The Underemphasized Role of Established Firms as …willm/bio/cv/papers/ICC1996...The...

The Underemphasized Role of EstablishedFirms as the Sources of Major Innovations

D A V I D M E T H S , A N A N D S W A M I N A T H A N AND

W I L L M I T C H E L L

(University of Michigan Business School, Ann Arbor, MI 48109-1234, USA)

We argue that established firms, including industry incumbents and diversifying

entrants, play vital and underemphasized roles as the sources of major innovations in

many industries. The study presents empirical evidence showing that established firms

are often sources of major innovations in the telecommunications and medical sectors. We

conduct an exploratory investigation of him variation in the types of technical and

market change influence whether industry incumbents, diversifying entrants or new

companies are the sources of major innovations.

1. Introduction

This study seeks to raise the profile of established firms as sources of majorinnovations. Traditional perspectives on technical change often assume thatnew firms are more likely than established firms to be the sources of majorinnovations. The traditional approach has two limitations. First, research has

^ only recently begun to provide general explanations for the many cases in~ which industry incumbents introduce major innovations. Second, researchersj rarely address the innovative role of diversifying entry, in which firmsI introduce innovations in industries beyond their current markets. We argue

that established firms, including industry incumbents and diversifying1 entrants, play vital and underemphasized roles as the sources of major£ innovations in many industries. The study presents empirical evidence§> showing that established firms often are sources of major innovations in theo telecommunications and medical sectors. We conduct an exploratoryfe investigation of how variation in the types of technical and market change2£ influence which types of firms are the sources of major innovations.

•| Research on innovation has spawned many typologies of technical change,

2 © Oxford University Press 1996

1181

Established Firms as Sources of Innovations

including references to competence-destroying and competence-enhancingchange, radical and incremental change, modular and architectural change,technological paradigms, technological trajectories, innovation envelopes, anddominant designs (see Adler, 1989). A common thread through theinnovation literature is that the more changes in knowledge a new product orprocess needs, the greater the change the innovation requires. For ease ofpresentation, we will refer to innovations that require substantial newknowledge or substantial reconfiguration of existing knowledge as majorinnovations. This definition follows Schumpeter (1950) in assuming thatmajor innovation often stems from recombining knowledge from previouslyunrelated uses. Major innovations contrast with incremental innovations,which are improvements, refinements, minor modifications, and extensions ofcore or complementary product and process technology that enhance theutility or reduce the cost of an existing product (Utterback and Abernathy,1975; Abernathy and Clark, 1985).

2. The Innovative Role of New and Established Firms

We begin by assessing the role of established firms in fostering newtechnology that leads to major changes within an existing industry or creationof entirely new industries. Scholars dating back at least as far as Schumpeterhave debated the relative roles of established firms and new firms as theprimary sources of major innovations. On the one hand, established firmsmight have a dominant innovative role owing to their financial and technicalresources, coupled with valuable linkages to customers, suppliers and otherconstituents. On the other hand, new companies might be the most likely toidentify path^breaking opportunities outside the bounds of existing practice.Schumpeter himself took both positions during his career: in Theory of EconomicDevelopment (1934) he argued that entrepreneurs in new companies werethe progenitors of most major innovations, whereas in Capitalism, Socialism,and Democracy (1950) he argued that large established firms with R&Dfacilities were the principal source of major innovations. For ease ofpresentation, we will divide the subsequent literature into two main streams:economic studies and behavioral studies. In recent years, both streams ofresearch have emphasized the role of new firms as the source of majorinnovations.

Economic studies emphasize the effect of market concentration, potentialsales cannibalization and firm size on research and development activities(Adler, 1989; Cohen and Levin, 1989). Research is inconclusive about theimpact of market concentration on R&D effort. Some studies have found a

1182


positive relationship between concentration and R&D (e.g. Scherer, 1983a).Others found no significant relationship (Cohen et al., 1987). The earlytheoretical contributions of Feller (1951), Arrow (1962) and others focusedon the disincentives for established firms to undertake major innovations,arguing that firms have economic incentives to avoid cannibalizing existingsales with major innovations.

Many studies have investigated how R&D expenditures, patent output andother research activities vary with firm size (e.g. Eeinman and Fuentevilla,1976; Scherer, 1984; Reinganum, 1985). The most.general conclusion of theempirical studies of firm size and innovativeness is that R&D inputs tend toincrease more than proportionately with firm size, but R&D outputs asmeasured by patents tend to increase less than proportionately with firm size(Acs and Audretsch, 1990, pp. 38-43). The evidence on the patentoutput-R&D input relationship suggests either that R&D effort suffers fromdecreasing returns or that large firms tend to patent fewer inventions (Scherer,1983b). Nonetheless, Acs and Audretsch (1990) found that in manyindustries large firms play a significant role in innovation. In 122 of 448industries, Acs and Audretsch (1990, p. 53) found that the large-firminnovation rate exceeded the small-firm innovation rate. Further, in another170 industries the innovation rates of large and small firms wereindistinguishable. Acs and Audretsch attribute the relative innovativeadvantage of large firms to industry-level factors such as capital intensity,advertising intensity and market concentration. Additional factors, though,are likely to include firm level issues such as the knowledge base that firmsdraw upon in the innovation process, as we discuss in the next section of thepaper.

Paralleling the economic research, studies using behavioral frameworksexamine the options available to firms confronted by technical change. In thisresearch, the focus shifts somewhat from a distinction between large firms andsmall firms to one between industry incumbents and new entrants. Researchconducted by Cooper and Schendel (1976), Foster (1986), Tushman andAnderson (1986) and others suggests that major technical changes that affectan industry often originate in firms outside the established industry.

