Research Policy 42 (2013) 1210 1224

Contents lists available at SciVerse ScienceDirect

Research Policy

j o ur nal homep age: www.elsev ier .com

Techno e fDestru on

Anna Ber , Mia KITE Research Linkb CENTRIM (Ce ton Fa

a r t i c l

Article history:Received 15 NReceived in reAccepted 17 FAvailable onlin

Keywords:Creative accumTechnological discontinuityCreative destructionDisruptive innovationCompetence-destroying innovation

dust indu

expl-destsoletith de exp

undtial of new technologies and integrate them with existing capabilities. Moreover, we show how intensecompetition in the wake of technological discontinuities, driven entirely by incumbents, may insteadresult in late industry shakeouts. We develop and extend the notion of creative accumulation as a wayof conceptualizing the innovating capacity of the incumbents that appear to master such turbulence.Specically, we argue that creative accumulation requires rms to handle a triple challenge of simulta-

1. Introdu

The effeindustries ition and teccan lead tocomplete bClark, 19851993). As (Schumpetethe demisemain explato respond nation argudiscontinuoobsolete (Tnation focu

CorresponE-mail add

(C. Berggren), michaelghobd

0048-7333/$ http://dx.doi.oneously (a) ne-tuning and evolving existing technologies at a rapid pace, (b) acquiring and developingnew technologies and resources and (c) integrating novel and existing knowledge into superior productsand solutions.

2013 Elsevier B.V. All rights reserved.

ction

ct of discontinuous technological change on existings a central theme in the literature on industrial innova-hnological development. Technological discontinuities

intensied technological competition or even to areakdown of competitive patterns (Abernathy and; Anderson and Tushman, 1990; Utterback and Suarz,a consequence, a process of creative destructionr, 1942/1994) may unfold, which eventually leads to

of established rms. Extant literature suggests twonations for this observed difculty of incumbent rmsto discontinuous technological change. The rst expla-es that incumbents fail when competence-destroyingus innovations render their existing knowledge baseushman and Anderson, 1986), and the second expla-ses on customer and market dynamics, arguing that

ding author. Tel.: +46 13 282573; fax: +46 13 281101.resses: anna.bergek@liu.se (A. Bergek), christian.berggren@liu.sethomas.magnusson@liu.se (T. Magnusson),ay@gmail.com (M. Hobday).

incumbents are seriously challenged only when innovationsintroduce new performance attributes that incumbents perceiveas unimportant (e.g. Christensen, 1997/2003; Christensen andRosenbloom, 1995; Rosenbloom and Christensen, 1994).

While these two explanations present different perspectives onthe failure of leading rms, they share the common assumptionthat industry incumbents are burdened with core rigidities andthe legacy of old technology (Leonard-Barton, 1992). Therefore,technological discontinuities may open up possibilities for newentrants. This is a key assumption in several modern versions ofSchumpeterian creative destruction from the competence-basedstudies in Tushman and Anderson (1986) and Utterback and Suarz(1993) to the more recent market-based approach of disruptiveinnovation (Christensen, 2006). A complementary assumption isthat the absence of successful new entrants signies periods ofstability, in which industry incumbents will retain and strengthentheir competitive positions.

This paper challenges these assumptions. With detailed longi-tudinal analyses of two sets of comparative cases electric vehiclesand hybrid power-trains in the auto industry, micro turbines andcombined cycle gas turbines in the power generation sector weidentify over-optimism regarding new entrants abilities to disruptestablished industries, partially generated by the above theories.

see front matter 2013 Elsevier B.V. All rights reserved.rg/10.1016/j.respol.2013.02.009logical discontinuities and the challengction, disruption or creative accumulati

geka,, Christian Berggrena, Thomas Magnussona

Group, Department of Management and Engineering, Linkping University, SE-581 83ntre for Research in Innovation Management), The Freeman Centre, University of Brigh

e i n f o

ovember 2011vised form 14 February 2013ebruary 2013e 6 April 2013

ulation

a b s t r a c t

The creative destruction of existing inis a central theme in the literature oncompetence-based and market-basedously challenged only by competenceknowledge base or business models obpaper challenges these arguments. Windustries, we demonstrate that thesand disrupt established industries and/ locate / respol

or incumbent rms:?

chael Hobdaya,b

ping, Swedenlmer, Brighton, East Sussex BN1 9QE, England, United Kingdom

ries as a consequence of discontinuous technological changestrial innovation and technological development. Establishedanations of this phenomenon argue that incumbents are seri-roying or disruptive innovations, which make their existinge and leave them vulnerable to attacks from new entrants. Thisetailed empirical analyses of the automotive and gas turbinelanations overestimate the ability of new entrants to destroyerestimate the capacity of incumbents to perceive the poten-

A. Bergek et al. / Research Policy 42 (2013) 1210 1224 1211

We show how intense competition in the wake of technological dis-continuities, driven entirely by incumbents, may result in dramaticshakeouts in mature industries, but also how some incumbents arecapable of absorbing the new technologies and integrating themwith their eaccumulatiity of thesePavitt (1986which buildon our casedual naturethe accumuattackers aalso for incuare composof performaimplicationexploitativement ambidsearch and indicates thlation, rmthan separa

The papcompetencties of incumchange. Serationale foSection 4 dconsequencanalyses thgled in thecreative acConclusionspaper, and ithe organiz

2. Explainichange on

2.1. Schum

With t(1942/1994price compcost or quaof the profoundationsthis traditio1986) over invaders the destruand Christeest highlighdiscontinuorms in the(cf. Abernat

Several tunder whatlikely to reeral level, innovation ing values, have built their ability

as competence-destroying innovation (Tushman and Anderson,1986), architectural innovation (Henderson and Clark, 1990) anddisruptive innovation (Christensen and Rosenbloom, 1995) havebeen used to describe the main competitive challenge for incum-

Theationird cot-bas

mpe

ordines os e

s repabi1985eten

peteemend tendual ccing (Gilbsitio

theathyders

resuain

d anhoutrom bentsnd Ogo, 1vestminnoontrhange a te (Ttionsthe mpetg coMurmted

ew eon, ir pran af opd effMach

arket

rket-uousdustrtionstainin

trajxisting capabilities. We develop the notion of creativeon as a way of conceptualizing the innovating capac-

successful incumbents, a concept originally coined by) to describe the process of generating new knowledge,s on existing knowledge rather than replacing it. Based

analyses, we elaborate this concept and emphasize its, including accumulation as well as creativity: whereaslation aspect creates barriers to entry and offsets thedvantage, the creativity aspect poses severe challengesmbent rms, particularly in industries where productsed of a multitude of technologies and represent a varietynce attributes. This analysis has important managerials. A general suggestion for rms that need to deal with

and explorative activities simultaneously is to imple-extrous organizations, which separate various types of

innovation (Raisch and Birkinshaw, 2008). Our analysisat in a competition characterized by creative accumu-s need sustained efforts of dynamic integration rathertion.er is organized as follows. The next section outlinese-based and market-based explanations of the difcul-

bent rms to respond to discontinuous technologicalction 3 presents our research design, including ther case selection and methods for data collection, andescribes discontinuous technological change and itses in the two studied industries. Thereafter, Section 5e reasons why attackers failed and incumbents strug-

selected cases. It also elaborates on the concept ofcumulation as a means to explaining this. Section 6,, summarizes the analysis and the main ndings of thes followed by a nal section, Discussion, which outlinesational implications of creative accumulation.

ng the effects of discontinuous technologicalincumbents competitive positions

peterian innovation and creative destruction

he concept of creative destruction, Schumpeter, p. 84) was among the rst to draw attention frometition to competition which commands a decisivelity advantage and which strikes not at the marginsts and the outputs of the existing rms but at their

and their very lives. Much of the later work inn has emphasized the attackers advantage (Foster,incumbents: the creation is usually accomplished bynew rms or entrants from other industries whilection is suffered by the incumbents (Rosenbloomnsen, 1994, p. 656). The more general issue of inter-ted by Schumpeter is, however, the implications ofus innovation for the success or failure of innovatingir rivalry with both actual and potential competitorshy and Clark, 1985).heoretical frameworks have been developed to explain

circumstances discontinuous technological change issult in the demise of incumbent rms. On a gen-innovation scholars have argued that discontinuousexposes leading rms to situations where the exist-norms and structures upon which they traditionallya competitive edge, turn into rigidities that limit

to innovate (Leonard-Barton, 1992). Concepts such

bents. explanthe thmarke

2.2. Co

Accoutcomon rmvationand caClark, comptively.

