportunism (Foss, 1996; Mahoney, 2001;

Williamson, 1999). The firm capabilities litera-ture argues that transaction cost economics ignoresor underemphasizes differences in firm capabili-ties, such as their technical capabilities (Winter,1988). The literature on inter-firm relationships hasargued that transaction cost economics overlooksthe fact that past transactions between the buyerand supplier may generate processes that alter thecalculus for future transactions (Gulati, 1995; Ringand Van de Ven, 1994). Since a buyer will have adifferent history of interaction with each potentialsupplier, the expected costs of dealing with eachsupplier will differ.

Each of these concerns has some validity. Themodel I develop makes it possible to move beyond

simply pointing out what each literature fails toconsider. Rather, it incorporates the key concernsof each literature, recognizing that the relativeimportance of each will vary according to thesituation under consideration. This responds toDenzins (1989) call to bring diverse theories tobear on a phenomenon.

The papers second contribution is to providemanagers with a richer framework than previouslyexisted to guide their selection of suppliers. Themodel helps answer a question managers mustanswer on a regular basis: not Should I make or

buy this component? but From whom, inside oroutside of my firm, should I buy this? It alsoapplies to any situation in which a firm mustchoose a partner under uncertainty. With little orno modification, it can inform the choice of a jointventure partner or a partner for entry into a newgeographic market.

The third contribution is to address an impor-tant strategic issue in a way that the underlyingliteratures cannot: how should companies orga-nize in the long-run to access the capabilitiesneeded for competitive success. For example, anotebook computer manufacturer needs to haveongoing access to the capabilities necessary todevelop new flat-panel displays. It can developthese internally, develop long-term ties with capa-ble suppliers, assume it will be able to access themon the open market, or combine these strategies.Each approach carries its own combination of costsand risks. Our ability to understand which strategycarries the best balance of cost and risk is lim-ited by the fact that no single literature speaks toall of the issues underlying each of these strate-gies. By combining the insights of the underly-ing literatures, the integrated model and associated

empirical results provide guidance on the optimalstrategy.

The next section reviews theories related to tech-nical capabilities and buyersupplier relationshipsand uses these concepts to construct an integratedmodel of supplier choice. I then test hypothesesderived from the model using information on theprocurement of innovative flat-panel displays bynotebook computer manufacturers in Japan andthe United States. I test the hypotheses using alarge-scale empirical study, supplemented by mul-tiple interviews. After presenting and interpretingthe results, I show that the integrated model is

Copyright 2004 John Wiley & Sons, Ltd. Strat. Mgmt. J., 26: 7596 (2005)


3/22

Supplier Choice for a Technically Innovative Component 77

a significantly more robust predictor of supplierchoice than any of the underlying theoretical lens

applied individually. I then discuss the robustnessand potential limitations of the study. A discussionof the papers contributions concludes.

THEORY AND MODEL

Differences in both the technical capabilities ofsuppliers and in their relationships with the buyeraffect the calculus of supplier selection. However,

determining their relative importance requires com-bining them into a single model.To do so, I take as a starting point the reduced-

form model of the governance decision widelyused in transaction cost economics (Williamson,1996 : 106109). Transaction cost economics isnot more fundamental to my arguments than thecapabilities and inter-firm relationship literatures,but offers a formal model to use as a foundation forbuilding an integrated model of supplier selection.

As a baseline, consider Figure 1, in which abuyer can procure an innovative component froman internal (I) or external (E1) supplier. The hori-zontal axis represents the degree of technological

Technological uncertainty

Totalcosts

Internal supplier I External supplier E1

Figure 1. A model of supplier selection

uncertainty posed by the desired innovation.1 LinesI and E1 map the total costs the buyer incurs by

choosing that supplier at a given level of techno-logical uncertainty. Total costs include the costsof producing the component, adjusted for qualityand the likelihood of successful development, gov-erning the relationship to avoid potential oppor-tunism, and communicating during the develop-ment process.

Following the practice of transaction cost eco-nomics, I assume firms organize their transactionsto minimize total costs. Therefore, on the left sideof the graph, when uncertainty is low and the exter-nal supplier offers lower total costs, the buyer will

choose the external supplier. On the right side ofthe graph, uncertainty is high and the internal sup-plier offers lower total costs and is preferred. NextI discuss factors that influence the position of thecost lines.

Uncertainty

The cost lines slope upwards because greateruncertainty increases the costs of governance andcommunication over the development cycle for aninnovative component. In the context of techno-

logical innovation, uncertainty will be high whenthe desired innovation requires large advancesbeyond the current technological frontier. UsingGalbraiths (1977) definition of uncertainty, thereis a large difference between the amount of infor-mation required to develop the innovation andthe amount of information already possessed. InWilliamsons (1975) terms, because much infor-mation is missing, it will be difficult for a decision-maker to specify a complete decision tree.

This uncertainty increases communicationscosts. The technology is likely to be in a stateof flux (Teece, 1977: 249) and performance

and design requirements are often unknownex ante (Monteverde and Teece, 1982). Thiscondition requires ongoing communication asthe development process reveals unforeseendifficulties and information on the innovationsactual potential. This communication will bedifficult because information at or beyond thetechnological frontier is especially likely tobe uncodified and difficult to transfer between

1 In many applications of this model, the horizontal axis rep-resents asset specificity. I address the issue of asset specificitybelow.



4/22

78 G. Hoetker

individuals or groups (Hansen, 1999; Nelson andWinter, 1982).

Uncertainty also makes governance more dif-ficult. The buyer and supplier are more likelyto disagree about the appropriate technologicalapproach (Argyres, 1995) and to have difficultyformulating R&D goals (Armour and Teece, 1980).Once development is underway, the probability ofencountering a contingency that renders the initialtechnical approach inappropriate is high (Masten,1984). In response, the buyer and supplier caneither draft a contract covering a wide range ofcontingencies, which is costly and will never becomplete, or engage in costly renegotiation each

time an unanticipated contingency develops. In theevent of failure, it is difficult for the contractingparties (Arrow, 1971; Davidson and McFetridge,1984) or the courts (Masten, 1984) to determinefault. Studies in industries including aerospace(Masten, 1984), telecommunications (Globerman,1980), and electric components (Anderson andSchmittlein, 1984) have found evidence that uncer-tainty increases governance costs.2

Technical capabilities

Potential suppliers may differ in their technicalcapabilities and, thus, in their ability to produce acomponent according to the desired specificationsand schedule. Firms develop capabilities by car-rying out related activities repeatedly (Nelson andWinter, 1982; Nonaka and Takeuchi, 1995). Differ-ences in past activities thus lead to heterogeneouscapabilities.