In searching for explanations for the inertia of established firms, researchersbegan to explore the innovation process and its impact on industries andfirms. This stream of research examines the implications for firms of varioustypes of technical change. Tushman et al. (1986) distinguished betweenincremental and discontinuous technical change that, respectively, requiredfirms to take a frame bending or frame breaking approach to theircapabilities. Abernathy and Clark (1985) and Tushman and Anderson (1986)

1183 '•


distinguished between competence-enhancing and competence-destroyingtechnical change, arguing that innovations that decreased the value of thecapabilities of industry incumbents would tend to emerge from outside theestablished industry. Anderson and Tushman (1990) examined the cycle oftechnological development from discontinuity through ferment to theemergence of a dominant design. They found that firms needed to developheterogeneous organizational and interorganizational competencies to dealwith the diverse technical, organizational and interorganizational challengesposed by these cycles. Industry incumbents often lack critical capabilitiesneeded to address the challenges (Burgelman, 1983; Mezias and Glynn,1993). Henderson and Clark (1990) argued that entrants had greatereconomic incentives than incumbents to invest in new technology. They alsosuggested that organizational inertia constrains incumbents' ability todevelop capabilities needed for new technologies.

Organizational inertia often stems from the routines needed to carry outbusiness operation. Routines are identifiable patterns of activity embodied inhuman or capital assets (Winter, 1990). Firms develop internal routines thatallow them to perform existing tasks reliably and to accumulate knowledge(Hannan and Freeman, 1977; Argote et al., 1990; Cohen and Levinthal,1990). Firms also develop routines based on interorganizational relationshipswith customers, suppliers and other external constituents in the market(Hannan and Freeman, 1989; Baker, 1990; Fichman and Levinthal, 1991;Singh and Mitchell, 1996). The organizational and interorganizationalroutines tend to localize the search patterns for new technology, even whencompetition or foresight causes firms to attempt to innovate (Nelson andWinter, 1982). The political process of reallocating resources withinestablished firms also detracts from responsiveness (Pettigrew, 1973; Pfefferand Salancik, 1974; Mintzberg, 1983). Consequently, managers of establishedfirms do not observe many new opportunities or are slow to react toopportunities (Zajac and Bazerman, 1991; Leonard-Barton, 1992). Overall,this literature suggests that established firms will find it difficult to create ofadapt to innovations that require substantial changes in their organization.

In contrast to the inertia of established firms, theorists often credit newcompanies with substantial flexibility and responsiveness to opportunities.The reasoning flows from the obverse of the reasoning attributed toincumbents (see Bourgeois and Eisenhardt, 1987, 1988; Bahrami and Evans,1987). New firms have no existing sales to cannibalize, which createsincentives to initiate change. In addition, new companies can establish neworganizational routines that will allow them to initiate and exploit change.

The popular press reinforces the image of new firms as the dominant1184


sources of major innovations with frequent stories about the latest 'High Tech,High Flyer'. Reports of successful initial public offerings led Business Week(December 18, 1995) to term the current US economy as fueled by 'IPOcapitalism'. Conversely, innovation by established firms seems to garnerheadlines mainly when they fail. The idea that large firms can only innovatewhen they break themselves up into smaller units has gained wide currency,as shown by media reports surrounding the planned break-up of AT&T. Theconclusion of this line of thinking is that new firms play a dominant role asthe source of major innovations.

Nonetheless, several contradictions arise when one considers the argumentthat established firms play an insignificant role as the source of majorinnovations. Perhaps the most basic conceptual problem is that the conclusionprimarily applies in terms of average firm-level tendencies; that is, as anargument that many established firms will not initiate major innovations. Weagree with this argument. Nonetheless, the argument obscures the empiricalfact that some established firms play a critically important role in theinnovation process by developing, sponsoring and legitimizing the adoptionof new technologies (Wade, 1995).

In order to investigate the innovative role of established firms, it is usefulto consider examples of major innovation by two types of companies: industryincumbents and diversifying entrants. Industry incumbents are firms thatalready participate in an industry when they develop a new good or service.Diversifying entrants are established companies that enter an industry that isnew to them.

Many theoretical and statistical analyses portray industry incumbents asactive participants in the process of incremental innovation but minor playersas sources of major innovations. In practice, though, industry incumbents arethe sources of many major innovations. Anderson and Tushman (1991) foundthat 'veterans' often are the sources of major process innovations anddominant designs. Meth6 (1992) showed that incumbents were the sources ofmany major innovations in the semiconductor industry. Banbury and Mitchell(1995) found that industry incumbents were common sources of majorinnovations in the pacemaker industry. Many other examples arise in otherindustries, such as IBM and reduced instruction set computing (RISC) in thesemiconductor industry, RCA and Westinghouse and the liquid crystal display(LCD) in the electronics industry, and Fujitsu-Fanuc in the development ofindustrial robotics (Florida and Browdy, 1991; Klepper, 1988).

The argument that established firms rarely develop major innovationsweakens further when we consider diversifying entrants. Diversifying entrantsreceive some mention but rarely play central roles in theoretical and empirical

1185


analyses of technical change. However, diversifying entry by establishedcompanies is the most common form of entry into most industries (Dunne etai, 1988), and diversifying entrants are the source of many majorinnovations. Many of the firms in the Cooper and Schendel (1976)multi-industry study and the Henderson and Clark (1990) photolithographystudy were diversifying entrants. Methe" (1992) found that diversifyingentrants were key sources of major innovations in the semiconductor industry.Klepper (1988) credited a diversifying entrant as the initial source ofcommercial industrial robots. Other examples include Ampex in thedevelopment of broadcast VCRs (Rosenbloom and Freeze, 1985); Xerox in thedevelopment of icon-based and networked PCs (Smith and Alexander, 1988);banks in the development of ATM networks, Dornier and medicallithotripters; and RCA in the development of television networks (Brand andEmmons, 1992). Again, established firms play critical roles as sources ofmajor innovations.

Although one might view examples of major innovations developed byindustry incumbents and diversifying entrants as idiosyncratic exceptions, webelieve that major innovation by established firms is more common and moresystematic than we commonly recognize. Underemphasis of the innovativeroles of established firms occurs for several reasons. In part, the establishedfirms' role is obscure because their innovative activities tend to be small partsof the firms' overall activities. This means that the activities are less obviousthan in the case of new firms, for which the innovative activities may be thesole activities. Moreover, financial analysts, who provide much of the publicinformation about firms' innovative activities, have less incentive to followindividual innovation projects of established firms because a single projectoffers less potential for corporate financial growth than in the case of newcompanies. In addition, established firms' failures tend to involve more peopleand money than the failures of new firms, so that academic and popularattention often focuses on cases in which large firms fail rather than on casesin which they succeed. A further reason why the role of established firms insupporting and initiating innovation is underemphasized is that firms that arecritically important to the generation of successful new technology often donot benefit commercially themselves. Although innovative establishedcompanies sometimes become strong players in the market for the goodsincorporating the new technology, later entrants very often become the mostsuccessful commercially. The later entrants may be large firms that comecloser to fitting the popular perception of established firms as imitators(Davies, 1979). In other cases, the firms that benefit from the initial technicalactivity are small firms founded by employees of the initiating firm or

1186 =


entering in response to growing market opportunities created by theinitiating firm. The early innovative role of the established firms either oftennever becomes widely apparent or is quickly forgotten.

