Franco, April, MB Sarkar, Raj Echambadi and Rajshree Agarwal, Swift and Smart?

19 Pages·2009·0.2 MB·English

MANAGEMENT SCIENCE
informs
®
Vol.55,No.11,November2009,pp.1842–1860 doi10.1287/mnsc.1090.1061
issn0025-1909(cid:1)eissn1526-5501(cid:1)09(cid:1)5511(cid:1)1842 ©2009INFORMS
Swift and Smart: The Moderating Effects of
Technological Capabilities on the Market
Pioneering–Firm Survival Relationship
April M. Franco
RotmanSchoolofManagement,UniversityofToronto,Toronto,OntarioM5S3E6,Canada,
[email protected]
MB Sarkar
FoxSchoolofBusinessandManagement,TempleUniversity,Philadelphia,Pennsylvania19122,
[email protected]
Rajshree Agarwal, Raj Echambadi
CollegeofBusiness,UniversityofIllinoisatUrbana-Champaign,Champaign,Illinois61820
{[email protected],[email protected]}
W
e extend the concept of first-mover advantage to the context of high-technology industries with mul-
tiple product generations, and propose that the notion of first-mover advantage needs to be viewed not
only through a dynamic lens, but also in conjunction with technological capability. Our main finding is that
first-mover advantages are best understood in tandem with the firm’s technological capabilities; early entry is
beneficialonlyforpioneersthataretechnicallystrong.However,pioneersthatarelowontechnologicalcapabil-
itiessufferfrompoorsurvivalratesvis-à-vismarketrespondersornonentrantsintonewproductgenerations.
Key words: dynamic capabilities; first-mover advantage; strategy; industry evolution; entrepreneurship;
technologystrategy
History: Received July 4, 2006; accepted March 20, 2009, by Pankaj Ghemawat, business strategy. Published
onlineinArticlesinAdvanceSeptember11,2009.
Introduction research that indicates that new product generations
In spite of 839 publications on first-mover advantage withinanindustrycanleveltheplayingfieldbetween
(FMA) in peer-reviewed journals, its existence has both incumbents and entrants, as well as between
neither been conclusively proved nor refuted (Suarez pioneers and late movers. Thus, legacy-based advan-
and Lanzolla 2007). Notwithstanding methodologi- tages could be patently atheoretical in such situa-
cal advances that have alleviated concerns related tions. On the micro side, firm-level capabilities may
to survivor bias, sample selection issues, and defi- complement market-pioneering efforts, namely, sci-
nitional confounds in dependent variables (Boulding entific or technological know-how embedded in a
and Christen 2003, Lieberman and Montgomery firm’s product. This is another possibly serious flaw
1998,VanderWerfandMahon1997),thisinconclusive- becausetherealizationofpotentialfirst-moveradvan-
ness of FMA prompted Finkelstein (2002, p. 39) to tagesfrommarketinitiativesislikelytobecontingent
skepticallycommentthatthe“holygrailoffirstmover on a firm’s technological competence. This important
advantage is as elusive as it is exaggerated.” fungible resource (Granstrand et al. 1997, Danneels
We contend that the problem lies in two theoretical 2002) determines how competitive a firm’s products
misspecifications of models relating entry timing to are along accepted scientific parameters of perfor-
organizational outcomes: one on the “macro” indus- mance. Complementary resources play a crucial role
try level, and the other on the “micro” firm level. in entrepreneurial rents (Stieglitz and Heine 2007),
BothhavelargelybeenignoredintheFMAliterature. and heterogeneity in such resources across firms can
On the macro side, ignoring the evolutionary dynam- explain variance in firm performance (Robinson et al.
ics of the industry has led to a static, legacy-based 1992, Rosenbloom and Cusumano 1987, Teece et al.
view of FMA, which is predicated on the idea that a 1997,Klepper2002).Therefore,studyinghowmarket-
one-time pioneering act during the incipient stage of creating capabilities affect profits, market share, or
an industry could lead to sustained advantages over survival without considering the impact of comple-
time. This premise is, however, problematic given mentary capabilities is inherently problematic.