Of course, firms can and do add capabilities.However, doing so is time-consuming, costly, andoften risky, leading Teece and co-authors to statethat it is rarely an attractive proposition to try

and develop a collection of novel skills rapidly

2 Like the above cited studies, I focus on uncertainty, ratherthan frequency and asset specificity, the other two dimensionsof transactions normally considered by transaction costs analy-sis (Williamson, 1985: 79). Because I focus on the same buyers,suppliers, and technology, frequency does not vary across trans-actions. Asset specificity may co-vary with uncertainty, sincethe more uncertainties there are to be resolved, the more partner-specific solutions will be involved. However, interviews indicatethat beyond the specificity present in any innovative display,increasing uncertainty impedes communication and the ability tocontract for technical contingencies much more than it increasesasset specificity. Therefore, I focus on the primary impact ofuncertainty.

(Teece et al., 1994: 17). Firms should thus eval-uate potential suppliers on the basis of their cur-

rent capabilities (Henderson and Cockburn, 1994;Kamien and Schwartz, 1982). If a firm choosesa supplier other than the most technically capa-ble, either development will proceed with inferiorcapabilities, or investments of time and resourceswill be needed to improve the capabilities of thechosen supplier.

To incorporate technical capabilities into thesupplier selection model, I observe that as tech-nical uncertainty approaches zero, communicationand governance costs decline and make up anincreasingly smaller portion of the total cost. Pro-

duction costs, a function of the suppliers techni-cal capabilities, become the main driver of differ-ences in total costs. Therefore, technical capabil-ities enter the model via the y-intercept of eachsuppliers cost line.

Figure 2 introduces a second external supplier,E2, identical to the original external supplier, E1,except for having greater technical capabilitiesand, consequently, a lower y-intercept. Since theyare identical by construction in all other respects,E2 always offers lower total costs than, and willbe preferable to, E1. To capitalize on E2s greatertechnical capabilities, the buyer delays moving tothe internal supplier until technical uncertainty ishigher than previously.3 This leads to my firsthypothesis.

Hypothesis 1: The greater a suppliers technical

capabilities, the more likely a buyer is to select

that supplier.

While this hypothesis is highly plausible, it isimportant not only as a baseline, but also becauseit addresses an important point of tension between

the capabilities literature and transaction cost eco-nomics. Adherents of the firm capabilities litera-ture have criticized transaction cost analysis forignoring or underemphasizing such differences infirms capabilities (Barney, 1999; Winter, 1988).

3 In this and all of the illustrative figures, there is nothingdeterministic about the relative positions of the various suppliers.In Figure 2, it would be equally possible for the internal supplierto have technical capabilities superior to those of either externalsupplier. In that case, it would always offer the lowest totalcosts and the buyer would never choose an external supplier.I have chosen situations that are most useful in illustrating theprinciples in question.



5/22



Totalcosts


External supplier E2

Figure 2. The impact of differing technical capabilities

Theoretically, transaction cost economics en-compasses any factor, including technical capabil-ities, that affects either governance or productioncosts, but the critique is valid in terms of empiricalresearch. A few early empirical studies controlledfor differences in production experience and pro-duction costs between buyer and supplier (e.g.,Walker and Weber, 1984). However, as transactioncost economics moved forward, empirical worklargely set aside differences between firms to focuson cost comparisons between different modes oforganizing a given transaction (Williamson, 1999).In a rare exception, Argyres (1996) case studyfound that both governance issues and capabilitiesaffected the decision to make or buy a component.

The transaction cost literature has recentlyrediscovered the importance of capabilities, withWilliamson identifying firm capabilities as theobvious next element to be incorporated intothe calculus of transaction cost economics(Williamson, 1999). To do so, he suggestscomplementing the transaction cost frameworkwith the thread of the firm competence literaturethat focuses on the role of routines (e.g., Nelsonand Winter, 1982), as I do here. At the same time,Foss (1996) and Mahoney (2001) have argued theconverse, that the origin and impact of differentialfirm capabilities cannot be understood without

taking opportunism into account per the transactioncost economics literature. However, to date, little

empirical work has answered this call.

The impact of the buyersupplier relationship

A buyer must also consider the costs of manag-ing the relationship over the development pro-cess. Unlike existing goods, innovations cannotbe obtained by a one-off, anonymous transactionand require time to develop. During that time,the buyer and supplierpossibly divisions of thesame firm must interact to develop mutuallyacceptable specifications (Monteverde and Teece,

1982) and to design the final product to incorporatethe innovative component. This interaction poseschallenges of communication and governance.

A large literature has examined the impact ofpast relationships on the success with which abuyer and supplier can manage these challenges.Sustained relationships have been found toincrease suppliers innovativeness (Helper, 1991a)and involvement in product development (Wastiand Liker, 1999), as well as leading tofaster, lower-cost procurement (Dyer 1996a). Thisliterature has been critical of transaction costeconomics practice of treating each transactionbetween a buyer and supplier as discrete (Dozand Prahalad, 1991; Ring and Van de Ven, 1994).Nooteboom (1992) summarized these concernswhen he wrote that An embedding of a transactionin an ongoing process of exchange is required tomake transaction cost economics coherent.

Past interactions improve communication be-tween buyer and supplier in several ways. First,because the buyer and supplier have first-handknowledge of each others capabilities, they canmore effectively partition tasks so that the mostcapable party is assigned each task (Fichman

and Levinthal, 1991). Second, relationship-specificcommunication and coordination routines developover time (Mitchell and Singh, 1996). Third, abuyer and supplier will, over the course of multipleinteractions, develop a common language for dis-cussing technical and design issues (Buckley andCasson, 1976).

Past interactions also reduce the cost of gov-ernance by building trust, by which I mean theexpectation that ones exchange partner will not actopportunistically for one or more social, psycho-logical, or economic reasons. Although the futurebehavior of a partner is unobservable, first-hand



6/22

80 G. Hoetker

knowledge of a partners past behavior providessome information (Granovetter, 1985). Crocker

and Reynolds (1993) found that the U.S. Air Forcetook fewer steps to protect itself from opportunis-tic behavior by an engine supplier if that supplierhad refrained from opportunistic behavior in thepast.

Trust also develops between individuals engagedin repeated transactions (Uzzi, 1996). This trustcan become institutionalized, leading to trustbetween the organizations that endures despitechanges in the individuals involved (Zaheer,McEvily, and Perrone, 1998).