So far, our argument stresses that established firms are important sourcesof major innovations We do not wish to understate the critically importantrole of new firms. Rather, our goal is to raise the profile of anunderemphasized part of the innovation process. To further investigatewhether established firms are important sources of major innovations, wedetermined the firms that introduced several major innovations in thetelecommunications and medical sectors from the 1950s to the 1980s.

3. Sources of Major Innovations in the Telecommunications andMedical Sectors

We chose the telecommunications and medical sectors for several reasons: thesectors are sufficiently different to provide breadth to the study; there havebeen frequent important innovations in the industries; moreover, the sectorsprovide detailed data for archival research. We identified major innovationsfrom an examination of the technical and business press and verified thesethrough discussions with industry participants. We used an extensive reviewof business and government publications to identify the first five firms tointroduce each of the major innovations. Our primary assumption is that theearliest firms to introduce new goods or services are the primary sources of theinnovations or, at least, were integrally involved in developing the innovation.The analysis distinguishes between new product markets and majorinnovation within existing product markets. A new product market is aproduct or service that provides a new function. An example from thetelecommunications sample is satellite communications. Quite simply, untilsomeone launched a satellite into space and beamed communications signalsto and from it, no such function existed. We define a major innovation withinan existing product market as the introduction of a product into a market forwhich some currently existing product provides a similar function. The newproduct may provide the same funaion as well as new functions. An examplefrom the medical devices sample is the introduction of multidimensionalB-scan and then M-scan ultrasound devices, which followed the introductionof unidirectional A-scan ultrasound devices. Both products provided newfunctions for the already existing ultrasound devices in the market. Com-paring new product markets and major innovation within product marketshelps investigate the sources of major innovations in different contexts.

'• 1187


Entry into New Product Markets

Table 1 summarizes entry by new companies and established firms into 12newtelecommunication and medical device product markets. For this table wedefine an established firm as an existing company that diversified into theproduct market; that is, a diversifying entrant. We define a new company as acompany founded in order to enter the product market. We examined six newproduct markets in the telecommunications area and six in the medicaldevices area. We recorded entry based on year of entry, counting as ties firmsthat entered during the same year. We report statistics for both the firstentrant and the first five entrants in each productmarket. The dual listingreduces any concern about entry timing ambiguity and addresses thepossibility that the first entrants might be responding quickly to productsbeing developed by other firms. We believe that entry by the first five firmsprovides a long enough window to address these issues.

Table 1 depicts the frequency with which established firms, in the form ofdiversifying entrants, were the sources of major innovations in the tele-communications and medical sectors. Overall, established firms accounted for63% of the first entrants and 74% of the first five entrants in the 12 productmarkets, with similar proportions in both sectors. The results illustrate theimportance of established firms as sources of innovations that initiatedproduct markets within the telecommunications and medical device sectors.

Major Innovation within Established Product Markets

Table 2 summarizes the sources of 17 major innovations within establishedmedical device product markets. We compare entry by industry incumbents,diversifying entrants and new companies. Again, we examined which type offirm was the first entrant into the product market and also identified the firstfive entrants.

Table 2 indicates the frequency with which established firms are the sourcesof major innovations in the established product markets. Taken as a whole,established firms account for more than two-thirds (68%) of the first entrantsand almost three-quarters (73%) of the first five entrants. When we dividethe established firm category into incumbents and diversifying entrants, theinnovative role of the industry incumbents becomes particularly clear.Industry incumbents account for almost half of the early entrants, including47% of the first entrants and 44% of the first five entrants. Although newcompanies are common early entrants, industry incumbents and diversifyingentrants clearly play critical roles as major innovators.

: 1188


TABLE 1. Entry into New Telecommunications and Medical Device Produa Markets byDiversifying Entrants and New Companies

First entry First entrants) First five entrantsand ties: diversifyingentrants

Telecommunications sector product markets

Satellite manufacturing 1960

Satellite communications I960services

Value added networks 1975

Public electronic mail services 1979

Direct broadcast satellite 1984

Community healthcareinformation networks

1990

Telecommunications mean:diversifying entrants

Medical device product markets

Nuclear medical imaging 1932scanners

Ultrasonic scanners 1956

Vascular grafts 1961

CT scanners 1972

Magnetic resonance imaging 1980scanners

Digital radiographic scanners 1980

Medical device mean:diversifying entrants

diversifying entrant 100%


diversifying entrant & 80%new company (tie)

new company 90%


new company

58%

new company

diversifying entrant



new company


67%

60%

80%

40%

60%

50%

78%

75%

100%

67%

Overall mean: diversifying entrants 63% 74%

The results in this section reinforce our argument that establishedcompanies are important sources of major innovations. In the next section weoutline factors that we expect to influence whether established firms or newcompanies tend to be the sources of major innovations.