Comimprovedge aand exindividenhanrms tive poexploit(Abernand Anas thetion (Mthat olthrougtions fincumCes aOrsenithat infuture

In ctally cproducobsoleinnovalower ous coexistin1990; associaring nAndersby the(Tushmways orespon1992;

2.3. M

Macontinand ininnova

Susmance rst two of these concepts are competence-baseds of incumbent failure, and can be contrasted withncept, disruptive innovation, which can be seen as aed explanation (cf. also Macher and Richman, 2004).

tence-based explanations of creative destruction

g to competence-based explanations, the competitivef a discontinuous innovation depend on its inuencexisting resources, skills and knowledge: some inno-ne and improve existing technological competenceslities, whereas other destroy them (Abernathy and). Tushman and Anderson (1986) label these types

ce-enhancing and competence-destroying, respec-

nce-enhancing discontinuities are order-of magnitudents in price/performance that build on existing knowl-skills (Tushman and Anderson, 1986). They rene

an established product design, usually by improvingomponents (Henderson and Clark, 1990). Competence-discontinuities are generally introduced by incumbentert, 2012), and also tend to reinforce the competi-ns of these rms, since they permit incumbents toir existing competences and increase barriers to entry

and Clark, 1985; Henderson and Clark, 1990; Tushmanon, 1986). Few, if any, new rms enter an industrylt of a competence-enhancing discontinuous innova-e and Garnsey, 2006). This explains the observationd large rms seem to have better chances of survival

industry evolution (Klepper, 1996) as well as observa-many sectors of stable patterns of innovation, in which

play the dominant role as persistent innovators (cf.rsenigo, 2001; Granstrand et al., 1997; Malerba and996; Pavitt, 1986). Several authors have also suggestedent in complementary assets and capabilities support

vation (Jacobides et al., 2006; Teece, 1986).ast, competence-destroying discontinuities fundamen-e the knowledge and skills required to develop andproduct and, therefore, make existing knowledge

ushman and Anderson, 1986). Competence-destroying tend to be introduced by new entrants and alsobarriers to entry for other new rms, since previ-ence-based competitive advantages are removed whenmpetence becomes obsolete (Anderson and Tushman,

ann and Frenken, 2002). These innovations are alsowith major changes in industry leadership, favou-ntrants at the expense of incumbents (Tushman and1986). Industry incumbents tend to be handicappedevious successes with the old technological paradigmnd Anderson, 1986); their existing skills, abilities anderating constrain their actions and make it difcult toectively (Abernathy and Clark, 1985; Leonard-Barton,er and Richman, 2004).

-based explanations of creative destruction

based explanations of the competitive outcomes of dis- change focus on the impact on performance trajectoriesies of a different dichotomy of innovations: sustaining

and disruptive innovations respectively.g innovations reinforce the established product perfor-

ectories in an industry (Christensen and Rosenbloom,

1212 A. Bergek et al. / Research Policy 42 (2013) 1210 1224

1995), by giving existing customers in mainstream markets some-thing more or better in the performance attribute they alreadyvalue (Bower and Christensen, 1995; Christensen, 1997/2003;Christensen and Bower, 1996). They build on established valuenetworks and, therefore, require no change in the innovatingrms strategic direction (Christensen and Rosenbloom, 1995;Rosenbloom and Christensen, 1994).

Industry incumbents are not always the rst to commercial-ize sustaining innovations (Christensen et al., 2002), but they willeventually take the lead in all sustaining innovations regardlessof whether they are competence-enhancing or competence-destroying (Christensen and Rosenbloom, 1995): . . . rarely haveeven the most radically difcult sustaining technologies precipi-tated the fa

In contrperformancnologies an1995).1 Theof performRosenbloomthey tend toued by mainthey will onthat value tBower and the mainstrKopalle, 20improve moto mainstrecompetitiveBower, 199and Christeogy (Adner,

A centraovershootinerally imprultimately aitated by suproduct pecustomers ences alongtheir attentmance requwill therefoattributes itet al., 2001)

Disruptientrants (Binitially by tive relativealternativesare inferiorcompanies(Christensecult to devolow-margin

1 Christensecompanies canspective is, howdisruptive innwith the threaperspective.

2 Performaat par with thamainstream cu

initiatives addressing the needs of known and powerful customers(Bower and Christensen, 1995).3

When the disruptive innovation starts to invade the main-stream market, incumbents will realize the threat, but by then itis often too late and the pioneering new entrants will come todominate the market (Bower and Christensen, 1995; Christensenand Rosenbloom, 1995). This is a pattern of change that canoccur in any product or service market (Christensen et al., 2004,p. 270). Thincumbentsand establi1994). In einability of

nsenonce

sust also

edict

comhat incuf dishat ifts. Thpproorgalocatcallytencory ates rce, tousltive otencnovaarketencforceccess

chalssumrial s

earc

se se

otedoth tpurpIn thontinting antaryses inticalle car

mben-end

2008)ilure of leading rms (Christensen, 1997/2003, p. xviii).ast, disruptive innovations imply a different package ofe attributes than those provided by mainstream tech-d valued by existing customers (Bower and Christensen,y also tend to redene the level, rate and direction

ance improvements in an industry (Christensen and, 1995). When these innovations are rst introduced,

underperform in the performance attributes most val-stream customers (cf. also Rosenberg, 1972). Therefore,ly attract niche customers in small or emerging marketsheir nonstandard performance attributes (Adner, 2002;Christensen, 1995), or the price-sensitive low end ofeam market (Christensen et al., 2003; Govindarajan and06). However, over time the disruptive innovation willre rapidly than the established technology with regardam performance attributes, making it performance-

also in the mainstream market (Christensen and6).2 It can then invade the mainstream market (Bowernsen, 1995) and even displace its established technol-

2002).l element of this framework is the idea of performanceg. According to Christensen (1997/2003), rms gen-ove technologies faster than their customers need orre willing to pay for. The invasion from below is facil-ch overshooting by incumbents technologies; when

rformance is higher than what the market demands,will no longer make product choices based on differ-

established performance parameters, but will turnion to alternative parameters. Once the main perfor-irements of mainstream customers are satised, theyre embrace the disruptive innovation based on the new

offers, e.g. reliability, convenience or cost (Christensen.ve innovations are most often pioneered by newower and Christensen, 1995) and tend to be ignoredincumbents, for whom they are nancially unattrac-

to existing prot models and competing investment (Christensen, 2006). Because disruptive technologies

to mainstream products in the beginning, the leading most attractive customers typically will not use themn et al., 2001), and these rms will nd it very dif-te sufcient resources to develop solutions for smaller

segments, especially when such projects compete with

n et al. (2002) describe disruptive innovation as a strategy that choose to turn ideas into plans for new growth businesses. This per-ever, quite different from that presented in most of the literature on

ovation, which rather takes the perspective of incumbent rms facedt from disruptive innovation. In this paper the focus in on the latter

nce-competitive does not necessarily mean that the performance ist of the established technology; it merely has to be sufcient to satisfystomers (Adner, 2002; Govindarajan and Kopalle, 2006).

Christeative chave a94; see

2.4. Pr

Thesomewwhich wake oment tentranboth adue to fore alspecicompetrajectattribu

Henbe seridisrupcompethat intake mcompeto reinfew suvationthus, aindust

3. Res

3.1. Ca

As nused bon the cases. of discof exisplemeprocestheoreand th

3 Incufrom lowDruehl, e main reason is that disruptive innovations require to rethink their business models (Christensen, 2006)sh new value networks (Rosenbloom and Christensen,ssence, the attackers advantage thus stems from theincumbent rms to change strategies (Rosenbloom and, 1994). This implies that disruptive innovation is a rel-pt: something that is disruptive in one company canaining impact on another (Christensen et al., 2001, p.

Christensen, 2006).

ions of competitive outcomes: a synthesis

petence-based and market-based explanations makedifferent predictions about the circumstances undermbents might lose their competitive positions in thecontinuous technological change. Still there is an agree-

incumbents lose their positions, they lose them to newe explanation for the attackers advantage is similar inaches: incumbents are unable or unwilling to respondnizational, technological and strategic inertia and there-e insufcient resources in response to the threat. More, incumbents lose their positions because their oldes are destroyed or because the current performancend value network are disrupted as new performanceeplace existing ones as the prime basis for competition.he general prediction is that industry incumbents willy threatened by technological discontinuities that arer competence-destroying. Under such circumstances,

e-based as well as market-based explanations predicttions will be pioneered by new entrants which willt shares from incumbents. Conversely, sustaining ande-enhancing technological discontinuities are predicted

the competitive positions of incumbents; there will beful new entrants and incumbents will master any inno-lenges they are faced with. Implicitly, both approaches,e that the absence of new entrants is a key indicator oftability and the persistence of industry incumbents.

h design

lection

by Gerring (2008), case study methodologies have beeno test existing and to generate new theories. Dependingose, different principles may be applied for selecting theis paper we seek to further develop the understandinguous innovation by critically examining the predictionsttackers advantage theories, and by developing a com-

approach for understanding discontinuous innovation specic types of industries. For this purpose, we havey sampled two industry cases: the gas turbine industry

industry (in particular power-trains). These industry

ts may be particularly inclined to overlook innovations that diffuse market segments upward toward high-end segments (Schmidt and.

A. Bergek et al. / Research Policy 42 (2013) 1210 1224 1213

cases have two characteristics which make them suitable for thepurpose of this paper:

(1) Both industries have been characterized as Schumpeter Mark IIindustripopulaton the obe comppatternbe espe

In botnologicaand potket posincumbdevelopant com(2001, potentiaargued tive. Incharacttion tecto serveplayers (1997/2disrupticle is noarchitecentire vDyersonindustrion the or discoSeawrig

(2) Both inencing)the prenicantincumbcompetlished cremainepetitivecase, indforced ship is bThe twoi.e. casethe genetherefortions (c

The selewhere the destroying at the samehave been fthan existin

The cassince such commerciature both trms, we stdestroying and electric

competence-enhancing innovation (combined-cycle gas turbines(CCGT) and hybrid-electric vehicles respectively).