Dyers (1994, 1996b) studies of supplier net-

works in the Japanese automotive industry showthat these relationship-specific communication andgovernance efficiencies can be very valuable. If thesupplier relationship were to end, this value wouldbe lost, which serves as a brake on opportunismfor both parties (Williamson, 1983). So long as thebenefit of opportunism does not exceed the valueof the relationship, neither party is likely to risk thelong-term value of the relationship for short-termgain, either by blatant acts of opportunism suchas breach of contract or more subtle actions suchas below-peak performance (Klein, 1980; Pisano,

1989). If the buyer and supplier believe that unex-pected contingencies will be resolved with neitherparty taking opportunistic advantage, developmentof the innovative component can occur under theaegis of an agreement short of an elaborate con-tract, reducing ex ante contracting costs and theneed for costly formal control mechanisms (Pow-ell, 1990).

Each of these communication and governanceefficiencies increases as a buyer and supplier worktogether over time. Helper (1991a) and Sako andHelper (1998) found that trust between buyer and

suppliers increased the longer they had workedtogether. Mutual knowledge, inter-firm routines,and a common language develop in the courseof repeated interaction of people from the partnerfirms (Ring and Van de Ven, 1994), increasing thevalue of the relationship and serving as a strongerbrake on opportunism.

Thus, a buyer and supplier should be better ableto address challenges of communication and gov-ernance the more they have transacted in the past.Empirically, Parkhe (1993) found that a prior his-tory of cooperation between two firms reducedtheir expectations of opportunism, while Heide and

Miner (1992) found that buyersupplier coopera-tion increased with higher frequency of contact in

the existing relationship.Despite these advantages, past transactions with

a supplier do not necessarily make that supplierthe most attractive choice. The value of past trans-actions depends on the technological uncertaintyposed by the desired innovation, a fact that canbe represented graphically. In Figure 3, supplierE2 still has superior technical capabilities, but E1is now posited to have transacted with the buyersignificantly more than E2 has. Because the buyerand E1 have better mechanisms for managing thegovernance and communications challenges that

accompany higher uncertainty, E1s costs increasemore slowly with increasing uncertainty than doE2s. The slope of line E1 is flatter than that of E2.

At low levels of uncertainty, there are few com-munication or governance costs and the total costconsists primarily of production costs. Since dif-ferences in technical capabilities determine dif-ferences in total cost, the difference in technicalcapabilities overwhelms differences in past trans-actions. The buyer will choose supplier E2, whichoffers the lowest total cost.

As uncertainty increases, communication and

governance costs comprise an increasing proportion


Totalcosts


External supplier E2

Figure 3. The impact of differing technical capabilitiesand buyersupplier relationships



7/22


of the total cost. The advantages of prior transac-tions outweigh differences in technical capabili-

ties and E1 offers the lowest total costs. In thisrange of uncertainty, the buyer will prefer a tech-nically mediocre, long-term supplier to a techni-cally superior supplier with whom it has not dealtextensively.4

Furthermore, the benefits of long-term relation-ships may come at the cost of diminished incen-tives. Parties in long-term relationships may givesecond chances more frequently, expect due pro-cess before termination, and are more willing tonegotiate unexpected cost increases. For example,Uzzi (1997: 43) quotes a contractor as saying With

people you trust, you know that if you have a prob-lem with a fabric theyre just not going to say, Iwont pay or take it back. If they did then wewould have to pay for the loss.

Therefore, I make no prediction for their maineffect of prior transactions. Their benefits aresmall under low uncertainty and may even be out-weighed by the impact of diminished incentives.However, the benefits should become more impor-tant as the uncertainty of the desired innovationincreases, increasing the positive impact of pasttransactions on the probability of a supplier beingchosen. Consistent with this prediction, Podolny

(1994) found evidence of this increasingly posi-tive effect in investment bankers choice of co-managers for debt offerings. In contrast, Wasti andLikers study of the U.S. and Japanese automotiveindustries found the effect of experience on thedegree of supplier involvement in product designwas positive, but did not increase with uncertainty(Wasti and Liker, 1999).

Hypothesis 2: The more uncertain the desired

innovation, the more positive will be the effect

of prior transactions between a buyer and sup-

plier on the likelihood of a buyer choosingthat supplier.

Internal suppliers offer many of the same advan-tages as long-term suppliers, although for differ-ent reasons. Transaction cost economics arguesthat when the buyer and supplier belong to thesame firm, challenges in governance are mitigated.

4 Note that I am not arguing that technical capabilities mightnot also take on greater importance with increasing uncertainty.Rather I argue that since communication and governance costscomprise an increasing portion of the total cost, the relative valueof relationships becomes greater as uncertainty increases.

Within the firm, governance costs are loweredby distinctive governance mechanisms (Mahoney

1992: 567569) including managerial fiat, agreater legal burden on employees as opposed tocontractors to disclose information (Masten, 1988),and judicial forbearance (Williamson, 1991).

Belonging to the same firm can also reducecommunication costs. Frequent, intense communi-cation within the firm leads to the development ofcommunication routines and a common languagefor describing technical issues (Nelson and Winter,1982). Of course, intense and repeated interac-tion could occur between firms. However, Sosaet al. (2002) found that, within geographically dis-

tributed product development teams, individualsbelonging to the same organization communicatedmore than did individuals from different organi-zations. This echoes Allens (1977) finding thatmore technical communication occurred over theR&D process when individuals belonged to thesame organization. Kogut and Zander (1996: 502)argue that individuals identification with the firmgenerates social knowledge that supports coor-dination and communication, while Williamson(1985) argues that organizational integrity, due to asense of shared destiny, leads to better informationdisclosure.

This reasoning suggests that, on average, a firmcan better manage the interactions that occur overthe innovation process with an internal supplierthan an external one. Returning to Figure 3, aninternal suppliers cost line will have a flatter slopethan that of an external supplier. As uncertaintyincreases, the advantages of working with an inter-nal supplier become increasingly important. At thehighest levels of uncertainty, the extreme right inFigure 3, the internal supplier offers the lowesttotal cost and will be the preferred supplier.

Firms must weigh this advantage against the

lower incentives and greater bureaucratic costs thatoccur in within-firm transactions (Helper, 1991b;Williamson, 1985: 140 153). Graphically, lowerincentives and greater bureaucratic costs raise they-intercept of the internal suppliers cost lineabove that of an external supplier with equal tech-nical capabilities.