1189 —


TABLE 2. Major Innovations within Medical Device Product Markets Introduced byIncumbents, Diversifying Entrants and New Companies

Innovation

Vascular grafts: velour

Vascular grafts: PTFE

Vascular grafts:umbilical

Nuclear medical:cameras

Nuclear medical:computers

Nuclear medical: SPECT

Nuclear medical: PET

Ultrasound: B-scan

Ultrasound: M-scan

Ultrasound: doppler

Ultrasound: mechanicalarray

Ultrasound: linear array

Ultrasound: duplex

Ultrasound: phasedarray

Ultrasound: annulararray

Ultrasound: digital

Ultrasound: coloured

Mean

Incumbent

Diversifying entrants

New companies

Firstentry

1966

1974

1979

1962

1968

1976

1976

1957

1961

1963

1965

1968

1969

1975

1977

1980

1984

First entrants)

new company

diversifyingentrant & newcompany (tie)

incumbent

incumbent

new company

incumbent


new company

new company


incumbent


incumbent

incumbent

new company

incumbent

incumbent

47%

21%

32%

First five entrants and ties:

Incumbents Diversifying

50%

0%

25%

80%'

25%

100%

0%

. 20%

40%

20%

75%

20%

60%

86%

29%

80%

43%

44%

entrants

0%

50%

50%

0%

38%

0%

60%

20%

40%

60%

13%

60%

40%

14%

43%

0%

14%

29%

Newcompanies

50%

50%

25%

20%

38%

0%

40%

60%

20%

20%

13%

20%

0%

0%

29%

20%

43%

26%

1190


4. Factors Affecting Major Innovation by Established FirmsWe believe that the issues affecting innovation by established firms can bestbe understood by examining factors that create the conditions that favor onetype of firm's entry over another. The ongoing debate on whether start-ups orincumbents are more innovative would be better understood if recast in termsof what conditions favor new companies, diversifying entrants or incumbents.

We consider two general sets of factors: the type of technical changerequired to undertake an innovation arid the impact that the innovation willhave on current markets. The type of technical change is important becauseit will affect the type of knowledge a firm requires in order to innovate. Thetype of market change is important because a new product must attractcustomers. We also discuss competitive influences following the discussion oftechnical and market change.

We develop a simple typology of technical and market change. Withintechnical change, we distinguish between reconfiguration of existingknowledge bases and the creation of new knowledge bases. By knowledgebase we mean the research and development, production and marketing skillsrequired to commercialize an innovation. Innovations that require recon-figuration of a firm's existing knowledge base would encompass both themodular and architectural types of innovation discussed by Henderson andClark (1990, pp. 10-12). Within market change, we distinguish betweeninnovations that substitute for existing goods and services, innovations thatcomplement existing goods and services, and innovations that create hewmarkets. We argue that variation along the technical and market dimensionswill affect which types of firms are the sources of major innovations.

We start with a framework that is consistent with the dominant themes ofexisting arguments concerning the impact of technical arid market change onthe sources of innovation. The empirical study addresses innovations whichresult in new products or services rather than innovations in processes forexisting products or services, but the logic of the discussion applies to bothproduct and process innovation. In the following section of the paper, we willreport empirical results classified according to the framework. The resultsshow partial consistency with traditional expectations, along with some keydifferences. We then use the combination of the theoretical perspectives andthe empirical results to develop implications for continuing research. Table 3summarizes the initial framework.

We first consider the technical change dimension of Table 3. The commonargument is that new companies will be the source of most innovations thatrequire substantial new knowledge. The logic stems from the behavioralresearch related to the concept of routines (see Nelson and Winter, 1982;


TABLE 3. Influences on Sources of Innovation with Variation in Technical Change and MarketChange

Market change Technical change

New knowledge base Knowledge basereconfiguration

Substitute use New companies Diversifying entrantsIncumbents (?) Incumbents (?)

Complementary use New companies IncumbentsDiversifying entrants

New use New companies Diversifying entrants

Tushman et al., 1986; Hannan and Freeman, 1989; Singh and Mitchell,1996). Routines are costly and difficult to create. Once firms have createdroutines, changing routines is both costly and subject to organizationalresistance. New knowledge bases require the creation of new routines, so thatestablished firms may have to bear the costs of disrupting established routinesand will often have to overcome organizational resistance to change. Newcompanies bear only the cost of establishing the new routines. To the extentthat financial capital isavailable, new firms will be more likely to develop theroutines needed to create and use the new knowledge.

In contrast to new knowledge bases, established firms will be commonsources of innovation when innovations require reconfiguration of existingknowledge bases. Firms can often reconfigure existing knowledge bases byusing existing organizational routines, so that established firms have bothbehavioral and economic incentives to pursue innovative opportunities.Moreover, established firms will be the most common repositories of much ofthe reconfigurable knowledge, which often provides the impetus forinnovative activity by the firms (Henderson and Clark, 1990;Leonard-Barton, 1992).

The vertical dimension of Table 3 reflects the common argumentconcerning the types of firms that will be sources of innovations requiring newor reconfigured knowledge. As we noted earlier, though, a limit of thisargument is that it applies to average firm-level tendencies. That is, thestrongest argument is that most established firms will be unlikely to introduceinnovations requiring substantial new knowledge. The argument is weaker at

1192


the level of the innovation in the sense that outliers among established firmsmight well be common. Although existing routines might bind mostestablished firms, a few firms in any industry might well be willing and ableto invest time and money in exploring new knowledge. Financial incentivesor simply research curiosity to carry out exploratory experiments might drivethe investment. Also, as we noted earlier, such early experiments often fail inthe market. If such failures quickly disappear from sight and memory, thenestablished firms might be more common sources of innovations involvingsubstantial new knowledge than we commonly recognize.

Cases in which incumbents are sources of innovations requiring substantialnew knowledge might well be most common in innovations involvingsubstitute uses. As we discuss below, the traditional argument suggests thatmany incumbents will avoid substitute innovation. However, thoseincumbents that do undertake innovation requiring new knowledge will oftenfocus on substitute goods and services, which involve local search in the senseof their existing market even if the innovations require distant search for newtechnical knowledge. Therefore, we add incumbents as possible sources ofsubstitute innovations that require substantial new knowledge.

We now turn to the market change dimension of Table 3, addressingsubstitute innovations, complementary innovations and innovations thatcreate new product markets. The common argument regarding innovationsthat substitute for existing products is that incumbents have strong incentivesto avoid cannibalizing sales of their existing product lines. As we noted earlier,several studies argue that incumbents will tend to avoid innovations thatwould reduce the value of existing assets, including both organizational andphysical assets (see Arrow, 1962; Abernathy and Clark, 1985; Henderson andClark, 1990). Instead, this research suggests that new companies anddiversifying entrants will be common sources of major innovations thatsubstitute for existing goods and services. Substitutes may provideopportunities for new companies and diversifying entrants to enter anestablished market. This result has not only found support in empiricalstudies (Tushinan and Anderson, 1986; Henderson and Clark, 1990) buthas also been derived in theoretical models of product cannibalization (seeReinganum, 1985; Conner, 1988). Coupled with the previous discussion oftechnical change, therefore, this research stream suggests that new companieswill be common sources of substitute innovations that require substantial newknowledge and diversifying entrants will be common sources of substituteinnovations that require knowledge reconfiguration. Consistent with ourdiscussion above, though, we add incumbents as possible sources of substituteinnovations. This may seem to run counter to empirical observation but,

1193


again, incumbents might often be forgotten sources if after-the-face empiricalobservations tend to omit failed innovations.