1842
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
ManagementScience55(11),pp.1842–1860,©2009INFORMS 1843
Both the micro and macro gaps have a com- Regarding the lacuna at the “micro” firm level,
mon feature; because of the evolutionary nature of we propose that technological capabilities, concep-
the technology and product market, it is important tualized in a dynamic setting, complement the rela-
to consider the relevant know-how, i.e., market- tionship between entry timing and firm performance.
pioneering and technological capabilities, in an evolv- Our focus on this complementary asset is motivated
ing context. This relates to recent strategic thinking by research that indicates that such assets not only
on dynamic capabilities (Eisenhardt and Martin 2000, foster successful entry into new product markets
Helfat2007)thatemphasizesachievingnewandinno- (Wernerfelt 1984), but also explain firm survival over
vative forms of competitive advantage through exer- the industry life cycle (Helfat and Lieberman 2002,
cisingstrategicchoicesthatguidecapabilityevolution Klepper and Simons 2000). Recent strategy research
(Lee 2008). At its essence, the idea of dynamic has also found firm performance to be a function
capabilities concerns change, and the “capacity of of interdependent capabilities (Agarwal et al. 2004),
a firm to purposefully create, extend, or modify thereby emphasizing the need to consider portfolios
its resource base” (Helfat 2007, p. 4). In this view, of resources that are complementary and together
firms continually realign and redesign their resources
enable value creation and appropriation (Steiglitz
through a process of search and selection to “achieve
and Heine 2007, Moran and Ghoshal 1999, Teece
new resource configurations as markets emerge”
1986). Whereas firms need technological capabili-
(Eisenhardt and Martin 2000, p. 1107). Continual
ties to engage in scientific inventions (Bierly and
efforts at strategic renewal—through creating new
Chakrabarti 1996, Cohen and Levinthal 1990), they
market spaces or responding to existing ones—while
also need marketing know-how to appropriate the
simultaneously sustaining its technological and prod-
potential economic rents embodied in their techno-
uct market position is critical to a firm’s survival
logical breakthroughs (Dierickx and Cool 1989, Teece
in rapidly changing environments (Burgelman 1991,
1986). Their complementary nature creates a valuable
FloydandLane2000,GansandStern2003,Huffetal.
synergy (Day 1994) that maximizes the value of mar-
1992). We draw insights from the strategy literature
ket opportunities to a firm (Park and Zaltman 1987,
to address this important “static” problem in entry
Walker and Ruekert 1987) and inhibits competitive
timing research where the dynamics of evolution-
imitation due to the inherent difficulty of develop-
ary change, both on market- and technology-related
ing the two capabilities simultaneously (Grant 1991,
domains of firm capabilities, have been ignored.
LippmanandRumelt1982,ReedandDeFillippi1990).
To address the “macro” concern, we consider an
From this perspective, it is imperative for FMA stud-
evolutionary context where technological discontinu-
ies to consider technological know-how as a critical
ities create a set of upwardly shifting S-curves in
resource that is complementary to market creation.
an industry (Foster 1986). Herein, we consider firms’
In an empirical setting marked by rapid changes
abilities to create new markets through pioneering
and the emergence of new technological subfields
entry into emerging product generations. Research
andproductmarkets—thediskdriveindustryduring
has documented how “technological subfields” in the
the 1977–1997 period—we examine the outcomes of
disk drive and medical diagnostic industries created
firms’ capabilities in market creating and responding.