Our investigation of the competence-destroying and poten-tially disruptive innovations includes new entrants and their

s well as the actions taken by incumbents in face of theial thherstaiweredustrm Eugest has ne inne maand

a rahave-train2011

se ch

induered, engts ancienft anginetencals, hter-aedges, bugrateases

the

uct tacteral goized et ch

simstry ate gegiorencer maventcles o

perssed ersopetitgies i

owinuld bbineed ads.nt paes, i.e. industries with a concentrated and rather stableion of innovators (cf. Breschi et al., 2000). This suggestsne hand that incumbent rms in these industries shouldetent innovators that are not easily shaken by normal

s of innovation, and on the other hand that they mightcially vulnerable to discontinuous technological change.h industries, there have also been predictions that tech-l discontinuities in the form of competence-destroying

entially disruptive innovations would threaten the mar-itions of incumbent rms. In the gas turbine industry,ents were predicted to be challenged by new entrantsing microturbines. For example, management consult-pany Arthur D. Little (2000) and Christensen et al.

p. 93) both explicitly identied microturbines as ally disruptive innovation and Christensen (2001, p. 13)that microturbines could prove to be very disrup-

addition, Hart (2005) predicted that the disruptiveeristics of microturbines and other distributed genera-hnologies would create opportunities for new entrants

the needs of the users much better than incumbentin the energy sector. In the car industry, Christensen003) argued that electric vehicles have the smell of ave technology (p. 239) and that . . . the electric vehi-t only a disruptive innovation, but it involves massivetural reconguration . . . that must occur . . . across thealue chain . . . (p. 252) (cf. also Aggeri et al., 2009;

and Pilkington, 2004; Sierzchula et al., 2012). The twoes can, thus, be described as typical cases of industriesverge of creative destruction, suitable for conrmingnrming the theory of the attackers advantage (cf.ht and Gerring, 2008).dustries have experienced (or are currently experi-

discontinuous change of a very different kind thandicted attackers advantage: hardly any new sig-

entry but instead intensied competition betweenents. These developments could be interpreted asence-enhancing and sustaining innovation, since estab-ompetences and basic performance attributes haved important. However, the effects on incumbents com-

positions have still been dramatic. In the gas turbineustry leadership changed and major incumbents were

out of the industry; in the car case industry leader-eing profoundly challenged after 80 years of stability.

industries can, therefore, be seen as deviant cases,s that demonstrate a surprising pattern in relation toral understanding of technological discontinuities, and,e, are useful when searching for alternative explana-f. Seawright and Gerring, 2008).

cted cases, thus, represent two important industriespredictions of existing frameworks on competence-and disruptive innovation have not materialized, while

time the innovation processes actually taking placear more dramatic and difcult for incumbents to handleg theories tend to assume.e studies focus on innovations in core subsystems,innovations are likely to have the most signicant

l implications (Gatignon et al., 2002). In order to cap-he predicted and the actual challenges for incumbentudy two sub-cases in each industry: one competence-and potentially disruptive innovation (microturbines

vehicles (EVs) respectively), and one sustaining and

fates apotentitly), wand suthere car iners frothe larChina tinctivChinesturers GM asrms poweret al., EVs.

3.2. Ca

Theenginedesignponenset of s(Rycrotion encompematericompuknowlsystemto inteedge b

But

Prodcharcapitdard

Markwithinduoperand rdiffeof caintervehi

Timea clo(Andcomnolo

Folltry wogas turidentiital goodifferereats (although all incumbents did not respond explic-eas the investigation of the competence-enhancingning innovations primarily includes incumbents as

no, or very few, new entrants. The study of they is delimited to the established core manufactur-rope, Japan and the U.S. China has recently becomesingle car market, but in spite of this strong growth,ot produced any major car manufacturer with a dis-ovation strategy and international presence, and therket continues to be dominated by foreign manufac-their local afliations, with Volkswagen as No. 1, andther distant second. Despite high ambitions, Chinese

also failed to develop cost competitive alternatives, electric or hybrid-electric (cf. McKinsey, 2012; Yao) and are therefore not included in the section on

aracteristics and comparability

stries are similar in that they both supply composite, products. The development of such products involvesineering, sourcing and integration of both physical com-d sub-systems as well as the integration of a diversetic and engineering knowledge bases and capabilitiesd Kash, 2002). Improving the performance of combus-s, for example, requires a combination of technologicalies from elds of knowledge as diverse as mechanics,eat transfer, combustion, uid ow, electronics andided design (Granstrand et al., 1997). The requisite

may be related to individual components and sub-t also to architectural knowledge, i.e., knowledge of how

various components and relevant engineering knowl-(Henderson and Clark, 1990; Takeishi, 2002).industries also differ along several dimensions:

ype and production scale. The gas turbine industry isized by customized, small-batch production of complexods, whereas the car industry is characterized by stan-mass production of complex consumer goods.aracteristics. The gas turbine industry is strongly globalilar customer demands across major markets. The caris also global in the sense that major manufacturerslobally, but markets are strongly inuenced by nationalnal regulations and customer demands, with importants between North America, Japan and Europe. The fatesnufacturers are also inuenced by direct governmentions, such as incentives in support of specic types ofr bailouts to avoid bankruptcies of national icons.pective. The gas turbine case can be considered ascase, in the sense that the studied era of fermentn and Tushman, 1990) has passed. By contrast, theion in the car industry between rms and between tech-s still ongoing and uncertainty remains high.

g the taxonomy outlined by Pavitt (1984), the car indus-e categorized as a scale-intensive industry, whereas the

industry would be included in one of the sectors Pavitts an extension of his original taxonomy: expensive cap-

The two industries can therefore be expected to displaytterns of innovation.

1214 A. Bergek et al. / Research Policy 42 (2013) 1210 1224

Table 1Overview of data types and sources.

Industry case Sub-case Interviews Annual reports Trade press Other sources

Gas turbine industry Microturbines Corporate manager ABB (2000), Sales performance Technical

egies,enges

Car industry cial d

3.3. Data so

The straadapted to vides an ovindicators a

Annual CCGT sub-cbecause of panys aftesuccessful iused to getmicroturbinextensive inwhen the blected fromhad access piled by Jim2008). In thactive compMagnussonby manufacstatistics frtry, the studbusiness pusuch as Autfrequent tessource of inlaunches in

These setechnical einterviews well as issentrants things. Structstudies of tthis speciized codingindustry, wmarket achuct performproblem ofrounded byas the car ileading carhydrogen-bBakker, 201

cont

e gas

Micro

he ead elehe esn ec

plancity s

userted bn los

dem conbus

ells, fs, they onis waat haicrotuse ane, pre wpor

lectrf the cale tly hR&D manager Alstom (2001), R&Dmanagers, Marketing managersand CEOs at Capstone (2001, 2002)and Turbec (2001), Customers andEnergy experts (2001, 2002)

CCGT Marketing managers, Engineersand R&D managers at ABB (1998,1999) and Siemens (2004, 2008)

Stratchall

Electric vehicles Technical experts Finan

Hybrid-electricvehicles

Hybrid R&D managers at Toyota(1999, 2006), GM (2007), VW (VPR&D 2008)

urces

tegies for data selection and collection have beenthe differences between the sub-cases. Table 1 pro-erview of the types of data collected, and the differentnd sources used.reports have been used extensively, especially in thease. The reports from GE were especially valuabletheir level of detail and frankness regarding the com-r-delivery problems (maybe because GE was the mostncumbent in solving them). Annual reports were also

nancial data in the electric vehicle sub-case, and fores, where the leading manufacturer Capstone providedformation to infuse condence among investors evenusiness was developing slowly. Market data were col-

several different sources. In the CCGT sub-case, weto a database containing global market orders com-

Watson and colleagues at SPRU (see also Bergek et al.,e microturbine sub-case, sales performance data for theanies were retrieved from their annual reports (see also, 2003). In the car industry, product announcementsturers were complemented by publicly available salesom, e.g., the U.S. Department of Energy. In this indus-y could also benet from a range of publicly availableblications and trade journals: professional publicationsomotive News as well as popular magazines with theirting of new products. Trade press was also an importantformation about technical performance and product

gas turbine industry.condary data were complemented by interviews withxperts and R&D managers. At the incumbents, thesefocused on technical forecasts and comparisons, as

4. Dis

4.1. Th

4.1.1. disrupt

In ttributewith twesterpowerelectrion thesuppormissiomatch

Thenal comsolar coptioning onlthat thogy thThe mcould kerose

Themost imtion elevel olarge-snicanues related to future product development; at newey also covered commercial prospects for new offer-ured interview guides were prepared based on carefulhe relevant sub-industries and rms. The analysis ofc type of expert interviews did not build on formal-

protocols; more important, in particular in the caras to check statements related to performance andievements with identiable public sources, e.g. prod-ance tests and sales data. This is related to the crucial

evaluating trends and projections in an industry sur- so much publicity and hype regarding new technologiesndustry. The bold predictions in the 1990s, also from

industry executives, concerning a rapid transition toased vehicles here served as a cautionary tale (cf.0).

than the lain other pefor simple cthe excess hresidential that standagrid, an inscould be compreparationplant. Withdesign was the unit cosof a completurbine wastechnologyperformance

TechnicalperformanceProduct launches

Database on global orders

ata Sales & pricesProduct testsMarket data

Product announcements bymanufacturersResearch reports (case studiesof new entrants)