Innovations involving complementary uses have received less attentionthan substitutes in traditional analyses, but several arguments are relevant.We define complementary goods as those goods that together make up asystem. As an example, Intel makes microprocessors, motherboards on whichmicroprocessors are mounted and personal computers in which motherboardsare installed. Each product cannot function alone and often innovation in onestimulates innovation in the others. Intel is an incumbent in all threeindustries, making complementary products. As a further example, a personalcomputer and a laser printer make up a system. If a company making personalcomputers also decides to make laser printers, then it is producing acomplementary product. An incumbent in the case of complementary goodsis a company that is making any of the various parts of the system.

All three types of firms have incentives and opportunities to developcomplementary products. Incumbents may gain economies of scope fromexisting assets by developing complementary goods and services. Searchroutines also tend to localize around current markets and technologies, whichwould create opportunities for innovation by industry incumbents, especiallyin cases involving knowledge reconfiguration. Consistent with the earlierdiscussion of technical change, new companies might tend to be the sourcesof complementary innovations that require substantial new knowledge. Thus,incumbents and new companies might be common sources of complementaryinnovations. However, complementary innovations also offer competitiveincentives for diversifying entrants, especially in the case of knowledgereconfiguration. A diversifying entrant would be a company entering from anunrelated industry, such as an aerospace company entering the personalcomputer or printer markets with an innovation. Such firms would have feweropportunities to gain scope economies, relative to incumbents. At the sametime, however, complementary innovation may provide a means of enteringan industry without competing directly with incumbents. Thus, diversifyingentrants also might be common sources of complementary innovations.

Finally, we turn to innovations involving substantial new uses, which createnew product markets rather than substitute for or complement existingproducts. Industry incumbents are not relevant to this category of innovation,because the innovations create new industries. We expect new companies tobe common initiators of new product markets requiring substantial newknowledge, while diversifying entrants will be common sources in casesinvolving knowledge reconfiguration.' In summary, we expect variation in technical change and market change to

1194

Established Firm as Sources of Innovations

influence the types of firms that are the sources of major innovations. Westress that our approach permits multiple influences to act simultaneously, sothat different types of firms will sometimes be sources of similar innovations.We believe that this breadth is a strength of the approach because it isconsistent with the empirical outcome in many industries.

Other factors will also influence which types of firms are the sources ofmajor innovations. Acs and Audretsch (1987), Dorfman (1988) and othershave argued that industry conditions may influence the relative contributiontowards innovation of types of firms. For instance, industry incumbents facestrong incentives to innovate when they face substantial current or potentialcompetition (Mitchell, 1989), because the threat of losing sales to other firmsoutweighs the disincentives of cannibalization. Despite the importance ofother factors, the technical and market change factors warrant investigation.Moreover, the technical and market change factors will tend to influence howstrongly other factors, such as competitiveness, affect established firms. Forinstance, industry incumbents will often be better able to respond tocompetitive pressure involving substitute products when innovation requiresthe reconfiguration of existing knowledge rather than creation of newknowledge.

We use the data from the telecommunications and medical sectors toconduct an exploratory investigation of how technical and market changeinfluences the sources of major innovations. The purpose of this empiricalanalysis is to help develop an understanding of the innovative role of differenttypes of firms, rather than to provide a conclusive test of predictions.

5. Empirical Analysis of Market and Technical ChangeTable 4 reports the first firms to introduce the major medical andtelecommunications innovations that we listed earlier, based on variation inmarket change and technical change. The market change classificationdenotes whether the innovation substituted for existing goods, complementedexisting goods or created a new use. The technical change classificationdenotes whether the innovation required substantial new knowledge. Weassigned values for the classifications based on discussions with industryparticipants. The numbers reported in bold type are the types of firmspredicted in Table 3 to be common sources of major innovations relevant toeach cell of Table 4.

Several aspects of Table 4 are noteworthy. First, on the technical changedimension, new companies are common sources of innovations requiring newknowledge bases, while established firms are more common in cases involving

1195


TABLE 4. Variation in Technical and Market Change and First Entrants for Major Medical andTelecommunications Innovation (29 innovations; 2 ties among first entrants)

Market change

Substitute use

Complementaryuse

New use

Total

New companies

Incumbents


New companies

Incumbents


New companies

Incumbents


New companies

Incumbents


Technical change

New knowledgebase

1

2

1

1

0

0

2

-

1

4

2

2

Reconfiguration

1

5

5

5

1

3

1

-

2

7

6

10

Total

2

7

6

6

1

3

3

-

3

11

8

12

The figures reported in bold italic type are the types of firms predicted in Table 3 to be common sourcesof major innovations relevant to each cell of the table.

reconfiguration. Nonetheless, new companies also appear in severalreconfiguration cases and established companies appear in several casesinvolving new knowledge. Second, on the market change dimension,diversifying entrants were common sources of substitute products involvingreconfigured knowledge. Incumbents also were common sources of majorinnovation involving reconfigured knowledge for substitute products—apoint that we will return to later in the paper. Third, diversifying entrantswere common sources of complementary innovations involving knowledgereconfiguration. Unexpectedly, new companies also were common sources ofcomplementary innovations involving knowledge reconfiguration. Fourth, inthe new use category, new companies were more common than diversifyingentrants in the new knowledge base cell, while diversifying entrants weremore common in the reconfiguration cell. One must interpret thecomparisons cautiously becasue of the small sample and broad measures, butthe implications are intriguing.