new market segments that had far-reaching conse-
We further investigate the interaction effects of tech-
quences on both industry structure and composition
nological capability and timing of entry. Our paper
(Mitchell 1989, Christensen 1997a). Such new product
thus extends the concept of first-mover advantage to
generations presented both incumbents and potential
industry contexts with multiple product generations
entrants with multiple strategic opportunities: be a
market pioneer, respond through late entry, or not to examine two specific relationships: first, the main
enter. Importantly, these punctuations in the industry effects of market-pioneering and market-responding
life cycle may serve as potential levelers in a Schum- behaviors on the rate of survival, and second, the
peterian (Schumpeter 1942) sense because they may role of a firm’s technology position as a moderator in
disturb the status quo, negating legacy advantages the market pioneering/responding–survival relation-
and offering the opportunity to redefine the competi- ships. We find that incumbents from prior product
tivescenario.ThisleadstoadynamictheoryofFMA: generationsareequallylikelytopioneernewmarkets
Early entry is not a one-time act that creates a legacy- as new entrants in the industry. Moreover, firms that
based advantage, but one that needs to be examined in fail to keep up with the evolving industry, in terms
a dynamic context characterized by the emergence of of either new markets or technological development,
multiple product generations as a result of industry are more likely to exit the industry. Specifically, we
evolution. In other words, we need to consider entry find that when timing of entry is considered in isola-
timing through the lens of a firm’s dynamic capability tionofothercapabilities,marketrespondingprovides
to pioneer or respond to new market opportunities. more benefits than either market pioneering or not
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
1844 ManagementScience55(11),pp.1842–1860,©2009INFORMS
responding to emerging market opportunities. How- removesmanypath-dependentadvantagesassociated
ever, when considered in conjunction with the com- with early entry that may have accrued to the mar-
plementary dynamic capability in the technological ket pioneers of the prior product generation. How-
realm, our results show a more nuanced relationship. ever, existing literature has tended to emphasize a
Pioneering elevates survival rates for firms that sus- static, legacy-based advantage associated with entry
taintechnologyleadership,andthisgroupisthemost timing, and has overlooked the evolutionary forces
advantaged relative to others over time. Conversely, that reshape industries as a result of technological
when pioneering efforts are not matched with efforts inventions and market innovations. In this context,
at keeping up with the changing technological fron- relatingcompetitiveadvantagetobeingamarketpio-
tierovertimeandfirmsfallbelowacertainthreshold neer during the inception of an industry (or for that
of technological capability, pioneering activities hurt matter a prior product generation) is a potentially
a firm compared to late or even nonentry into later flawed thesis, because such legacy-based advantages
generations. are unlikely to be sustained in high-velocity markets
that are undergoing disruptive changes (Eisenhardt
Theoretical Framework and and Martin 2000, Suarez and Lanzolla 2007).
Investigating what firms do when faced with mul-
Hypotheses
tiple opportunities over time allows us to study the
Market Creation and Response impact of firms’ strategic capabilities to create, or
as Firm Capabilities respond to, such technological and market shifts.
Although industries and products are typically char- Extant entry timing literature has ignored whether
acterizedasevolvingthroughsmoothandpredictable firms are advantaged by being able to morph along
S-curves (Gort and Klepper 1982), such heuristic por- with discontinuous changes in their product markets.
trayals may mask occurrences when emerging new However, this ability to create new markets is part of
product generations cater to entirely new user seg- an important class of organizational capabilities that
ments, and in doing so transform the competitive has been termed “dynamic capabilities” (Teece et al.
landscape of the industry. This is consistent with the 1997, Eisenhardt and Martin 2000), which both con-
description of technological progress as one where strainsandenablesafirm’sabilitytoadapttochanges
long periods of incremental innovation are punctu- in its competitive environment (Teece et al. 1997).
ated by bursts of radical change (Abernathy and Viewing entry timing through the lens of dynamic
Utterback 1978, Tushman and Anderson 1986). The capabilities thus suggests that a relevant metric to
new technical subfields and new product markets considermaynotbewhetherafirmwasamarketpio-
create fresh entrepreneurial opportunities within the neer in first product generation, but whether it pos-
industry. Yet, much variance can underlie such shifts, sesses the ongoing capability to engage in the act of
making them distinct phenomena from the perspec- creating or responding to new markets in changing
tive of both entrants and industry incumbents. In industries.
the case of incumbents, some resources and capa-
bilities that were useful in an earlier market may Market Pioneers, Market Responders,
remain relevant in the new segment of the market, and Survival
whereas others may become obsolete (Christensen Legacy-based FMA (resulting from pioneering the
1993,Mitchell1991).Forexample,withinthediagnos- very first product generation in an industry) may
ticimagingindustry,thefivemajortechnicalsubfields be difficult to sustain in high-velocity environments
that emerged represented very different domains of (Suarez and Lanzolla 2007). Rapid evolution of tech-
scientificknowledge,yetthespecializedmarketassets nologies and markets can annul experience curve
retained their value (Mitchell 1989). However, in data advantages (Lieberman 1989) and lead to obsoles-
storage, even though the innovations drew on sim- cence (Tushman and Anderson 1986). Incumbent
ilar underlying scientific knowledge, the disruptive firms often fail to enter new market niches because
technological innovations often created new product- of routinization and inertia (Miller and Chen 1994).