Sales & pricesProduct testsMarket data

Technical reportsDoE sales statistics

inuous technological change in two industries

turbine industry

turbines: a distributed technology that failed to

rly 2000s, there were high hopes that the concept of dis-ctric power generation would represent a major breaktablished structure of electric power supply in mostonomies. Instead of relying on a few centrally locatedts and long-distance transmission lines, distributedystems would comprise a large number of generators

side of the distribution grid, producing power locallyy smart grids. This would result in reduced power trans-ses and enable quick, incremental capacity additions toand.cept included both mature technologies such as inter-tion engines and more novel ones such as photovoltaicuel cells and microturbines. Among these technological

microturbine stood out by its simple design, compris-e moving part, and microturbine manufacturers argueds the only new small-scale power generation technol-d reached a competitive cost level (Magnusson, 2003).urbine was also exible in terms of the fuel source; it

variety of hydrocarbon fuels, e.g. natural gas, biogas,ropane and diesel.ere also several drawbacks, however. In terms of thetant performance parameter for thermal power genera-ical efciency the microturbine only achieved half thecombined cycle gas turbine, which was the dominatingtechnology. Moreover, the microturbine featured sig-igher capital cost per unit of power generation capacity

rge turbines. However, the microturbine offered valuerformance attributes, such as exibility and a potentialombined heat and power (CHP) applications, in whicheat from the power generation would be used to heat

buildings, hotels, ofces, swimming pools, etc. Providedrds were available for interconnection with the electrictallation could be completed within a few weeks. This

pared with the several years of planning, negotiations,s and construction work needed for a new large power

a standardized product conguration, the microturbinealso much simpler and due to the much smaller unit size,t of a microturbine generator was much lower than thatte combined cycle gas turbine power plant. The micro-, thus, more exible and convenient than the large scale.

A. Bergek et al. / Research Policy 42 (2013) 1210 1224 1215

Located in southern California, the new technology based rmCapstone Turbine took the role as the leading microturbine manu-facturer, entering the market in 1998. In the early 2000s, Capstoneoffered microturbine based solutions for a broad range of applica-tions includsewage plaextender fopower geneoperated inthe market,a joint ventand the jet in 1998 wiR&D activitbased rm Ingersoll-Raover, establthat they p(Magnusson

Accordinstrategies fgies, eithermarkets. Toa number oing marketson the existwhich the lond, new enon a low-coit should (1do it themswho will wmore easily(Christense

The highit difcult strategy. Mfor the opepower prodcantly lowemicroturbinmarkets. Thwas completion, e.g. hoHaving neitpower prodsimple desithe microtustable acceload, when possibility tto the Califoparticularlyturbine supcapacity exapplicationthe technol

Bower abent rms keep the incumbent microturbinas an indepeorganizatiomicroturbinthe Italian

sales. The second large rm to enter the market with a microtur-bine CHP unit, American Ingersoll-Rand, was more persistent, butin 2011 also this rm divested their microturbine business. Otherrms that announced activities in microturbine technology in the

000s, e.g. GE, Alstom and Bowman, either did not present anyts or failed to get any market foothold. In 2011, ten yearshe initial optimism and thirteen years after the rst prod-nch, Capstone was the only microturbine manufacturer thaty hadncreut stn terleadtion.h theas tu) inies anork

emesrupch aics oould

le. Thg maouldn.ing e tecctriced tly hat perw mly lim

erosrematechnbentsponetion , the

on mts. Th

mation wer, srers. t efis unenou

Compent ile droturied pmeAdva

tone resp

(502ing resource recovery (using waste gases from landlls,nts, wellheads, etc.), hybrid-electric vehicles (ranger battery-electric buses), CHP (combined heat andration) and secure power (auxiliary power units to be

case of power outage). Other manufacturers entered such as the start-up Turbec (Sweden), emanating fromure between the large scale turbine manufacturer ABBengine manufacturer Volvo Aero. This rm was set upth the intention to commercialize results from jointies since the early 1990s. The British new technologyBowman and the large and diversied American rmnd also entered the market in the early 2000s. More-ished manufacturers such as Alstom and GE announcedursued R&D in the area of microturbine technology

et al., 2005).g to Christensen et al. (2002) there are two alternativeor market entry with potentially disruptive technolo-

via the low-end tiers of existing markets or via new identify the most viable strategy, these authors offerf litmus tests. To enter via the low-end tiers of exist-, it is necessary to pass two litmus tests. First, productsing market need to overshoot the level of performanceeast demanding customers are willing to pay for. Sec-trants need to create a business model which is basedst strategy. If the rm instead enters via new markets,) target customers who in the past havent been able toelves for lack of money or skills, (2) aim at customerselcome a simple product and (3) help customers do

and effectively what they are already trying to don et al., 2002, pp. 2425).

capital cost per unit of power generation capacity madefor microturbine manufacturers to adopt a low-costoreover, fuel expenses were still a determining factorrating costs of thermal power generation and electricucers would hardly adopt technologies with signi-r efciencies than other available technologies. Thee manufacturers therefore preferred to enter via newey explicitly targeted customers whose core businesstely different from large scale electric power genera-tels, restaurants, ofces and small manufacturing rms.her the skill, nor the money or ambition to become largeucers, these customers appreciated the low unit cost,gn and easy installation of the machine. By means ofrbine, it was possible for these customers to ensure a

ss to electricity, and by running the machine at peak-electricity prices were high, the microturbine offered ao reduce the sensitivity to volatile electricity prices. Duernian energy crisis in the early 2000s, this had become a

urgent issue in the home market for the leading micro-plier Capstone. Since the microturbine enabled quicktensions with high systems efciency if used in CHP-s, policy makers also supported commercialization ofogy through nancial incentives.nd Christensen (1995, p. 52) argue that also large incum-can manage disruptive innovation, provided that theydisruptive organization independent. ABB, the onlyturbine manufacturer that entered the market with ae, strictly followed this advice. They established Turbecndent business unit, separated both geographically andnally from its large scale parent. However, in 2004 thee business had not taken off and ABB sold off Turbec torm API Com S.r.I., which still by 2011 had very limited

early 2producafter tuct lauactuallstone iUSD, beries ito the genera

Botbent gTurbecturbinframewManagas a ditoo muthe logeters wvaluabexistinkets whappe

Durturbinand eleremainstill ontal cosThe neseverein anystone niche incum

ProdisrupClearlymentsbenestreamdisrupHowevufactumodesket, it good bines.

4.1.2. incumb

Whof micintensdevelobines.

4 Caps91.9 MWturbine a position on the market. Between 2007 and 2011 Cap-ased its revenue almost four-fold, from 21 to 82 millionill struggled to make any prot. Moreover, its deliv-ms of installed power were marginal when compareding manufacturers of large turbines for electric power4

new entrants (such as Capstone) and some incum-rbine manufacturers (such as ABB through its start-uptially believed in the disruptive potential of micro-d some even incorporated the disruptive innovation

as part of their marketing strategy (cf. Capstonent, 2009; Murelio, 2000; Tanner, n.a.). The classicationtive technology meant that they did not have to worrybout the inferior efciency of their product. Followingf disruption, the value of existing performance param-

diminish and new parameters would become moree new markets would grow and eventually erode thein markets for large turbines. The entry via new mar-

set these disruptive processes into motion. This did not

the time it had been available on the market, micro-hnology developed very slowly both in terms of costsal efciency. In 2011, its efciency performance, whichhe key performance parameter on main markets, waslf of the performance of large-scale CCGT and the capi-

kW between two and three times as high (see Table 2).arkets targeted by microturbine manufacturers wereited in size and did not grow at a pace that resulted

ion of the main market for large turbines. Thus Cap-ined a niche player and the microturbine remained aology that did not pose any real threat for industry.nts of disruptive innovation theories may argue thatis only a matter of time (cf. Nair and Ahlstrom, 2003).re is a limit at which the cost for additional improve-ainstream performance parameters will exceed the

is limit will determine what is good enough on main-rkets. According to this argument, opportunities forill emerge, when the large turbines reach this plateau.

uch an argument is of limited use for microturbine man-Given the physical limitations of the technology and theciency improvements during its rst decade on the mar-likely that microturbine technology ever will becomegh in a comparison with large combined cycle gas tur-

etition in large gas turbines: a life and death race forrms

isruptive innovation theorists focused on the prospectbines, the real industrial drama was played out in antechnological competition among incumbents in thent of even more efcient and powerful large gas tur-nced gas turbines are a key building block in combined

s booked orders for the scal years 2010 and 2011 were 77.2 MW andectively. This can be compared to the size of one single large-scale gas000 MW, see Table 1).

1216 A. Bergek et al. / Research Policy 42 (2013) 1210 1224

Table 2Performance characteristics of microturbines and combined cycle gas turbines.

Microturbine Combined cycle gas turbine

2002 2011 2002 2011

Typical size W Capital cost Thermal elec

Sources: Green EPA (2

cycle poweachieve elethose of othket for elecin the latebines for cointroduced launched ations, whicpower planogy.