We also used binomial logistic regression (Yamaguchi, 1991) toestimate the influence of market and technical change on the likelihoodthat new companies, diversifying entrants and incumbents would be thesources of major innovations. The logistic regression modeb took theform ln(P;/(l - P,)) = (X + PX. In this equation, P,is the probability that afirm of type / would be the first entrant into a new industry or subfield. Thelog odds of the probability is held to be linearly affected by a matrix ofcovariates X with coefficient vector P and intercept a. The effect of a one-unit change of covariate j on the probability of a firm of type / being thesource of a major innovation is PyPXl — Pi). We used the SAS statisticalpackage to obtain the maximum likelihood estimates. The logistic regressionprocedure allows us to verify and summarize the interpretation of thecontingency tables.

We analyzed three binary logit models. Each logit model compared onetype of entrant with the other two types of entrants (new companies,diversifying entrants, incumbents). Thus, the 0—1 dependent variable foreach equation denoted whether the first entrant to a new product market ornew product class within an existing product market was the specified typeof firm. The covariates included dummy variables for market and technicalchange. The market change variables denoted whether the innovationsubstituted for existing goods, complemented existing goods or created anew use. The technical change variable denoted whether the innovationrequired creation of substantial new knowledge. We did not investigate theinteraction of market and technical change in the regression analysis becasueof the sample size. The analysis of incumbents includes only the majorinnovations within existing product markets. Table 5 reports the results of theregression analysis.

The results reported in Table 5 are consistent with our earlier discussion.New companies are particularly likely to be first entrants when an innovationrequires new knowledge bases. This result supports the argument that newcompanies are more likely than established firms to be able to create the newroutines needed to adapt to fundamentally new knowledge. By contrast,according to this logic, established companies are particularly likely toundertake major innovation that requires reconfiguration of existingknowledge bases. The market change results, consistent with thecross-tabulation in Table 4, show that incumbents were especially likely to bethe sources of innovations that substituted for existing uses. Conversely, newcompanies were particularly unlikely to be the sources of substitutes.

We conducted several sensitivity analyses of the regression estimates. Wefound that the reported results were robust to the addition of variables

1 1 9 7


TABLE 5. Binomial Logistic Regression Estimates of Market and Technical Change Influenceson the Type of Firm that Introduced Major Telecommunications arid Medical Sector Innovations

Technical change

New knowledge base

Market change

Substitute use*

Complementary use*

Intercept

Model log likelihood (At)

Cases (first entrants)

Influence on likelihood

New company

2.09**

-2.48**

0.84

-0.29

11.7(3)**

29(11)

that first entrant was a:

Diversifying entrant

-1.03

-1.00

-1.44

0.84

2.1 (3)

29 (12)

Incumbent

-0.22

2.71**

-1.79**

5.6 (2)*

17(8)

All variables are 0-1 dummies; positive coefficient indicates that type was more likely than others.

"Substitute use and complementary use compared with new use.

•P < 0.10, **P < 0.05 (one-tailed tests).

denoting entry year, telecommunications versus the medical industrial sector,service versus manufactured-goods innovations, and new product marketsversus major innovations within product markets. We also found similarresults in linear regression estimates that specified the dependent variable asthe proportion of each type of firm among the first five entrants.

The empirical results are consistent with several aspects of the initialframework outlined in Table 3. Diversifying entrants were common sourcesof substitute products that required knowledge reconfiguration. Diversifyingentrants were common sources of complementary innovations that requiredknowledge reconfiguration. New companies were common sources of new-use products that required new knowledge bases. Diversifying entrantswere common sources of new-use products that required knowledgereconfiguration.

At the same time, though, there are two major contradictions between theinitial framework in Table 3 and the empirical results in Tables 4 and 5. First,incumbents were the sources of many innovations that substituted for existinggoods. Second, new firms were the sources of many complementary productsthat involved knowledge reconfiguration. In the next section, we explore theanomalies in more detail.

1198


6. Implications for Future ResearchThe two contradictions between the initial framework and the empiricalresults suggest intriguing implications for continuing research. The marketchange results in Tables 4 and 5 raise the question of why incumbents oftenare the sources of substitute products. One possible explanation for the resultsis that high technical and market uncertainty for innovations mean that it isoften unclear whether a new good will substitute for an existing good or createcomplementary or new uses. In such cases, incumbents might introduce newgoods that they expect to create new or complementary uses but whichactually turn out to be substitutes. A more likely explanation for the results,though, is that incumbents are more proactive with substitutes than they aresometimes credited as being. As we noted earlier, the effect of competitionmight be at play as incumbents attempt to pre-empt rivals. Moreover, the factthat incumbents are often embedded in a web of relationships with buyers,suppliers and competitors will often assist the incumbents. Much of the workthat has examined the spill-over effects of innovations (Jewkes et at., 1958;von Hippel, 1988) has shown that many innovative ideas originated withcustomers and suppliers. Similarly, Mettie" (1985, 1991) showed that supplierand customer relationships contributed to the rise of Japanese semiconductorfirms in the 1980s. Not only are these partners sources of ideas forinnovations, but innovation by one of the partners may also necessitateinnovation in an incumbent firm. The foresight that some incumbents gainthrough prior experience with the technology, market and other firms mayallow them to make early judgements about innovative opportunities.

The lack of incumbent firms in the case of complementary innovationsmight suggest that there were few economies of scope available. In order tounderstand why this situation might result, we can re-examine the Intelexample cited earlier in the paper. In the example, we suggested thatinnovating in microprocessors, motherboards and PCs may provide economiesof scope for Intel. However, each of the goods falls within a different productmarket, with its own set of suppliers, customers, competitors and technology.In learning and managing these diverse industries, incumbents may tend toincur costs that lessen the advantage gained from scope economies.

In addition, complementary innovations tend to create fewer competitivethreats for incumbents than do substitute innovations. As in the case ofsubstitutes, incumbents may be aware of potential entrants for com-plementary goods but may expect to gain from overall increased demand ifanother firm successfully introduces a complementary product, withouttaking the risk of unsuccessful innovation or incurring the costs involved bymanaging in diverse markets. The importance of the incumbent carrying out

1199


the innovation lessens further because, unlike the case of substitutes, theincumbent does not lose a market to the efforts of the entering firm.Therefore, incumbents may welcome entry by other firms in order to takeadvantage of innovation in complementary goods. (Kogut et al., 1995). As wenoted earlier, the incentives of the entrant may be higher in the case ofcomplementary goods than in the case of substitute goods. For the industryentrant, complementary goods avoid direct competition with an entrenchedcompetitor. Also, entrants may have opportunities to ally with incumbents,which would allow an entrant to gain access to complementary assets and tolearn about the industry it has just entered (Freeman, 1990).