These two dynamics determine the likelihood ofan internal supplier being chosen. Absent uncer-tainty, bureaucracy and diminished incentivesincrease costs. At high levels of uncertainty, how-ever, the communication and governance advan-tages of working with an internal supplier offset



8/22

82 G. Hoetker

these costs. Therefore, it is not possible to deter-mine the net effect of common firm membership

independent of the level of technical uncertainty.However, the greater the uncertainty, the greatershould be the net value of a buyer and supplierbelonging to the same firm.

Hypothesis 3: The more uncertain the desired

innovation, the more likely that a firm will select

an internal supplier.

As Figure 3 illustrates, the model implies thatthe value of long-term supply relationships willbe greatest under moderate uncertainty.5 At low

uncertainty, the advantages of prior relationshipsare irrelevant. Under extremely high uncertainty,the far right of Figure 3, the advantages of work-ing with internal suppliers are at their highest. Thelarger the advantage offered by internal supply,the less impact past experience with any externalsupplier will have on the choice of supplier. Sta-tistically, this would correspond to an inverted-Ushape for the effect of prior experience.

This observation is only possible via an inte-grated model. Transaction costs economics initiallyfocused on the contrast between interactions withinand between firms. The literature on interfirm net-

works (Ahuja, 2000; Dyer, 1997, 1998; Ring andVan de Ven, 1992) subsequently drew heavily onthe logic of transaction cost economics and theknowledge-based view of the firm to understandchanges, primarily positive, that occur in commu-nication and governance as firms interact over time(Gulati, 1995). These streams of literature haverarely intersected, leaving it unclear to what degreehierarchy can provide equivalent or superior ben-efits to long-term inter-firm relationships.

One consequence of this gap has been diffi-culty in understanding the conditions under whichfi

rms should build networks of long-term relation-ships to access the capabilities they need and whenthey are best served by building these capabili-ties internally. This model suggests that long-termrelationships are most beneficial in an environ-ment of medium uncertainty, where they providea firm with efficient access to capabilities withoutthe cost of developing them internally. For a series

5 Recall that the specific configuration of values for each variablerepresented in Figure 3 illustrates the principals in question. Theprincipals discussed in this paragraph apply to any configurationof values, but the identity of the optimal supplier depends on thespecific values.

of relatively low-uncertainty innovations, relation-ships provide little benefit. Indeed, if they prevent

the firm from making selections based on techni-cal criteria, they may even be counter-productive.Assuming there are capable suppliers available,there is also less value in developing internal capa-bilities in such an environment, since the firmcan acquire the needed capabilities from suppliers,regardless of their past relationship or lack thereof.When innovations are expected to have very highuncertainty, acquiring capabilities from any exter-nal supplier, even a long-term supplier, is costly.A firm will be better served if it has developed therequired capabilities internally.

I do not formally hypothesize this relationshipfor two reasons. First, it is unclear if any obser-vations in my data approach this high a level ofuncertainty. Second, as discussed in the methodssection, the statistical model appropriate to modelsupplier choice is not conducive to testing such arelationship. However, I find evidence from bothmy statistical results and interviews that is congru-ent with this implication of the model.

DATA AND METHOD

Empirical setting

I test these hypotheses by studying notebook com-puter manufacturers sourcing decisions for inno-vative flat-panel displays from 1992 to 1998. Thissetting offers several advantages.

First, displays are a key component in notebookcomputers because they are a large part of theuser experience. To distinguish their product, note-book computer manufacturers continually striveto incorporate more advanced displays into theirnotebook models. Therefore, notebook computermanufacturers take great care in their selection ofsuppliers for innovative displays.

Second, notebook manufacturers chose suppliersunder varying degrees of uncertainty. Because ofcompetitive pressures, computer firms cannot waitfor an innovative display to exist before begin-ning to design the computer into which it will beintegrated. The race to introduce a notebook witha 13-inch display was underway when the largest



9/22


available notebook display was 12 inches. How-ever, the technical challenges to be overcome var-

ied in number and complexity across innovations.6

This makes it possible to identify low-uncertaintyand high-uncertainty innovations.

Third, the engineering effort to develop aninnovative display and integrate it into a newnotebook computer model requires the type ofmanagement-intensive buyersupplier relationshipthis model addresses. This process can easily takeover 9 months and requires continuous communi-cation between notebook maker and display sup-plier, partially because many relevant parametersare highly subjective.7 As a result, even though the

initial specifications from the notebook manufac-turer are usually very demanding, the manufacturerand supplier will negotiate compromises duringdevelopment on a wide spectrum of specificationsincluding driving method, driving voltage, inputsignal, the dimension of the module, and connectorshapes.

Fourth, it is easy to identify discrete innova-tions. Since the early days of the industry, thekey features through which to differentiate lap-tops were size and resolution, both of which Istudy. Each has advanced in discrete steps, e.g., thejump from 12.1-inch to 13.3-inch displays or fromXGA resolution to the higher SXGA resolution in12.1-inch displays. This provides a series of iden-tifiable innovations for which I can model buyersdecisions.

Lastly, all of the decisions took place in the sameperiod 1992 98, and involve the same buyers,suppliers, and basic technology. This eliminatesmany potentially confounding factors.

Data

My primary data source was the COMTRAK

database compiled by Stanford Resources, one

6 Producing larger displays requires new handling equipmentand processes. For example, it is necessary to apply processessuch as vapor deposition or photolithography uniformly over anever-increasing surface area. Equipment has to be faster andhave greater throughput. Resolution increases require puttingmore circuitry on the same-size display, demanding reducedline-widths, tighter tolerances, and more driver chips with morechallenging packaging.7 For example, although it is possible to specify and measureabsolute brightness, designers can determine consumer accep-tance of a given brightness only in the context of other parame-ters, including color matching, brightness uniformity, and bright-ness leakage.

of the industrys leading consulting firms. Foreach model produced by a notebook manufac-

turer, COMTRAK provides the size and resolutionof the display, as well as the firm that suppliedit. Stanford Resources compiled the informationin COMTRAK from interviews with display sup-pliers and notebook manufacturers from 1992 to1998. Stanford Resources had a strong incentiveto compile these data accurately; COMTRAK sub-scribers paid $1250 per quarterly issue and werewell-informed participants in the industry. TheCOMTRAK data allowed me to compile a com-plete inventory of a notebook manufacturers rela-tionships with display suppliers and to identify the

supplier for a manufacturers first introduction ofa given sizeresolution combination. I augmentedthese data with the trade press and issues of Lap-top Handbook and Buyers Guide. Valuable infor-mation on the industrys development came fromMurtha, Lenway, and Hart (2001).