From jusity in 1987share to ovchallenge frhouse was in collaboraThis new msuch as supof the four slightly betABB was thrable to GEyear later, htheir GT24/sequential pressor, whefciency. Idisastrous tnew genera501G in 199Unlike ABBponent imp

All compthus, able tthe same rfered from During the of new metgies. This oturbine blaing systemsturbine inlehigher efcnew proble

Accordinrequires a takes a comunderstoodbe a disastethe only chinnovation very high unto run full-neers to venew produc

he med w

thatth aflity w

bothghou

wayry.an bbal m80s er wmpege foth W

GE g. 1bm a k

wass of ilemeand d

basand litieso-stcould

accut bacive i

fromcreatneympey repcate

the e resulved

marcientmpeartnce coss, cacal, logiced to

racerange (MW) 0.030.4 MW 0.030.25 M(USD/kW) 12001700 12901410 trical efciency 1430% 2333%

e and Hammerschlag (2000), Smith (2002), California Energy Commission (2012),

r plants, which combine gas and steam turbines toctrical efciencies that are almost 50% higher thaner fossil-fuel power stations. Whereas the total mar-tric power generation equipment grew very slowly

1980s and early 1990s, sales of advanced gas tur-mbined-cycle applications suddenly took off when GEits innovative Frame 7F in 1986. With Frame F, GE

combination of architectural and component innova-h increased the thermal efciency of combined-cyclets to 54%, far above that of any other fossil-fuel technol-

t a small fraction of installed power generation capac-, combined-cycle gas turbines increased their marketer 35% in 1993. To cope with this expansion and theom GE, all its competitors invested heavily. Westing-the rst to respond with its own F turbine, developedtion with Mitsubishi. Siemens followed with its V94.3.achine comprised many advanced component features,ersonic blades in the compressor and cooling in threeturbine blade stages. It was larger (200 MW) and alsoter in terms of thermal efciency than GEs 7F turbine.e last of the incumbents to launch a turbine compa-s Frame 7F, introducing the GT13E2 turbine. Only oneowever, ABB took the technological lead by introducing26. This was a very different machine, featuring a newcombustion chamber design and a new type of com-ich gave another 2 percentage point increase in thermalt sold very well initially but after a few years ran intoechnological difculties. The other companies launchedtions with similar thermal efciencies: Westinghouses4, Siemens V84.3 in 1995, and GEs Frame 7G in 1995.

s machine, these turbines were largely based on com-rovements such as blade design, materials and cooling.anies that started in the large gas turbine race were,o launch successive generations of turbines at aboutate and performance. However, all of them also suf-difcult periods of customer-experienced problems.

development process, the rms had introduced a rangehods, new materials and new manufacturing technolo-pened up possibilities in terms of more heat-resistantdes, more optimized designs and more intricate cool-. But the use of e.g. elevated compression ratios, highert temperatures and lean combustion processes to reachiencies and lower NOx emissions resulted in entirelyms, which the engineers could not foresee.g to R&D managers, developing new gas turbinestruly multidisciplinary process, and if one disciplinemanding role without considering the implications as

from other disciplinary perspectives, the result couldr. Balancing and integrating expert knowledge is not

when tbe solvmeansing wireliabition ofWestinfactoryindust

As cthis glolate 19(togethfour cochallencally. Bleaving(see Fi

Frothat GEaspectative ebroad vided aother hcapabitial, twwhich cientwithoua massturbinewith in& Whihigh tequentlto allocausednomichad soon theinsufbine coto its presourbusinenologitechnodeploy

The

allenge in this development process. A factor makingin advanced gas turbines particularly uncertain is theit cost of a new machine and the large facilities needed

load tests of new turbines. Thus it is difcult for engi-rify the real-world performance and reliability of theirt. The problems had, therefore, not yet been discovered

power genenot be descrEstablishedmance parno new enantees for 502000 MW 502000 MW500+ 500+4857% 5060%

008) and Capstone (2011).

achines were delivered to the rst customers and had tohen the turbines had already been commissioned. This

introducing advanced products was not enough. Deal-ter-delivery problems and upholding high operationalas a moment of truth, requiring on-the-eld applica-

specialized and integrated knowledge. Neither ABB, norse was able to solve their technical problems in a satis-

and they both had to exit the entire power generation

e seen from Fig. 1a, the position of the leading rms onarket (measured by market shares) was quite even in

and early 1990s. GE, Siemens, ABB and Westinghouseith its partner and licensee Mitsubishi) were the toptitors. The intense technology battle implied a greator the companies, and their fortunes differed dramati-

estinghouse and ABB were forced to exit the industry,as the obvious leader and Siemens as a distant second).nowledge and innovation perspective it may be argued

able to master both the creation and the accumulationts innovation process. The company embedded the cre-nts, the Frame F design with its new components, in aeep pool of knowledge and experience. This also pro-

is for its superior problem-solving abilities. ABB on thesuffered from a lack of knowledge depth and integrative

at key R&D positions. Its GT24/26 design with a sequen-age combustion process involved high technical risks

not be tested in its in-house facilities. In spite of insuf-mulated knowledge, a risky sales process was launched,k-up in deep problem-solving abilities. The result wasnnovation failure. In contrast to ABB, the new V84.3A

Siemens was based largely on its previous products,sed temperature and airow to improve efciency. Pratt

also added new knowledge regarding blade designs,rature materials, and cooling systems. Siemens subse-orted vibration problems in its new turbines and hadsignicant resources to solving these problems, whichntire power generation segment to report negative eco-lts in 1999. However, a few years later the companythese problems and could remain a strong number twoket with a share of just over 20%. Westinghouse had

depth in CCGT technology, and much of its gas tur-tence pertaining to its original design was transferreder Mitsubishi. Its turbine problems were aggravated bynstraints due to signicant losses in its nuclear powerused by litigation costs. Thus, the company lacked tech-nancial, and managerial resources to participate in theal race, especially when substantial efforts had to be

solve after-delivery problems. to develop advanced gas turbines for combined cycle

ration is a case of discontinuous innovation that can-ibed according to the attackers advantage frameworks.

competences were not destroyed, the main perfor-ameter, thermal electrical efciency, prevailed, andtrants succeeded. Still, the incumbents had no guar-sustaining their position, and two incumbents with

A. Bergek et al. / Research Policy 42 (2013) 1210 1224 1217

28%

13%

9%

8%

1987-1991a b

4%

1%

11%

8%1%

1998-2002

GE

Fig. 1. Global E liceprimarily Mits r is re

a legacy ofthe marketestablishedillustrates tmajor indusand compet

4.2. Power-

4.2.1. ElectrElectrica

were outco70 years inwhen GM uCalifornia lmandate, wemission-frsales: 2% in 2009). Follotery electricand Toyota 1500 units. aged to ghat all majorbattery elec

Ten yeaemerged. Ccommerciateries baseoptimistic China. In tputting onekey milestoensuring thfacturing ineven more electrics oning Chinese2012).

Howeveproductionthe crucial plug-ins, w

5 The electriits EVs. After bat the end of li

f 1212 shgureions iforn

rancadmand,ationformsive havenal ICer, ae wilayeccelehortrformfar frablish incing E

in 1 a

a trtakent, annag

mpanankr

Wheber oation19%

18%

5%

21%

market share based on total CCGT-orders in 19871991 vs. 19992002. Note: Gubishi. (For interpretation of the references to color in this gure legend, the reade

industry leadership were forced to withdraw from because of their failure to innovate on the basis of

technologies. The case of advanced large gas turbineshat intensied technological competition can triggertry change, also when it does not entail any disruptiveence-destroying innovations.

train competition in the car industry

ic vehicles: a source of creative destruction?lly powered vehicles are as old as the car industry, butmpeted by gasoline cars in the 1910s. However, after

obscurity they received a new chance in the 1990snveiled its electric Impact car (Shnayerson, 1996) andegislated the so-called Zero Emission Vehicles (ZEV)hich required major car manufacturers to introduceee power-trains for an increasing part of their vehicle1998, 5% in 2001, and 10% in 2003 (Sperling and Gordon,wing the ZEV mandate, all incumbents launched bat-

cars, such as GM EV1, Honda EV Plus, Ford Ranger EV,RAV4 EV, but none of these were produced in more thanWhen lawyers engaged by the automotive majors man-t back the ZEV mandate in 2002, EV-production ceased

car manufacturers and no incumbent tried to continuetric vehicles as a commercial operation.5

rs later, however, a new breed of electric vehiclesompared to the previous period these vehicles have al orientation and feature important advances in bat-d on lithium-ion technologies, which have inspiredofcial forecasts of EV penetration, from the U.S. tohe U.S., President Obama set the goal in 2009 of a

million electric vehicles on the road by 2015 as ane toward dramatically reducing dependence on oil andat America leads in the growing electric vehicle manu-

sales ofor 20bleak limitatthe Caltions innickel 1990s expectEV perof mastimes ventiohowevpetitivniche prapid atively sthe peis still the est

BotemergStartedmodeltive tobeing sold ounot mathe cowent b2011).a numexpectdustry (USDOE, 2011). In China, the government set anchallenging goal: ve million electric and plugin hybrid

the road by 2020, with the similar purpose of achiev- leadership in the new promising industry (McKinsey,

r, despite major investments in product development, and marketing, sales have remained disappointing inU.S. market. In 2011, less than 18,000 EVs includingere sold in the U.S. (which could be compared with total

c cars were physically destroyed, especially so by GM who only leasedeing pressured by lobbyists and customers, Toyota made an exceptionfe of its RAV4 EV and actually sold 300 vehicles.

delivered 2order to preof 2012 (Kafunding anits equally not secure 2012). Seve20082009and a survWheego, Ap(Webb, 201challenging

Instead,in interest54%

Siemens

ABB

Westinghouse

Westinghouse licensees

GE licensees

Other

nsees include primarily Alstom; Westinghouse licensees includeferred to the web version of the article.)