markets with new, very different user characteris- Experience with a particular set of operating rou-
tics that rendered redundant market assets that were tines can restrict an organization’s ability to pro-
effective in earlier generations (Christensen 1993). duce new products, acquire new resources, and enter
Importantly, new product generations can level the new market niches (Teece et al. 1997, King and
playing field among incumbents as well as between Tucci 2002). However, a firm’s dynamic capability
incumbents and entrants. The emergence of a new to create new markets and pioneer emerging prod-
productgenerationcreatesafreshopportunitytocap- uct generations is likely to offset such debilitating
italizeonpotentialstrategicandeconomicadvantages effects of change on legacy advantages, and thereby
associated with being a market pioneer, even as it enhance its likelihood of survival. Firms that can
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
ManagementScience55(11),pp.1842–1860,©2009INFORMS 1845
buck the trend and strategically transform them- may offset the risks of being first, yet share the gains
selves through periodic creation of new markets ofagrowingmarketsegment.Thiswait-and-seestrat-
exhibit the capability of ambidexterity in being able egyisespeciallybeneficialinindustriesinwhichnew
to take care of the present even as they are creating marketsrenderearliersegmentsobsolete(Christensen
the future (Tushman et al. 1997). Organizations that 1993, Mitchell 1991). The choice is not a trivial one in
possess such entrepreneurial capabilities for strate- dynamic product markets that face rapid commodi-
gic renewal are likely to exhibit enhanced perfor- tization: firms may be inclined to “sit it out” rather
mance (Agarwal and Helfat 2009). than invest in markets that are moving toward price-
Although market pioneers bear high product and based competition. Yet, by not responding, they face
market development costs, and face greater levels the risk of being locked out forever from the market.
of risks emanating from demand and technologi- There is reason to believe that entry into new mar-
cal uncertainties (Lieberman and Montgomery 1988, kets, even if late, is better than not entering at all.
1998), firms that consistently engage in market pio- Christensen (1997a) documents the failure of firms
neeringacrossnewproductgenerationsarealsolikely that paid close attention to their current customer
to benefit from various demand- and supply-side iso- needs but missed opportunities to enter new prod-
lating mechanisms that underlie first-mover advan- uct generations. Conversely, Tegarden et al. (1999)
tage (Rumelt 1987). By repeatedly getting a head report that in the personal computer industry, firms
start on the learning curve, preempting scarce assets improved their chances of survival if they undertook
in strategic factor markets, and increasing buyers’ the strategic decision to embrace the new product
switching costs, firms that exhibit sustained market- generation even if they were late to enter. King and
pioneering capabilities are likely to enjoy strategic Tucci(2002)findthatincumbentsthatbroketheiriner-
and economic benefits due to advantageous posi- tia and entered emergent market segments enjoyed
tions in resource space and creation of entry barri- higher sales, and thus gained value relative to firms
ers (Lieberman and Montgomery 1988, Suarez and that did not respond. Firms that are able to under-
Lanzolla2007).Thesebenefitsarelikelytobestronger take such transformational experiences, even if late
when products can be sharply differentiated from in the game, are able to maintain their adaptability
older generations, and early market entry offers an and reduce their inertia, and thereby enhance their
escape from commoditization and intense price com- survival chances (Katz and Allen 1985, Tushman and
petition. In such situations, reputation effects and Romanelli 1985). Accordingly,
temporary price umbrellas are likely to provide an
Hypothesis 2 (H2). Sustained market-response capa-
important monopoly window to firms that systemat-
bilities increase the likelihood of survival.
ically engage in market pioneering of new product
generations (Adner 2002). Therefore, Technological Know-How: A Complementary
Hypothesis1(H1).Sustainedmarket-pioneeringcapa- Dynamic Capability
bilities increase the likelihood of survival. Thusfar,ourfocushasbeenonhowentrytiminginto
new product generations impacts firm survival. Lit-
Onceanewmarkethasemergedthroughtheefforts
erature suggests that, in high-velocity environments,
of market pioneers (whether a new entrant or an
having a number of distinct capabilities enables firms
incumbentfromearliergenerations),incumbentfirms
to alter their resource bases and adapt to chang-
in prior market generations of the industry must
ing competitive conditions (Eisenhardt and Martin
decide whether or not to respond and move into
2000). Specifically, the notion of dynamic capabil-
the new product generation space.1 Potential respon-
ities spans a diverse range of routines, including
ders thus face a difficult trade-off. Having lost the
those related to technological capabilities, market-
market-pioneering opportunity, should they enter the
ing, product development, and resource acquisition,
new market? Their decision relates to the trade-off
recombination, and integration (Teece et al. 1997).