The data include 116 introductions of innova-tive displays, meaning that a manufacturer incor-porated a new sizeresolution combination, by 13manufacturers. For each innovation a manufacturerincorporated, I generated an observation for eachpotential supplier, including measures of the sup-pliers technical capabilities and the characteristicsof the manufacturersupplier relationship at thattime, a vector of control variables, and an indi-cator of whether the manufacturer selected thatsupplier for its initial introduction of that inno-vation. Table 1 summarizes my conceptual vari-ables and the empirical data used to test them.Table 2 presents descriptive statistics and correla-tions.

I supplemented the quantitative data with inter-views at three notebook computer manufacturersin the United States and Japan. At the two firmsthat also produced displays, I interviewed individ-

uals in both the notebook computer and displaydivisions. Interviews were semi-structured, lastingbetween 45 minutes and 2 hours, with participantsincluding procurement managers, engineers, andothers. To assure maximum clarity during the inter-views in Japan, I began with a pre-translated setof questions in Japanese. Depending on the lan-guage ability of the respondent, responses andfurther questions took place in Japanese, English,or a combination. I also interviewed principalsat three major consulting firms in the industry(two in the United States, one in Japan), rele-vant officials at Japans Ministry of International



10/22

84 G. Hoetker

Table 1. Conceptual variables and empirical data

Conceptual variable Empirical data Variable name

Independent variablesTechnical capability of supplier Number of U.S. display-related patents in

previous 5 yearsPATENTS

Nature of the buyersupplierrelationship

Number of years in which buyer andsupplier have transacted

YEARS

Are the buyer and supplier part of the samecompany? (0/1)

INTERNAL SUPPLIER

Technical uncertainty ofinnovation

Technical advance required as perceived byindustry

PERCEIVED UNCERTAINTY

Technical advance required as determinedby need for new processes (0/1)

NEW PROCESS REQUIRED

Control variables To what degree do the buyer and suppliercompete in the notebook market?

COMPETITOR

Common nationality (0/1) SAME NATIONALITYDependent variable

Is this the supplier selected for the giveninnovation? (0/1)

CHOICE

Table 2. Descriptive statistics and correlations

Mean S. D. Min. Max. 1 2 3 4 5 6 7 8

1 CHOICE 0.03 0.17 0.00 1.00 1.002 PERCEIVED

UNCERTAINTY2.82 1.24 1.00 5.00 0.00 1.00

3 NEW PROCESSREQUIRED

0.25 0.43 0.00 1.00 0.00 0.78 1.00

4 COMPETITOR 9.04 7.42 0.00 34.40 0.07 0.12 0.12 1.005 SAME

NATIONALITY0.40 0.49 0.00 1.00 0.04 0.02 0.00 0.04 1.00

6 PATENTS 181.52 200.8 0.00 926 0.22 0.12 0.06 0.22 0.11 1.007 YEARS 0.25 0.74 0.00 5.00 0.30 0.09 0.05 0.06 0.04 0.35 1.008 INTERNAL

SUPPLIER0.04 0.20 0.00 1.00 0.17 0.01 0.01 0.24 0.24 0.07 0.24 1.00

Significant at 0.10 level

Trade and Industry,8 and other long-time industryparticipants. The combination of qualitative andquantitative methods allows me to capture a morecomplete, holistic, and contextual portrayal of thephenomenon (Jick, 1979: 603, emphasis in origi-nal).

Dependent variable

In the set of observations related to a manufac-turers introduction of a given innovation, this indi-cator variable was set to 0 for all suppliers exceptthe one the manufacturer chose. For that supplier,

8 Now the Ministry of Economy, Trade and Industry.

it was set to 1. Note that while a buyer may even-tually use a number of suppliers for a given typeof display, I focus on the single supplier used in amanufacturers initial introduction of a notebookwith a given display type.

Independent variables and measures

Technical capabilities

Patents are widely used to measure technical capa-bility (Hall, Jaffe, and Trajtenberg, 2000). I mea-sure a suppliers technical capability by the numberof display-related U.S. patents it applied for inthe previous 5 years (PATENTS). This figure isupdated annually. I defined display-related patents



11/22


as those containing the terms liquid crystal dis-play or LCD or classified in International Patent

Classification section G02F 1/, G09G 3/, G09F9/3-, or G09F 13/. I selected these patent clas-sifications according to Spencer (1997) and con-firmed that they were the classifications commonto all patents selected for inclusion in the Liq-uid Crystal Display section of Industry and Tech-nology Patents Profiles: Electronic Displays and

Display Applications, published by Derwent Infor-mation/Thompson Scientific, a leading publisherof patent information. An industry informant con-firmed the appropriateness of this measure for thedisplay industry: Although decision-makers rarely

if ever actually count the number of display-relatedpatents each potential supplier holds at a giventime, the number of patents held by a supplier cor-relates closely with its technical capabilities andwith other firms beliefs about those capabilities.

The manufacturersupplier relationship

I measure two aspects of the relationship betweenthe manufacturer and each of its potential suppli-ers: the extent of past transactions and whetherthey belong to the same company.

I measure past transactions by the number ofprevious years in which the manufacturer hadpurchased at least one notebook computer displayfrom the supplier (YEARS). If a manufacturer andsupplier had relationships prior to their entry intomy data, my measures of their relationships willbe truncated. To control for this possibility, I donot model decisions made in the first 3 years ofa manufacturers presence in the data. Sixty-eightof the original 116 innovations remained in thesample.

The indicator variable, INTERNAL SUPPLIER,is set to 1 if the supplier is part of the same

company as the computer manufacturer. I includethe interaction of INTERNAL SUPPLIER andYEARS, since the value of previous transac-tions may differ for internal and externalsuppliers.

I measure technical uncertainty in two ways.The first, PERCEIVED UNCERTAINTY, ratesindustry perceptions of the advance beyond exist-ing technology that each innovation required. Itranges from 1 to 5. A researcher in a lead-ing consulting company, an 18-year industry vet-eran with experience as both a product engi-neer and product marketing manager, provided

this rating for each innovation I observed in mydata. Since this measure was constructed after the

fact, it may be biased by knowledge of whichinnovations ultimately proved the most difficult.However, it will correlate closely with a pri-ori perceptions of uncertainty unless there wereinnovations that proved surprisingly difficult orsimple, which my informants indicate did notoccur.