.8 million passenger vehicles this year), and forecastsowed a very small increase (Wahlman, 2012). Theses are related to the persistent price and performanceof electric vehicles. The EVs launched as a response toian ZEV mandate were highly priced, had severe limita-ge and speed, plus long charging times. Developments inium and nickel metal hydride technologies in the late

more recently, in lithium ion-batteries, have createds of a breakthrough for EVs. But as shown in Table 3,ance has improved only slowly since the 1990s. In spitetechnological investments, driving range and charging

not changed signicantly, and remain far from con-E-cars or hybrids. Prices are coming down somewhat,

nd with government subsidies EVs could become com-th advanced conventional cars in specic segments. Ar such as Tesla Roadster (not in table) stands out with itsration, longer driving range (about 200 miles), and rela-

charging time (less than 4 h) but so does its price. Thus,ance of EVs with regard to key performance attributes

om competitive enough to make them able to challengehed technology.umbents and new entrants have been active in theV market. One of the EV pioneers was Norwegian Think.991, Think developed and introduced a rst productioncompact city car positioned as a convenient alterna-aditional car but then went bankrupt in 1998. After

over by Ford, Think produced 1000 cars, before Fordd production ceased (Roste, 2009). The new owners dide to make any business out of the EV production andy has been haunted by nancial problems. In 2011 itupt for the fourth time in its 20-year history (Bolduc,n the EV market came to life again in the late 2000s,f high-proled U.S. startups contributed to the highs on EVs (Guilford, 2011). From 2008 to 2011, Tesla

500 of its roadster before discontinuing production inpare for a new model to be launched in the second halfrlberg, 2011). Fisker raised 860 million USD in privated a substantial low-interest USDOE-loan and launchedhigh-priced plug-in Karma in January 2012, but couldresources for its second model (Rechtin and Lindsay,ral other new entrants announced their U.S. entry in, but three years later they had not started production,ey in Automotive News showed these rms Bright,tera Motors, ZAP Jonway and Coda to be in dire straits2). None of the new entrants have, thus, succeeded in

the established automakers. incumbents have gained the most from the new surge

for electric vehicles, beginning in the 1990s. The

1218 A. Bergek et al. / Research Policy 42 (2013) 1210 1224

Table 3Price and convenience a comparison of six EVs 19972011.

19972003 20092011

Toyota RAV4EV GM EV1 Ford EV MIEV Leaf Renault Fluence ZE

Price (USD, excl. subsidies) 42,000 40,000 50,000 Range (miles) 80120 100 75 Charging time (h), std outlet 5 8

Sources: Prices: trade journals. Actual prices are inuenced by subsidies, such as USD 7500 per EV in feRange gures are approximate, and depend on driver behaviour, trafc conditions and the use/non-use

a Renault w are le

revival of tincumbentsafter failingNissan/Rencles when itNissan prodMarch 2011cal year (Gra part of a companys lturing systesame lines. to 5000 in duction ramother indusvehicles as wEV (2012), D(2012), HonVW E-Up mbeen limiteLeaf cars wesame period

Disruptihave arguedcaused by thstrategies tin order tonary users.makes EVs range. Howother strateas complemfor long drishare. Teslaprice tags otheir cars hare far fromcles. An altewhich can tproducts. Sof so-calledand China. LSEV produaccommodaTen years laelectric vehfrugal low-countryside

6 The Natioventure by deallowed LSEVsand Kimble, 20

on inequipationus o

areationstion,mbleomp

to 1 pothis mly deccesst ar

to e or ction

beenc veh; al

it waar mch anot nt reienc

rers thatess. Bthem

The atitionile er criing -elecere ped w

warmerlin. in 2ecam

ruaryho is part of the Nissan-Renault Group pursues a different strategy, where batteries

he EV market in the late 2000s was seen by Japanese as a possibility to build a new competitive platform

to compete with Toyota in the hybrid race (see below).ault emerged as a leading proponent of electrical vehi-

launched the compact Leaf for the mass market in 2010.uced 10,000 Leaf cars in its rst scal year ending in, and planned to produce 50,000 units in the next s-eimel, 2011). At Mitsubishi, electrical vehicles becamenew growth strategy, which tried to make use of theong history in battery research and its exible manufac-m, where conventional and electric cars are built on theMitsubishi produced 1500 EVs in scal 2009, increasingscal 2010 and similar to Nissan, planned major pro-p-ups in 20112012 (Orihashi et al., 2010). A range oftry incumbents started or announced sales of electricell (Guilford, 2011): Audi e-tron R8 (2012), BMW Miniaimlers Smart EV (on sales in 2011), Ford Focus electricda Fit EV (2012), Toyotas new RAV4 electric (2012) andinicar (2013). As noted above, the market success hasd so far, however. In the rst half of 2012, 3148 Nissanre shipped to U.S. consumers (compared to 3875 in the

in 2011), and only 333 Mitsubishi EV subcompacts.ve innovation theorists (e.g. Christensen, 1997/2003)

that the competitive problems of electric vehicles aree cramming tendency of established makers, i.e. their

o build EVs as similar to conventional cars as possible minimize differences in driving experience for ordi-

According to this argument, the cramming strategycomplicated and costly, and also reduces their drivingever, some of the new entrants, e.g. Think, have triedgies, but without much success. Think positioned its carentary to conventional cars, which reduced the needving ranges, but failed to gain any substantial market

and Fisker have focused high-end market niches withn their rst models exceeding USD 100,000. Althoughave a longer driving range than most other EVs, they

entering main markets with cost competitive vehi-rnative, disruptive strategy is to develop frugal EVsarget low-end or underserved markets with affordableuch markets are actually being targeted by producers

LSEVs, low speed electric vehicles, in both the U.S.Global Electric Motorcars (GEM), the leading Americancer, produced its rst vehicle in 1998: a 48-volt car thatted two passengers and had a top speed of 20 mph.6

ter, GEM reported an accumulated production of 40,000icles (Automotive News, 2008). In China, even morespeed electric vehicles have gained popularity in some

based safety nese n(and thties inexempinnovaand Kiket is csoaring

Theenter tis highmay suand costituteenter all funso haselectriof GEMyears, main cany sucle canstringeconvenufactureasonthing lnot in tage.

4.2.2. compe

Whanothesurprishybridtion wcombinglobal car (Spthe U.SPrius bin Feb locations. Similar to the GEM-species, these LSEVs are

nal Highway Trafc Safety Administration (NHTSA) supported thissignating a new class of motor vehicle, the low-speed vehicle which

to be driven on public roads if they met certain safety criteria (Wang12).

had reacheCongress, 2

Honda wtheir Insighmodels Civshare of thcompetitorthe eld d27,990 32,780 c. 29,000a

100 100 10067 8 68

deral U.S. subsidies (2011). Range and charging time: company data. of climate systems.ased separately.

expensive lead-acid batteries, but in the Chinese casement is omitted entirely to reduce the total cost. Chi-al authorities do not recognize them as a road vehiclewners do not pay any taxes), but provincial authori-s where LSEVs are produced have created temporary. Some observers have presented them as a disruptive

referring to their low cost, utility and simplicity (Wang, 2012). In China as in the US, however, the auto mar-letely dominated by conventional vehicles, with sales4.4 million cars in 2011.ential of the makers of low-speed electric vehicles toainstream market or challenge incumbent automakersbatable. Low cost-products with limited performancesfully occupy specic niches, but reduced performancee not enough to make a technology a competitive sub-xisting technologies. The critical test is its ability toerode main markets and successfully upgrade over-ality in competition with established products. To do

beyond the technical conguration of the low-speedicles. This is indirectly conrmed by the experiencethough it was owned by Chrysler for more than tens never used as a basis for any low-end entry into thearket and nor has its new owner Polestar announcedmbitions. The explanation seems to be that a vehi-enter main car markets without satisfying a menu ofquirements, including speed, efciency, driving range,e, safety, styling, utility, etc. In other words, if car man-pursue cramming strategies, they do it for the simple

consumers and regulators are not satised with any-etter environmental performance and/or lower cost doselves provide any breakthrough competitive advan-

dvent of hybrid-electric vehicles: technological driven by incumbentslectrical vehicles failed to disrupt the car industry,tical discontinuity did occur. This was triggered by therelease in late 1997 of Toyotas rst mass-producedtric car, Prius I. Initial sales of this sustaining innova-oor, but the launch of successively rened new models,ith soaring fuel prices and a rising public awareness ofing made Prius a symbol of the modern environmental

g and Gordon, 2009). Toyota sold 88,000 hybrid cars in004 and as many as 350,000 three years later. In Japan,e the best-selling car model in 2009 and 2010, and

2011, Toyota announced its cumulative hybrid sales

d more than 3 million vehicles worldwide (Green Car011).as the rst to respond to Toyotas initiative, launchingt model in 1999 and hybrid versions of their standardic and Accord a few years later, gaining a signicante U.S. market at that time (see Fig. 2a). The Americans reacted much more slowly and their early R&D inid not generate any new products. However, when

A. Bergek et al. / Research Policy 42 (2013) 1210 1224 1219

%1%

1%0%

0%

2011a b2000

Fig. 2. Hybrid nd 201the reader is r

Source: USDOE

hybrids begtheir technhybrid cars2004. GM wmodels in 2entered theAmerican mToyota. At tportfolio bluxury Lexuand the cosmanufactur(2011) andVolkswagencatch up bnext roundextended bgrid. Its Chresponded plug-in veras against U

By the emajor car mthe U.S. mar9 different mhad achievnicant salWhereas samonths of 2cars (Rechthybrids, ToFord and th(see Fig. 2bPrius was mJapan (Colia

The salesales numbous sectionthe dominaindustry, inestimates otric vehicledominant velectrics ma

et arform

raneve

ectriectris bas werice/

theacturlties

but ahen

ives r hyb

o selt Eng

stor invoe sig

discit a. Thetionsake i11%

10%

7%

5

59%

41%

car market shares for different car manufacturers on the U.S. market: year 2000 aeferred to the web version of the article.)