between minimizing the threat of obsolescence and
Importantly, Teece (1982) called attention to exam-
the need to focus resources on their extant markets.
ining capabilities related to the market (as in our
By entering new markets after they have been estab-
emphasis on market pioneering and responding) and
lished by the market pioneers, market responders
to the underlying technological domain. While not
discounting the importance of other capabilities, we
1We note that one way in which firms may choose to respond to focus our attention in this paper on dynamic techno-
emerging product generations is by “skipping” a certain market logical capabilities as important to firm survival, par-
altogether and entering or initiating the next product generation ticularly because it complements our dynamic timing
(de Figueiredo and Silverman 2007). For the focal product gener-
of entry capabilities.
ation,however,thefirm’sdecisionisthusdefactotonotrespond
Acrucialdeterminantofsurvivalisafirm’spropri-
toanopportunity.Intheempiricalsection,weconductrobustness
checksrelatedtotheincidenceofskippingasaresponse. etary technical/innovative capability (Schoonhoven
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
1846 ManagementScience55(11),pp.1842–1860,©2009INFORMS
et al. 1990). Because technological knowledge is typi- even as organizations greatly improve their technolo-
cally tacit and developed over time, it is a source of gies in absolute terms, they often move backwards
competitive advantage. Technological capabilities can in relative ranks over time because rivals improved
create differentiation advantages for firms and thus even faster. This is further supported by work by Lee
relatepositivelytosurvival,particularlyundercondi- (2008), which points to the “Markovian” property of
tions of intense competition. Furthermore, in process- capabilities,suchthatinitialcapabilitiesmayhavelit-
enablingindustries,suchdiskdriveproduction,rapid tle impact on current capabilities. This highlights the
and smooth introduction of new manufacturing pro- issue of contemporaneous rank. In industries with
cesses is highly important for competitive perfor- rapidlyevolvingtechnologies,lagginginpositionrel-
mance. Technology development is often targeted at ative to the contemporaneous technology frontier not
discovering parameters whose unfamiliarity causes onlysignalslowerquality,butalsoincreasestheprob-
defects in manufacturing. Early specifications of a ability of failure. Thus, the ability to sustain tech-
process technology generally fail; engineering exper- nological leadership, or maintain a contemporaneous
imentation and analysis generate knowledge and superior technological position, based on scientific
thereafter codification of the parameters of the tech- breakthroughs and complex organizational learning,
nology (Hatch and Macher 2002). This sequence enhances firm performance. Therefore,
requires both conscious learning by doing and scien-
Hypothesis 3 (H3). Sustained technological capabili-
tific breakthroughs. For example, in the case of the
ties increase the likelihood of survival.
disk drive industry, technological capabilities were
dependent on capabilities related to miniaturization, Given the nature of the relationship between
error tolerances, and clean rooms. As the underlying market-related and technological capabilities, the
science behind the technology of manufacturing disk potential complementarities between the two deserve
drives evolved, obsolescence could only be averted close attention. A key reason relates to the mixed
throughdynamiccapabilitiesthatensuredcontinuing empirical findings on the main effects of each.
mastery over new sets of technological challenges. Just as scholars have documented the inconclusive-
In industries that experience rapid advancement, ness of FMA research (Suarez and Lanzolla 2007),
technological competition often resembles an arms there is increased skepticism regarding the unbri-
race, where instead of absolute goals, the relative dled optimism surrounding technological innovation
goal of staying ahead of other competitors becomes (Adegbesamn and Ricart 2007). Recent findings show
paramount. In such contexts, technology capability that an emphasis on innovation neither translates
needs to be conceptualized in both dynamic and rel- into strategic advantage, nor improves bottom lines
ative terms such that it is measured in comparison as expected (Linder et al. 2003, Andrew and Sirkin
to other competitors over time, instead of in abso- 2003). Despite the potential of leveraging technolog-
lute,staticterms.Fromthisperspective,thereappears ical inventions across multiple market applications
distinct ways in which technology-related capabili- and domains (Danneels 2002), firms often fail to real-
ties can impact survival. First, environmental selec- izethebenefitsfromtechnologicalresources(Thomke
tion processes will lead to the exit of those that are andKuemmerle2002).Thisfailuretoleveragetechno-
relatively weaker on the parameter of importance. As logical competence (Danneels 2007) and appropriate
noted by Lee (2008), the survivor principle suggests valuelatentinthetechnologicalinventionshaspoten-
that firms that occupy lower positions on technology tially led to conflicting findings regarding technologi-
frontiers will be eliminated from the industry, thus calinnovativenessandfirmperformance(Caponetal.