The second measure of technical uncer-tainty, the dichotomous variable NEW PROCESSREQUIRED, measures whether suppliers couldproduce the display by refining existing techniquesor if new techniques were necessary. A resolution

increase was highly uncertain (NEW PROCESSREQUIRED = 1) if it required new process tech-nology or breakthroughs in metallurgy, rather thanexecuting existing materials and processes better.A size increase was highly uncertain if it requiredthe assembly of an entirely new fabrication line, asopposed to being possible through improvementsin an existing fabrication line. It was clear beforedevelopment began when the limits of currentmaterials or production lines had been reached. Asenior industry participant who worked in the dis-play division of a firm that made both displays and

notebooks suggested this measure. For each inno-vation I observed, he identified whether it requireda new process or not.

The two measures are consistent, but not iden-tical. All of the innovations rated 5 on the PER-CEIVED UNCERTAINTY scale required new pro-cesses. Every innovation that required a new pro-cess was rated either 4 or 5 on the PERCEIVEDUNCERTAINTY scale. There were 20 innovationsof above-average perceived uncertainty that did notrequire new processes.

Controls

Some notebook manufacturers and potential sup-pliers had inter-organizational relationships inother fields. For example, Acer and Texas Instru-ments had a joint venture to manufacture memorychips starting in 1989. These relationships mightgenerate trust and communications ability betweenthe firms. To control for the effect of these rela-tionships, I use the Quadratic Assignment Proce-dure (QAP) from network analysis, as describedin the Appendix. This procedure controls for



12/22

86 G. Hoetker

the unobserved heterogeneity in buyersupplierdyads, including unobserved relationships.9

Working with a supplier from another countrycreates several challenges, including language dif-ferences (Arrow, 1971) and, possibly, lower trust(Zucker, 1986). These effects may have diminishedgiven the pervasive globalization of the electron-ics industry, but I include an indicator variable,SAME NATIONALITY, set to one if the supplieris of the same nationality as the manufacturer, tocontrol for possible effects.

Firms occasionally procure displays from suppli-ers with which they compete in the notebook com-puter market, despite the possibility of information

leakage and opportunistic behavior in the displaymarket to influence competition in the notebookmarket. Industry informants suggest that the highminimum efficient scale and significant cycles ofsupply demand imbalance in the display indus-try drive this behavior. Downstream competitorsrely on cross-selling of displays to even out theseimbalances. Reputation effects mean that a bla-tantly opportunistic supplier would be unable tosell to downstream competitors in the future, apowerful incentive to avoid opportunistic behav-ior. Since the impact of a potential supplier being

a competitor in the notebook market is unclear,I include a variable (COMPETITOR) measuringthe intensity of competition between the buyerand supplier in the notebook computer market. Ifthe supplier does not produce notebook computers,this variable is set to 0. Otherwise, it is set to thesum of the manufacturer and suppliers worldwidemarket share in the notebook computer industry.10

9 An extensive literature review identified some of these relation-ships. However, many relevant relationships, such as purchasesof other components from a supplier, never appear in the pressand the nature of these relationships varied enormously. There isno theoretical basis for rating these relationships on a continuousbasis and a dichotomous variable is not appropriate. Fortunately,QAP controls for their effect, although it does not allow me tomeasure their impact.10 As is usually the case, I lack information on the price offeredby each supplier. However, two factors make this lack lessimportant in this setting than most. First, I focus on the earlystages of a displays life cycle, during which sales to earlyadopters generate very large margins, dwarfing differences inprices offered by suppliers. Over the life of a given model, otherdisplay suppliers will come online, and competition will driveeveryones price to a market level (interview with industry infor-mant, confirmed by an unrelated industry informant). Second,my interest is in comparing decisions made under low and highuncertainty. There is no reason to suspect that price would havea stronger impact on one than the other.

METHOD: THE DISCRETE CHOICEMODEL

For each innovation, I wish to model the effectof each potential suppliers characteristics on theprobability of the buyer choosing that supplierrather than any other potential supplier. I thereforeapply McFaddens discrete choice model (Domen-cich and McFadden, 1975; McFadden, 1973).

Let the potential suppliers be indexed by s andthe computer manufacturers making the choice bym. The vector of variables associated with suppliers at the time a supplier was being chosen forinnovation i is xsi and the vector of coefficients

to be estimated is . The utility a manufacturerobtains by choosing supplier s is

Usi = xsi + si (1)

where si is an unobserved random term that is anidentically and independently distributed extremevalue, independent of and xsi (Amemiya, 1981).

The probability of manufacturer m choosingsupplier s for innovation i is given by

P (ymi = s|xsi ) =

exp(xsi )s

j=1

exp(xj i )(2)

Since there were 32 potential suppliers for each ofthe 68 new model introductions, this yielded 68 32 = 2176 observations. Hausman and McFad-dens (1984) test indicates that my sample exhibitsindependence of irrelevant alternatives, a keyassumption of the discrete choice model.

To test my hypotheses, I must evaluate the statis-tical significance and relative magnitude of covari-

ates for low-uncertainty vs. high-uncertainty inno-vations. There are several statistical approaches tomaking such a comparison.

One approach is to estimate a single regres-sion for all innovations (Aiken and West, 1991). Adummy variable is set to 1 for one type of innova-tion, e.g., systemic, and interacted with the relevantcovariates. Considering only one variable, x, andletting the dummy variable, H, be set to 1 for high-uncertainty innovations, the equation would be ofthe form

Yi = 0 + 1xi + 2(Hi xi ) + ei (3)



13/22


An estimate of 2 significantly different than 0indicates that the impact ofx varies between low-

and high-uncertainty innovations. The sign of 2indicates whether the impact ofx is diminished orincreased for highly uncertain innovations.

As indicated by the presence of a single errorterm, ei , this approach assumes that the errorvariance of low- and high-uncertainty innovationsis the same (Darnell, 1994: 111; Pindyck andRubinfeld, 1991: 107). If this is not accurate,differences in residual variation will confound anyattempt to compare coefficients across groups anddifferences in the estimated coefficients tell usnothing about the differences in the underlying

impact of x on the two groups (Allison, 1999:190).11

Unobserved heterogeneity likely differs betweenlow- and high-uncertainty innovations. Low-uncertainty innovations require less managementof the design process, lowering the cost ofchoosing a hard-to-manage supplier and givingmore play to unobserved factors. Managersmay also simply pay less careful attentionto selection, increasing the variance for low-uncertainty innovations. Although untestable, thislogic casts severe doubt on the results thatcombining low- and high-uncertainty innovationsinto a single estimation would generate.

Therefore, I estimate high- and low-uncertaintyinnovations in separate regressions. This avoids theunsupportable assumption of common variance atthe cost of complicating comparisons of the effectof covariates across innovation types.