(2012).

an to gain market shares, GM and Ford ramped upological activities and started to present their own. Ford was rst, introducing their Escape hybrid inas slower in response, launching their rst two hybrid007 and four additional models in 2008. Chrysler also

market in 2008. However the sales of the big threeanufacturers were minuscule compared to those ofhe same time, Toyota continued to expand its hybridy gradually migrating the hybrid technology to thes line, as well as the mass market Camry (in 2006)t-sensitive subcompact Yaris (Prius C) (in 2012). Otherers also entered, including Nissan (2007), Hyundai

the European manufacturers BMW, Mercedes and (20102011). After a late awakening, GM sought to

y accelerated efforts to develop technologies for the of hybrid evolution, plug-in hybrids, i.e. cars withattery capacity that can be charged from the electricevrolet Volt was launched in 2010. Toyota, however,to this leapfrogging attempt a year later by launching asion of Prius, at a considerably lower price (USD 32,750SD 44,600 for Volt before tax credits and subsidies).

nd of 2011, fourteen years after the launch of Prius I, allanufacturers had a hybrid on their product program. Inket over 30 different hybrid car models were offered byanufacturers. However, thus far no one except Toyota

ed a cost and performance level resulting in any sig-

(Weisskey pedriving

Howtery eland elsuch atems afrom pket andmanufdifcutrains,cost. Wexecutlevel fosible tProduc

Thelengescar thmulti-opmenlengesrestriccost, mes, giving Toyota an overwhelming market leadership.les of Chevrolet Volt reached 16,348 cars in the rst nine012, Toyota Prius sold eleven times more, or 183,340

in, 2012). Including its full range of Toyota and Lexusyota owned two thirds of the market, forcing Honda,e other 6 manufacturers to ght for the remaining third), in spite of the fact that the 2011 production of Toyotauch reduced because of the March 2011 earthquake ins and Rechtin, 2012).s gures for hybrid cars could be compared with theers for pure electric vehicles presented in the previ-. It is evident that hybrid-electric power-trains remainnt alternative power-train in the global automotive

spite of the hype surrounding EVs. According to recentf learning rates and price trends of hybrids vs. pure elec-s, HEVs (including plug-in hybrids) could become theehicle technology in the next two decades, while purey require extensive policy support until the late 2020s

therefore cture and nein close patest how easystem (Cotion of knowhybrid powture. Manydeveloped fcess innovaindustry. Stric motorsof electric vbattery syscompanies project ToyPanasonic. this joint v65%

Toyota

Honda

Ford

Hyundai

GM

Nissan

VW

BMW

Mercedes

1. (For interpretation of the references to color in this gure legend,

l., 2012). The main reason for this is that hybrids offerance advantages in comparison with EVs in terms of

ge, cost, ease of charging, etc.r, hybrids are considerably more complex than bat-c cars. To combine the virtues of combustion enginesc drives, innovation is required in new componentstteries, power electronics and electronic control sys-ll as in the power-trains overall architecture. Judgingperformance characteristics of hybrids in the U.S. mar-ir market shares, this is a challenge few established carers have mastered. The competitors of Toyota have hadnot only to design compact and efcient hybrid powerlso to build requisite industrial capabilities and reduce

GM introduced its hybrid models, for example, R&Dlamented the difculty of reaching a competitive costrid car production, claiming that it was virtually impos-

l hybrids without signicant losses (Vice President forineering, GM Power train Europe, May 2007).y of Toyotas hybrid development illustrates the chal-lved in developing a high-performing hybrid-electricnicantly increased complexity, the extended need for

plinarity and the necessity of parallel testing and devel-s well as the strategies employed to handle these chal-

basic rationale for Toyotas strategy is that system level, such as comfort, safety, serviceability, reliability andt necessary to optimize all components together. Toyota

hose an integrated effort, where both a new architec-w specialized components were developed in-house, orrtnership, in a process where the Prius teams had toch component integrated into the constantly changinghen et al., 2009, p. 70). This strategy required accumula-ledge of the new component technologies involved in

er-trains, both regarding their properties and manufac- components of the hybrid power-train were originallyor low-volume applications and not exposed to the pro-tion required to reach competitive prices in the car

o Toyota became a signicant mass producer of elec- and engaged in intimate collaboration with its supplierehicle batteries, Panasonic, to develop hybrid-electrictems to be integrated in the hybrid power-train. Theformed a joint venture and during the developmentota located a signicant number of vehicle engineers atAt a later stage, Toyota acquired a controlling interest inenture. Toyotas sourcing strategy indicates that when

1220 A. Bergek et al. / Research Policy 42 (2013) 1210 1224

the requisite knowledge base changes, developing new technicalinsights at the R&D level does not sufce; the company also needs toaccess knowledge related to manufacturing and its cost structures,either by joint ventures or in-house component production.

5. Analysis

5.1. Why at

There artially disrufailure to sakets, (2) theand (3) the infrastructu

A primeto capture undershooting customvehicles), asufced to range and ciency (micstate of affaHowever, invations deshas demonsformance ato hover arin CCGT-apyears. And iand efcienresulted in overall perfucts makingthe specic

A secondindustries hattributes, athe gas turdescribed ithat existinin relation disk drive icreating a scost and conhas never beter (electrin the best remains faron their efa huge maris the compdisruptive imeasured aand Christeagainst a coing experiereliability, suct has to imnew ones) wvery difcuone or two formance opotentially argued that

driving range or maximum speed and might be willing to substitutereduced emissions for speed, they are not willing to accept lowerperformance in attributes such as safety and price.

A third reasons for failure is that in both cases the dominantlogiehangs. A

stan for puteds at Suchour t

trajedednge th

hy in

howan sstudaintys enIn theasinSeveroyotthoubothges es. Inder

a dying g coms innof ined alogyd toecenvery ed ae nelogyd notpid pwas logy

indtrajelogiens frtechn

al comeep y seategraoduc

e cas

the as ca: explaining the competitive outcomes

tackers failed

e three basic reasons why the attackers and their poten-ptive innovations failed in both industries: (1) theirtisfy established performance demands in main mar-

absence of overshooting in the relevant main marketsembeddedness of the studied industries in establishedres and institutional frameworks.

reason why microturbines and electric vehicles failedany signicant share of mainstream markets is theiring in key performance attributes valued by exist-ers. Advantages in zero tailpipe emissions (electricnd simplicity and exibility (microturbines) have notcompensate for their inferior performance in terms ofexibility (electric vehicles) and thermal electrical ef-roturbines). As described in Section 2, this is the normalirs for many innovations when they are rst introduced.

contrast to examples of successful discontinuous inno-cribed in the literature, neither of these technologiestrated sufcient improvements in the established per-ttributes. In power generation, microturbines continueound half of the electrical efciency of large turbinesplications, which are continuously improving over then the car industry, increasing stringencies in emissionscy regulation, both in the U.S., Japan and the EU, havea new competition among incumbents to upgrade theormance, including efciency, of their established prod-

them more difcult to compete against and reducing advantages of electrical vehicles.

and crucial aspect is that competition in the studiedas not changed from established to new performances suggested in the disruptive innovation framework. Inbine case, this is related to a lack of overshooting. Asn Section 2, the disruptive innovation model assumesg suppliers tend to over-serve major market segmentsto key performance attributes (e.g. storage capacity inndustry). This is described as a general phenomenon,pace for new entrants competing on parameters such asvenience. However, in thermal power generation, thereeen any over-shooting in the key performance param-ical efciency) and with an efciency rate of 5060%combined-cycle applications, the gas turbine industry

from saturation. New entrants are always measuredciency performance and here microturbines fail withgin. In the car industry, the problem for new entrantslexity of the established performance trajectory. In thennovation framework, performance is assumed to belong a single critical performance trajectory (cf. Bowernsen, 1995). In contrast, car performance is evaluatedmplex set of non-substitutable criteria, including driv-nce, styling, fuel efciency, convenience, functionality,afety as well as price. To succeed, an innovative prod-prove some core performance attribute (or introduceithout sacricing the other dimensions. This makes it

lt for new rms to enter based on superiority only incharacteristics they have to offer good enough per-n all the other dimensions as well. This also offsets thedisruptive effects of any overshooting: even if it can be

many car buyers today are over served with respect to

technoicant cpolicierequirementsdistribstationusers. behavimanceembedchallen

5.2. W

As snot meunder uncertturbinebents. an incrrms. with Tfar, wi

In challenpetencand At intodestroexistintinuouscope remaintechnoreferremore r[i]n eimprovafter thtechnogies diat a rabents technothe carferent technosolutioCCGT tencesto staytheir dneousland inand pr