7. Conclusion

Our empirical analyses show that established firms are frequent sources ofmajor innovations in the telecommunications and medical sectors. This roletends to vary depending upon the change in technical and market factors. Theinnovator is more likely to be a new company than an established firm wheninnovation requires a new knowledge base. Although the results are onlysuggestive, there is some indication that new companies and diversifyingentrants often try to avoid competing head-on with incumbents and, instead,introduce goods that complement existing goods or which create new uses. Incontrast, counter to traditional predictions, incumbents often introduceinnovations involving substitute uses.

Why is the role of established firms as common sources of major innovationoften underemphasized? At least three factors seem relevant in addressing thisquestion. First, although some incumbents may be early innovators, manyincumbents are not. As a result, the average tendency toward lethargy is themost obvious. Second, discussion of innovation often focuses on casesinvolving the creation of substantial hew knowledge bases, which areparticularly likely to involve new companies. By contrast, major innovationsthat stem from reconfiguring existing knowledge bases, in which establishedfirms appear to play critical roles, receive much less attention. This lack ofattention is unfortunate because reconfiguration often creates fundamentalshifts in technology and markets. Third, many established firms fail in themarket even though they are among the earliest to introduce an innovation.The forces of change unleashed by a new technology often create needs fornew skills and push markets in unforeseen directions. These changes oftenoverwhelm the originating firms, which leave the market. Time and our ownperceptions of success then tend to create survival biases that cause us to lose

1200


sight of the first firms to bring about an innovation. These issues provide abase for continued analysis.

AcknowledgementThis paper was presented at the Conference of the Consortium forResearch concerning Telecommunications Policy and Strategy, May 1996,Northwestern University.

> : ReferencesAbernathy, W. J. and K. B. Clark (1983), 'Innovation: Mapping the Winds of Creative Destruction,' Research

Policy, 14, 3-22.

Acs, Z. J. and D. B. Audretsch (1987), 'Innovation, Market Structure and Firm Size,' The Review of Economics

andStatistia, 71, 567-574.

Acs, Z. J. and D. B. Audretsch (1990), Innovation and Small Firm. MIT Press: Cambridge, MA.

Adler, E S. (1989), Technology Strategy: a Guide to the Literatures,' Research in Technological Innovation,Management and Policy, 4, 25-151.

Anderson, R and M. L. Tushman (1990), 'Managing through Cycles of Technological Change,' Administrative

Science Quarterly, 35, 604-633.

Anderson, P and M. L Tushman (1991), 'Technological Discontinuities and Dominant Designs: a Cyclical

Model of Technological Change,' Research-Technology Management, 34 ,26-31 .

Argote, I_, S. L. Beckman and D. Epple (1990), The Persistence and Transfer of Learning in Industrial

Settings,' Management Science, 36,140-154.

Arrow, K. (1962), 'Economic Welfare and the Allocation of Resources for Invention,' in R. Nelson (ed.). TheRate and Direction of Inventive Activity. Princeton University Press: Princeton, NJ, pp. 609-625.

Bahrami, H. and S. Evans (1987), 'Stratocracy in High Technology Firms,' California Management Review,

30(FaU), 51-66.

Baker, W. E. (1990), "Market Networks and Corporate Behavior," American Journal of Sodotagy, 96, 589-625.

Banbury, C. M. and W. Mitchell (1995), The Effect of Introducing Important Incremental Innovations onMarket Share and Business Survival,' Strategic Management Journal, l6(Summer), 161—182. .

Bourgeois, L. J. and K. M. Eisenhardt (1987), 'Strategic Decision Processes in Silicon Valley: the Anatomy of

the Living Dead,' California Management Review, 30(Fall), 143-159.

Bourgeois, L. J. and K. M. Eisenhardt (1988), 'Strategic Decision Processes in High Velocity Environments:

Four Cases in the Microcomputer Industry,' Management Science, 34,816-835.

Brand, M. and W Emmons (1992), 'Note on Public Policy and the Media Industry,' Harvard Business School

Case No. 9-793-027, Harvard Business School.

Burgelman, R. A. (1983), "A Process Model of Internal Corporate Venturing in a Diversified Major Firms,'Administrative Science Quarterly, 28, 223-244.

Cohen, W. M. and R. C. Levin (1989), 'Empirical Studies of Innovation and Market Structure,' in R.Schmalensee and R. Willig (eds). Handbook of Industrial Organization. North-Holland: New York, pp.1059-1107.

Cohen, W. M. and D.A. Levinthal (1990), 'Absorptive Capacity: a New Perspective on Learning and

Innovation,' Administrative Science Quarterly, 35 ,128-152.

1 2 Q 1


Cohen, W M., R. C. Levin and P. C. Mowery (1987), 'Firm Size and R&D Intensity: a Re-examination,'

Journal of Industrial Economics, 26, 289-314..

Conner, K. R. (1988) 'Strategies for Product Cannibalism,' Strategic Management Journal, 9, 9-26.

Cooper, R. G. and D. Schendel (1976), 'Strategic Responses to Technological Threats,' Business Horizons, 19,

61-69.

Davits, S. (1979), The Diffusion of Process Innovations. Cambridge University Press: Cambridge.

Dorfmah, N. S. (1988), Innovation and Market Structure: Lesson Frmn the Computer and Semiconductor Industries.

Ballinger: Cambridge, MA.

Dunne, T., M. J. Roberts and L Samuelson (1988), 'Patterns of Firm Entry arid Exit in US. Manufacturing

Industries,' Rand Journal of Economics, 19,495-515.

Feinman, S. and W Fuentevilla (1976) "Indicators of International Trends in Technological Innovation. Final

Report to the National Science Foundation,' NTIS document PB-263-738, April:

Feller, W. (1951), 'The Influence of Market Structure on Technological Progress,' Quarterly Journal of

Economics, 55, 556-577.