RESULTS

Table 3 presents the results. Model (a) presents thecontrol variables, which I discuss in the later mod-els. Model (b) estimates the full set of coefficientson all innovations. PATENTS is positive and sig-nificant, indicating that buyers are more likely tochoose a supplier with high technical capabilities.

11 In the linear case, it is possible to test this assumption (Quandt,1960). However, in a logit model, this assumption is not testablebecause neither nor 2, the error variance of the underlyinglatent variable, can be identified. Only / can be identified,requiring an arbitrary assumption about the variance to identifythe model (Long, 1997: 47). The likely effect of using a singleequation despite differences in variance between the groups isthat the slope coefficients will not be found to be different(Gujarati, 1988: 527).

This result supports Hypotheses 1. Since Hypothe-ses 2 and 3 are contingent on the uncertainty of

the innovation, Models (c) through (f) divide theinnovations into low uncertainty and high uncer-tainty using the two measures of uncertainty. Mod-els (c) and (d) divide the innovations according toPERCEIVED UNCERTAINTY, the degree of per-ceived uncertainty reported by industry informants.Innovations rated higher than the mean rating(2.82) are classified as high uncertainty. Models(e) and (f) divide the sample according to the valueof NEW PROCESS REQUIRED, which measureswhether refinements of existing processes weresufficient or whether new processes were required.

Using either measure of uncertainty, PATENTSis positive and significant for both low- and high-uncertainty innovations, supporting Hypothesis 1.A suppliers technical capability is clearly animportant consideration when choosing a supplier.

For perceived uncertainty, Models (c) and (d),YEARS is significant only for high uncertainty,consistent with Hypothesis 2. The value of priortransactions is contingent on the level of tech-nical uncertainty. Buyers only value past expe-rience with a supplier when they perceive thatthe desired innovation will require a substantialadvance beyond the technical frontier. This resultindicates that I am not merely observing inertia insupplier selection.

Testing Hypothesis 2 further requires calculatingthe relative effect of technical capabilities and priorexperience under low and high uncertainty. Sincethe coefficients in any logit equation are scaled byan unknown variance, one cannot compare coef-ficients across samples. However, one can com-pare ratios of coefficients (Train, 1998: 237). Forexample, computing the ratio years/patients yieldsa value of 50.71 for low-uncertainty innovationsand 272.23 for high-uncertainty innovations. This

result means that under low uncertainty buyersvalue a year of experience as much as 51 patentsheld by a supplier. In comparison, under highuncertainty, buyers value a year of experience asmuch as 272 patents. For a high-uncertainty inno-vation, a buyer will be willing to give up almostfive times as much technical capability for an addi-tional year of past transactions as it would be will-ing to give up for a low-uncertainty innovation.This difference is significant at the 0.049 level,further supporting Hypothesis 2.

Further evidence comes from managers, whocited several benefits in long-term relationships.



14/22

88 G. Hoetker

Table3.

Resultsofdiscretechoicemodelpredictingsupplierchoice

(a)

(b)

(c)

(d)

(e)

(f)

Controls

All

Lowperceived

uncertainty

Highperceived

uncertainty

Lownew

process

uncertainty

Highprocess

uncertainty

COMPETITOR

0.108

(0.021)

0.043(0.029)

0.028(0.045)

0.064

(0.039)

0.055(0.035)

0.009(0.058)

SAMENATIONALITY

0.391(0.266)

0.908

(0.474)

1.082(0.740)

0.883(0.636

)

0.682(0.516)

1.832(1.235)

PATENTS

0.004

(0.001)

0.005

(0.001)

0.003

(0.001

)

0.004

(0.001)

0.003(0

.001)

YEARS

0.555

(0.158)

0.231(0.241)

0.817

(0.212

)

0.592

(0.182)

0.460(0.341)

INTERNALSUPPLIER

1.838

(0.708)

1.172(1.284)

2.156

(0.914)

1.069(0.893)

3.906(1.412)

INT.SUPP.YEARS

0.031(0.275)

0.903(0.563)

0.539(0.341

)

0.051(0.347)

0.093(0.522)

Decisionsobserved

68

68

31

37

51

17

Observations

2176

2176

992

1184

1632

544

Log-likelihood

221.9

151.01

68.22

79.16

115.42

33.35

Incrementalchi-squareover

acontrols-onlymodel

41.78

70.4

78.2

106.48

38.74

Standarderrorsinparentheses.

Significantat10%;significantat5%;significantat

1%



15/22


Although they mentioned trust, i.e., fewer fearsof opportunism, they saw other benefits as more

important. The predictability that develops overmultiple transactions helps buyers anticipate prob-lems that might arise. As a U.S. manager explained,If a company swears it will meet a schedule, butyou know they have missed every deadline forthe last five years, you can make adjustments. AJapanese manager agreed, focusing on his greaterability to anticipate and correct technical shortcom-ings. This ability is especially important becausedisplay problems often become evident only afterthe computer is in the hands of the customer.

Managers also valued improved inter-firm com-

munications with long-term suppliers. As one U.S.manager explained:

It seems like the development of a notebook com-puter should be a science, but in fact it is the artof getting the right people together. If you arentfamiliar with a company, you dont know who theright people are.

Continuing to examine perceived uncertainty,INTERNAL SUPPLIER is not significant for low-uncertainty innovations, but is highly significantand positive for high-uncertainty innovations.Calculating the relative value of a supplier beinginternal, internal/patients, indicates that a supplierbeing internal is as valuable as 256.7 patents forlow-perceived uncertainty innovations and 729.4patents for innovations with a high-perceiveduncertainty. This difference is significant (p =0.092), supporting Hypothesis 3.

Managers comments are consistent with thisresult. Negotiations with internal suppliers areeased by the fact that, as one Japanese managerexpressed it, we all wear the same uniform. How-ever, each side still views it as a business rela-tionship to be managed. Another Japanese man-

ager felt that he had more open communicationswith his internal supplier. In particular, he felt hereceived information about technological advancesfrom his internal supplier earlier than he wouldfrom an external supplier, allowing him more timeto prepare to take full advantage of the advance. AU.S. manager expressed similar sentiments, draw-ing attention to discussions about technologies intheir embryonic stage that are much more likely tooccur with an internal supplier.