5.3. Th

As theories are parts of large socio-technical systems. Any signif-e therefore requires major changes in institutions andwider diffusion of microturbines, for example, woulddardized interfaces to the power grid and legal require-ower utilities to buy power delivered by thousands of

generators. Electric vehicles require access to chargingconvenient locations and new driving patterns among

changes in institutional arrangements and consumerend to be even harder to achieve than changes in perfor-ctories or rm competences. Hence the infrastructuraless further aggravates the difculty for new entrants toe industry incumbents.

cumbents struggled

n in Section 4, the absence of successful attackers didtability in the selected industries. During the periodsy, rms in both sectors faced intensive technological

and dynamics. The competition in large, advanced gasded with the exit of two of the four leading incum-e car industry, the competition is still on-going, withg diversity of technological offerings from incumbental incumbents have made serious attempts to catch upas overwhelming lead in hybrid-electric vehicles sot any success. How can these patterns be explained?

sectors, the innovation efforts demonstrate theof building and applying requisite technological com-n contrast to the framework suggested by Tushmanson (1986), however, the studied industries do notichotomy of competence-enhancing vs. competence-innovation. Incumbents struggled even though theirpetences were not rendered obsolete by the discon-

ovations. A rst reason for this was that the pace andnovation in established technologies accelerated and

core factor in both industries. The improvement of old as a response to technological discontinuities is often

as the sailing-ship effect (Gilllan, 1935). Discussingt cases, Cooper and Schendel (1976, p. 67) argued thatindustry studied, the old technology continued to bend reached its highest stage of technical developmentw technology was introduced. But eventually, the new

carried the day. However, in our cases the old technolo- leave the stage. Instead they continued to be improvedace. A second reason for the struggle among incum-the challenge to combine their evolving established

in complex congurations with new technologies. Inustry, hybrid electric vehicles combined two very dif-ctories: electric power-trains and combustion engines. In the gas turbine industry, materials, methods andom the aircraft engine industry were incorporated intoology. Thus, in both industry cases existing compe-though still important and useful were insufcientpetitive; established rms had to maintain and evolveknowledge in established technologies, while simulta-rching and acquiring new complementary knowledge,ting this with existing knowledge in order to develope optimized product solutions.

e for creative accumulation: a triple challenge

nalysis above indicates, attackers advantage-orientednnot explain our industry cases. They correctly

A. Bergek et al. / Research Policy 42 (2013) 1210 1224 1221

predicted that new entrants would appear in the two potentiallycompetence-destroying and disruptive innovation cases (EVs andmicroturbines) and that there would be few (if any) new entrantsin the competence-enhancing and sustaining innovation cases(HEVs and CCGT). However, in contrast to the predictions of thesetheories the new entrants were not able to seriously challengeindustry incumbents and the main difculty for incumbents wasneither to rebusiness mneed for anWe proposeas such a cand creativ

In previobeen used cesses in soknowledge than replacever, wherincrementa1997; Hopkcreative acccreativity aoutside of150), whereon these ex

The creimprovemenologies oron changesboth (cf. Himplies thatrather thandescribed a(1986) sensknowledge this specicexpandingsystem regi

[E]xistipumps developbined, abthe shapspeeds, characte2006, p.

In creatiseen as a bathan a sourinnovation companies an unteste(Tushman aing and comways to devperformanc

Moreoveand the neimplies thaical trajecto

7 Gatignon acquisition.

trajectories simultaneously, which is a very different challengefrom the established understanding of competence-enhancinginnovations. This challenge is even more pronounced in complexproduct industries. Firms in such industries already have to managethe critical and broad Zirpoli, 200

e accxisti

illuste accuctiostry

sum challg tecchnoowle

To m andknow

exteartnlogiend s

clus

stingpmes conolot thfor cnedte (Ts, onged he rriste

Thetencrmsally

s paries, th thtors. t- anechnbreakdustestrnsened bevelavitto th

cap chatinuiulatit of place their existing competence base nor to create newodels. The two industry studies thus demonstrate the

innovation concept beyond the attackers advantage. an elaboration of creative accumulation (Pavitt, 1986)

oncept to understand processes of both accumulationity in complex products industries.us literature, the concept of creative accumulation has

to highlight the cumulative nature of knowledge pro--called mature industries, where the creation of newbuilds on existing knowledge and expertise, rathering it (Granstrand et al., 1997; Hobday, 1990). How-eas this literature presents cumulative innovation asl, step-by-step enhancements (cf., e.g. Granstrand et al.,ins et al., 2007; Kash and Rycroft, 2002), our notion ofumulation stresses the inherent tension between thend accumulation aspects: creativity implies responses

the range of existing practice (Schumpeter, 1947, p.as accumulation implies knowledge creation buildingisting practices.ativity aspect is manifested through substantialnts in cost, performance or quality over previous tech-

the introduction of new performance attributes, based in component technologies, product architectures orenderson and Clark, 1990). The accumulation aspect

this type of innovation builds on previous competence, destroys it. However, creative accumulation cannot bes competence-enhancing in the Tushman and Andersone, since it forces incumbent to go outside their existingbase and search for new competences. Geels (2006) calls

type of cumulative innovation process competence-,7 using it to explain why incumbents in the sewageme were able to adjust and reorient themselves:

ng knowledge about bricks, pipes, water ows andremained relevant. Additional knowledge had to beed about how different components were best com-out costs and benets of combined or separate sewers,e of sewer pipes, material choice, sewer slopes, owand soil conditions. But this knowledge had an add-onr, and could be learned by incumbent actors. (Geels,

1078)

ve accumulation, the existing knowledge base can besis for expanding the available choices of rms, ratherce of inertia (cf. King and Tucci, 2002). As seen above,characterized by creative accumulation also exposesto substantial challenges both in terms of masteringd and incompletely understood product or processnd Anderson, 1986, p. 444) and of acquiring, integrat-bining new types of knowledge with old ones in novelelop products with substantially improved or changede.r, there is no Archimedean xed point: both the oldw domains evolve and interact with each other. Thist rms cannot choose to follow only one technolog-ry but have to pursue several internal and external

et al. (2002) suggest the related concept of new competence

Creativuse of e

As creativintrodof indu2009).

To triple existinnew teing kn2011).searchhouse 1990),close ptechnonents a

6. Con

Exidevelocultieto techsuggesbasis threateobsolenationchallenalter tket (Ch2006).compebents drastictry.

Thiindusting boof secmarkewith tsingle such inative dChristedescrib

By dtion (Pboth tpete into thedisconaccumopmenbalance between deep component-related knowledgesystems-related architectural knowledge (Becker and3; Brusoni et al., 2001; Funk, 2010; Takeishi, 2002).umulation implies an added challenge of balancing theng knowledge and new knowledge on both these levels.rated by recent studies in telecommunications sector,umulation is not merely related to technology, as then of new technologies may prompt recongurations

standard setting processes and value networks (Funk,

up, creative accumulation requires rms to handle aenge of simultaneously (a) ne-tuning and evolvinghnologies at a rapid pace, (b) acquiring and developinglogies and resources and (c) integrating novel and exist-dge into superior products and solutions (Bergek et al.,eet this challenge, they have to engage in technology

extended experimentation as a basis for building in-ledge and absorptive capabilities (Cohen and Levinthal,nd their search processes to external sources throughering with new suppliers and specialists, and develops and systems that integrate new and existing compo-ubsystems into functioning products.

ions

theories on industrial innovation and technologicalnt have presented important explanations of the dif-nfronting leading rms as they attempt to respondgical discontinuities. Competence-based explanationsat technological competencies constitute the primeompetition and that the position of incumbents is

when new technologies make existing competenceushman and Anderson, 1986). Market-based expla-

the other hand, suggest that incumbent rms arewhen new technologies offer novel attributes whichank-ordering of performance attributes in the mar-nsen and Rosenbloom, 1995; Christensen, 1997/2003,se differences notwithstanding, both market- ande-based explanations argue that the position of incum-

is challenged only when technological discontinuitieschange the prime basis for competition in an indus-

per has analysed discontinuous innovations in twogas turbines and automotive power-trains, represent-e scale-intensive and expensive capital goods typesThe analysis shows the limitations of the existingd competence-based explanations when confrontedological discontinuities in these industries, where nothrough is likely to change the basis for competition. Inries, technological discontinuities seldom lead to cre-uction, neither disruptive innovation as proposed by

and others, nor competence-destroying innovation asy Tushman and Anderson.oping and extending the concept of creative accumula-, 1986), we have advanced an alternative explanatione challenges faced by new entrants trying to com-ital intensive and complex products industries, andllenges threatening incumbents when technologicalties intensify competition in these sectors. Creativeon highlights the importance of accelerating the devel-established technologies, acquiring new technologies

1222 A. Bergek et al. / Research Policy 42 (2013) 1210 1224

and integrating these technologies and attributes with existingones. Thus rms need to simultaneously develop their exist-ing knowledge and source and integrate new competencies. Thisperspective explains why potentially competence-destroying ordisruptive innovations failed to challenge the industry incumbentsin the gas turbine and automotive industries, and why only someincumbents, in spite of the failure of new entrants, managed tosurvive and improve their competitive positions.

Two main features of creative accumulation contribute to theunderstandelectrical vaccumulatiovant attribumarkets. Aknowledge develop andSecond, theenable thementrants cacar industreach othermoving tarence enableto competeand Krop, 2

The neednew knowleestablishedattackers inIn the case 1990s in spnew customnew entranspecialized or fuel cellsintegrating establishedmajor mark

Creativeand creativincumbentsand power-partially bucreative aspket positionWe