allowing firms better equipped to satisfy the mar- 1990, Leiponen 2000).
ket’s demands to survive. Because a firm’s position To profit from innovation, therefore, a firm needs
depends not only on its own efforts, but also on the to keep up with both new technologies and new mar-
industry-wide advancement of technology, the ones kets (Griffin and Hauser 1996). The complementarity
that are unable to keep up with the industry-wide of these capabilities creates a synergy that increases
advancement of technology will face exit pressures. a firm’s effectiveness and efficiency. As noted by
Second, the ability of a firm to integrate, build, and Danneels(2007),thecreationofcomplementaryassets
reconfigure its technological competencies to address requires resource allocation and transformation of
rapidly changing environments will determine sur- generic resources into specific resources. Developing
vival.However,whetherfirmsadapttothenewtech- new, complementary capabilities requires a strong
nological realities and merely succeed in improving strategic intent, because it involves a deliberate man-
their technological capabilities from one period to the agerial decision to divert resources away from the
nextisnotimportantperse;whatmattersistheirrel- production of output to the production of compe-
ative position in the technology race in the current tence(Dorrohetal.1994).AsindicatedbyChristensen
period. Barnett and McKendrick (2004) observe that (1997a, b), the “innovator’s dilemma” is predicated
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
ManagementScience55(11),pp.1842–1860,©2009INFORMS 1847
by the tendency for incumbent firms to focus on dynamic technological capabilities and that of pio-
theircurrentconsumers’needsandignoredeveloping neering versus responding. In other words, although
capabilities that may enable them to tap into emerg- we have argued that the ability to maintain tech-
ing, yet nascent segments. nological superiority is of greater help when firms
The inherent difficulty of developing these two possess the ability to pioneer new markets or to
capabilities simultaneously can be a valuable isolat- respondtothem,itisnotclearwhetherthesizeofthe
ing mechanism (Grant 1991, Lippman and Rumelt effects are different between the two market-related
1982, Reed and DeFillippi 1990), separating win- capabilities. This is important because discriminat-
ners from losers. Various organizational pathologies, ing effects would suggest that the interrelationship
which researchers call learning traps (March 1991, between the two sets of dynamic capabilities that
Levinthal and March 1993), localize search in prox- are fundamental to the renewal of firms in dynamic
imate areas (Ahuja and Lampert 2001, Cyert and environments, namely, technological (value creation)
and commercialization (value appropriation), is con-
March 1963), thus often causing firms to fail in
tingent on entry timing into new product genera-
realizing the full value of their capabilities (Moran
tions. At the core, this would contribute to a more
and Ghoshal 1999). Indeed, recent conceptualizations
nuanced understanding of Schumpeter’s (1942) dis-
of firms’ capabilities have emphasized the creation
tinction between invention and innovation, and how
of new sources of economic rents (Eisenhardt and
the payoff from a firm’s ability to engage in both
Martin 2000, Teece et al. 1997). Profiting from inno-
value creation and appropriation concurrently differs
vation requires a blend of research and development
based on the timing of commercialization.
(R&D) capabilities (value creation) and capabilities to
We contend that technology plays a stronger syn-
pursue new markets (value appropriation), particu-
ergistic role for market pioneers than for respon-
larly in environments that may have high imitabil-
ders.Marketpioneerswhoalsomaintainhigherlevels
ity and low dependence on existing complementary
of technological capabilities possess the cospecial-
assets (Gans and Stern 2003). Regardless of whether
ized ability to both create new markets and sustain
firms enter new markets relatively early or late,
them through technological inventions (Gambardella
such efforts at renewal help them shed inertia and
and Torrisi 1998, Teece 1982). Firms that are able to
develop new competencies that maintain their rele-
stay on top of the technology curve while demon-
vance during periods of change (Burgelman 1991).