A different pattern emerges in Models (e) and(f), which classify an innovation as high uncer-tainty if it required a new process. The relative

value of a year of experience is statistically indis-tinguishable for low- and high-uncertainty innova-

tions. Computing the ratio years/patients yields 153for high-uncertainty and 148 for low-uncertaintyinnovations. These values are statistically indistin-guishable (p = 0.95), failing to support Hypothe-sis 2. However, Hypothesis 3 is supported. INTER-NAL SUPPLIER is insignificant for low uncer-tainty, but highly significant and large for highuncertainty. The relative value of working with aninternal supplier is 4.5 times greater for an innova-tion requiring new processes than for one that doesnot. Although large, this difference is not statisti-cally significant (p = 0.15), perhaps reflecting the

diffi

culty of achieving statistical signifi

cance forthe difference between ratios of coefficients.Interviews put the apparent failure of Hypothe-

sis 2 in a different light, one consistent with theoverall model. An interviewee at a Japanese note-book computer company said that it is very hardto convince an external supplier to undertake theinvestments required when a new process is neces-sary. The capital required (up to $1.5 billion) andvery high degree of uncertainty make outside sup-pliers reluctant to move ahead, even if there is astrong relationship in place. Being part of the samefirm is a more credible commitment that the buyer

will ultimately purchase the output resulting fromthis capital investment. YEARS is not significantfor innovations requiring new processes becauseprior experience is not sufficient to overcome thechallenges posed.12 This finding is exactly whatthe model predicts: under extreme uncertainty, theadvantages of working with internal suppliers givethem a key advantage over even long-term externalsuppliers.

In combination, the empirical results provideinsights into the conditionally complementarynature of internalization and long-term relation-

ships. At low to moderate levels of uncertainty,internalization and long-term relationships can actas partial substitutes for each other. At high levels

12 The positive, significant result for YEARS for innovationsnot requiring new processes reflects the fact that many inno-vations not requiring new processes still demanded significantadvances beyond the technical horizon. This posed problemsof uncertainty, which past experience could help address. Themean value of PERCEIVED UNCERTAINTY for innovationsnot requiring new processes, Model (e), is 2.27. This is sig-nificantly higher (p = 0.001) than the mean value for the lowperceived uncertainty innovations in Model (c), 1.67. This dif-ference is consistent with that fact that 20 of the innovationsnot requiring new processes were rated as above average inPERCEIVED UNCERTAINTY.



16/22

90 G. Hoetker

of uncertainty, however, internalization is the supe-rior means of overcoming governance and commu-

nication difficulties.Among the control variables, COMPETITOR

is insignificant for innovations of low perceiveduncertainty, but significant and negative for highlyuncertain innovations, indicating that manufactur-ers avoid procuring displays from downstreamcompetitors when large technical advances arerequired. This finding is theoretically consistent.Since it is harder and costlier to protect one-self from opportunistic behavior as uncertaintyincreases, the possibility of a competitor actingopportunistically is more threatening for high-

uncertainty innovations. SAME NATIONALITY,indicating that the buyer and supplier are fromthe same country, is never significant, which mayreflect the impact of globalization in the electronicsindustry.

EXPLANATORY POWER OF THEMODEL

The models true worth comes if it can providebetter understanding of firm behavior. To explorethis, I compare the full model to several alterna-

tives, each reflecting a single lens for viewing theissue of supplier selection:

(i) random choice of suppliers, as a worst casebaseline;

(ii) only the control variables;(iii) the control variables plus PATENTS, reflect-

ing the firm capabilities lens;(iv) the control variables plus YEARS, reflecting

the interfirm relationship lens;(v) the control variables plus INTERNAL SUP-

PLIER, reflecting transaction cost economicstraditional internal vs. external lens.

I first compare the explanatory power of the mod-els by comparing their log-likelihoods (Table 4).Whether looking at all innovations together, low-uncertainty innovations (defined by PERCEIVEDUNCERTAINTY) or high-uncertainty innovations,adding PATENTS, YEARS, or INTERNAL SUP-PLIER individually improves fit over the controlvariables. However, the integrated model fits sig-nificantly better (p < 0.001) than the best of theindividual lens. While each theoretical lens offerssome explanatory power, integrating them pro-vides even more.

Next, I ask how often a model accurately pre-dicted the supplier that the buyer selected. This test

is challenging, as success requires that the buyeractually selected the single supplier that the modelpredicted had the highest probability of being cho-sen. As shown in Table 5, the same pattern holds.Any of the individual lenses improve predictionover the control variables. However, the model thatconsolidates the three lenses is more accurate thanany single lens.

By either measure of explanatory power, thecombined model is a success. It is a significantlymore robust predictor of supplier choice than anyindividual theoretical lens.

ROBUSTNESS CHECKS ANDLIMITATIONS

I argue that internalization provides advantages ofcommunication and governance. But, as William-son (1985: 151153) points out, internal opera-tions and investment decisions are especially sus-ceptible to politicization. In particular, there maybe an internal procurement bias due to managersseeking employment stability and a desire to pro-tect an internal supplier that might be at a rela-

tive disadvantage in the marketplace. If this biasexisted in the notebook computer industry, whichmy interviews suggest it might, we should under-stand each decision to procure internally as beingpartially motivated by this bias.

However, my empirical results remain robusteven in the presence of this procurement bias.Since my primary interest is in the relative valueof internal supply under different levels of uncer-tainty, a problematic statistical bias occurs only ifthe political bias towards internal procurement ismore pronounced for high-uncertainty innovations.

Nothing in theory or my discussions with industryinformants suggests that this is the case.

Since I had no a priori assumptions of howquickly the value of knowledge depreciates in thisindustry, I tested my models using display-relatedpatents in the last year and total display-relatedpatents, rather than the last 5 years. Empiricalresults were robust.

There is an additional issue regarding patents.Lanjouw and Schankerman (1999) and Hall et al.(2000) find evidence that weighting patent countsby subsequent citations provides a more accuratemeasure of firms technical capabilities than an



17/22


Table4.

Explanatorypower,comparisonoflog-likelihoods

(i)No

covariates

(ii)

Control

variables

(iii)

Capab

ilities

lens

(Controls+

PATENTS)

(iv)

Interfirmrelations

lens

(Controls+

YEARS)

(v)

Transactioncosts

lens(Controls+

INTERNALSUPPLIER)

(vi)

Integrated

model(

All

variables)

Allinnovations

Log-likelihood

235.67

221.90

168.66

174.69

201.26

151.01

Comparedtomodel

(i)

(ii)Co

ntrols

(ii)Controls

(ii)Controls

(iii)Capab

ilities

Incrementalchi-square(d.f.change)

27.54(2)

106.48(1)

94.42(1)

41.27(1)

35.30(3)

Significanceofimprovement

Documents

How Much You Know Versus How Well I Know You