Fichman, M. and D. A. Levinthal (1991) 'Honeymoons and the Liability of Adolescence: a New Perspective

on Duration Dependence in Social and Organizational Relationships,' Academy of Management Review, 16,

442-468.

Freeman, J. (1990), 'Ecological Analysis of Semiconductor Firm Mortality,' in J, V Singh (ed,), Organizational

Evolution: New Directions. Sage Publications: Newbury Park, CA, pp. 53-77.

Florida, R. and D. Browdy (1991), "The Invention That Got Away,' Technology Review, Aug^Sept., 43-54.

Foster, R. N. (1986), Innovation: Tie Attacker's Advantage. Summit Books: New York.

Hannan, M. T. and J. H. Freeman (1977), "The Population Ecology of Organizations,' American Journal of

Sociology, 82, 929-964.

Hannan, M. T. and J. H. Freeman (1989), Organizational Ecology. Harvard University Press: Cambridge, MA.

Henderson, R. M. and K. B. Clark (1990), 'Architectural Innovation: the Reconfiguration of Existing

Product Technologies and the Failure of Established Firms," Administrative Science Quarterly, 35, 9—30.

Jewkes, J., D. Sawyer and R. Stillerman (1958), The Sources of Invention. Macmillan: London.

Klepper, S. (1988), "Collaboration in Robotics,' in D. C. Mowery (ed.), International Collaborative Mntures in

U.S. Manufacturing. Ballinger: Cambridge, MA, pp. 223-265.

Kogut, B., G. Walker and D. J. Kim (1995), 'Cooperation and Entry Induction as an Extension of

Technological Rrvalry/'ftwarrf Policy, 24, 77-95.

Leonard-Barton, D. (1992), 'Core Capabilities and Core Rigidities: a Paradox in Managing New Product

Development,' Strategic Management Journal, 13, 111—125.

Mcthe.D. T. (1985), Technology, Transaction Costs and the Diffusion of Innovation. Doctoral Dissertation,

University of California.

MetW.D. T. (1991), Technological Competition in Global Industries: Marketing and Planning Strategies for American

Industry. Quorum Books: New York.

MetM, D. T. (1992), The Influence of Technology and Demand Factors on Firm Size and Industrial

Structure in the DRAM Market: 1973-1988," Research Policy, 21, 13-25.

Mezias, S. J. and M. A. Glynn (1993), 'The Three Faces of Corporate Renewal: Institution, Revolution and

Evolution,' Strategic Management Journal, 14, 77—101.

Mintzberg, H. (1983), Power In And Around Organizations. Prentice-Hall: Englewood Cliffe, NJ.

Mitchell, W (1989), 'Whether and When? Probability and Timing of Incumbents' Entry into EmergingIndustrial Subfidds," Administrative Sciena Quarterly, 34, 208-230.

1202


Nelson, R. R. and S G. Winter (1982), An Evolutionary Theory of Economic Change. Harvard University Press:

Cambridge, MA.

Pettigrew, A. (1973), The Politic; of Organizational Decision Making. Tavistock: London.

Pfeffer, J. and G. R. Salancik (1974), 'Organizational Decision Making as a Political Process,' Administrative

Science Quarterly, 19, 135-151.

Reinganum, J. E (1985), 'Innovation and Industry Evolution,' Quarterly Journal ofEconomics, 100, 81-99.

Rosenbloom, R. S. and K. Freeze (1985), 'Ampex Corporation and Video Innovation,' in R. S. Rosenbloom(ed.), Research on Tkbnological Innovation Management and Policy, 2. JAI Press: Greenwich, CT.

Scherer, F. M. (1983a), "Concentration, R&D and Productivity Change,' Southern Economic Journal, 50,

221-225.

Scherer, F. M. (1983b), The Propensity to Patent,' International JournalofIndustrialOrganization, 1,107-128.

Scherer, F. M. (1984), Innovation and Growth: Scbumpeterian Perspectives. MIT Press: Cambridge, MA.

Schumpeter, J. (1934), The Theory of Economic Development. Harvard University Press: Cambridge, MA.

Schumpeter,J. (1950), Capitalism,Sodalism and Democracy. Harper: New York.-

Singh, K. and W Mitchell (1996), 'Survival of Businesses Using Collaborative Relationships to

Commercialize Complex Goods,' Strategic Management Journal, 17, 169-196.

Smith, D. K. and R. C. Alexander (1988), Fumbling the Future: How Xerox Invented, then Ignored the First Personal

Computer. William Morrow: New York.

Tushman, M. L. and P Anderson (1986), Technological Discontinuities and Organizational Environments,'

Administrative Science Quarterly, 31,439-465.

Tushman, M. L , W. H. Newman and E. Romanelli (1986), 'Convergence and Upheaval: Managing the

Unsteady Pace of Organizational Evolution,' California Management Review, 29, 29-44.

Utterback, J. M. and W. J. Abernathy (1975), 'A Dynamic Model of Product and Process Innovation,' Omega,

3, 639-656.

von Hippel, E. (1988), The Sources of Innovation. Free Press: New York.

Wade, J. (1995), 'Dynamics of Organizational Communities and Technological Bandwagons: an Empirical

Investigation of Community Evolution in the Microprocessor Market,' Strategic Management Journal,

l6(Summer), 111-133.

Winter, S. G. (1990), 'Survival, Selection and Inheritance in Evolutionary Theories of Organization,' in J. VSingh (ed.), Organizational Evolution: New Direction. Sage: Newbury Park, CA, pp. 269-296.

Yamaguchi, K. (1991), Event History Analysis. Sage: Newbury Park, CA.

Zajac, E. J. and M. H. Bazerman (1991), 'Blind Spots in Industry and Competitor Analysis: Implications ofInterfirm (Mis)perceptions for Strategic Decisions,' Academy of Management Review, 16, 37-56.

1203

The Underemphasized Role of Established Firms as …willm/bio/cv/papers/ICC1996...The...

Documents

Transcript of The Underemphasized Role of Established Firms as …willm/bio/cv/papers/ICC1996...The...