strating the dynamic capability to create new mar-
Firms that invest in technological capabilities while
kets are likely to be better off than firms that are
also adapting to new market needs are better off
similarly endowed on the technology front but are
than firms constrained by inertia in any one dimen-
characteristically responders, or late movers into new
sion. Such strategic renewal efforts permit firms to
markets. The mechanism behind our prediction is
adapt to changing environments rather than stagnat-
twofold: First, technology leadership operates as an
ingandfallingvictimtoselectionprocessesunleashed
isolating mechanism through which FMA is derived
by industry evolution (Agarwal and Helfat 2009). (Lieberman and Montgomery 1988) by encompassing
Therefore, we expect that the payoff from pioneer- “cost advantages arising from advanced appropria-
ing and responding to new markets will be higher tion of scarce input resources, forestalling bids for
whenafirmisabletosimultaneouslymaintainhigher product characteristic spaces, and economies of scale
levels of technological capabilities. Because the two created from preemptive investment in the plant and
components of the dynamic capabilities required for equipment; and switching costs arise from habit for-
renewal, namely, market creating/responding and mation in buyers or from the installed-base effect in
technological,arelikelytohaveapositivesynergyon thepresenceofnetworkeffects”(SuarezandLanzolla
survival, we hypothesize the following: 2007, p. 379). Second, research has shown that late
movers can overthrow pioneers through a process of
Hypothesis 4 (H4). Technological capabilities posi-
continuous innovation based on learning and tech-
tively moderate the relationship between (a) market-
nological superiority (Shankar et al. 1998). However,
pioneering and (b) market-responding capabilities and the
whenpioneersareabletomaintaintheirtechnological
likelihood of survival.
leadership, such an avenue to losing potential FMA
to responders is preempted. Together, these mecha-
Thus far, we have focused our attention on the
nisms suggest that technologically superior pioneers
synergies between the dual dynamic capabilities—
are likely to have higher rates of survival than simi-
entering new markets and technological capabili-
larly endowed responders. Therefore,
ties, while without distinguishing how the order in
which firms enter may impact this synergy. Our Hypothesis 5 (H5). The synergistic effect of sustained
final hypothesis relates to the intriguing question technologicalcapabilitieswithmarket-pioneeringcapability
of the relative strength of the synergy between is stronger than with market-responding capability.
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
1848 ManagementScience55(11),pp.1842–1860,©2009INFORMS
Data and Methodology
Figure2 ArealDensityofDrivesbyDiameter
Empirical Context 10,000
14"
We tested the hypotheses in our study using data
5.25"
from the rigid disk drive industry over the period 2.5"
1977–1997.Diskdrivesaremagneticinformationstor- ch) 1,000 Highest areal density
n
age devices. In 1973, IBM introduced the first com- sq. i 83".75"
pletely sealed and removable disk drive, the 14-inch B/ 1.8"
M
Winchester, causing the industry to take off shortly y ( 100
thereafter. Figure 1 depicts the standardized values sit
n
e
fwoirththtehenuamnnbuearlovfalfiuremosfaenadchsavleasriainblethdeiviniddeudstbryy, Areal d 10
themaximumvalueobservedforitinthestudy.Both
variables conform to trends documented in industry
1
evolution studies (Agarwal 1998, Gort and Klepper
1976 1980 1984 1988 1992 1996
1982).Thefirst10yearsoftheperiodstudiedshowed
Year
a rapid increase in both number of firms and sales.
The39incumbentsthathadenteredbetween1973and eachsuccessivemarket.Arealdensity(megabytesper
1976 were joined by 153 new entrants after 1977, an inchofstoragecapacity)isacriticaltechnologicalper-
increase that fueled rapid growth in industry sales. formance parameter for disks, and Figure 2 depicts
A shakeout of firms ensued in 1986, and sales contin- the rapid increase in areal density within and across
ued to increase, though at a decreasing rate. disksofsuccessivediameters.Thus,themultiplenew
The smooth trends in number of firms and sales, marketsandtherapidchangeintechnologymakethe
however, mask discontinuous changes within the disk drive industry an ideal empirical context for our
industry. As shown in Figures 1 and 2, architec- study.
tural innovations led to five new disk diameters
Data Sources
during the 20-year period considered in our study.
We collected data from sources that documented
Each diameter innovation created a new market. The
eventsatthetimeofoccurrenceandtrackedinforma-
smaller diameters enabled the creation and growth
tion in the industry for all firms entering and exit-
of new markets including desktop computers, lap-
ing it, to avoid survivor bias. Specifically, we used
topcomputers,personaldigitalassistants,geographic
information compiled from the Disk/Trend Report, a
positioning systems, and other handheld electronic
market research publication that has covered the disk
devices. Importantly, as Christensen (1993, 1997a)
drive industry since 1977 and has been used in sev-
documents, not only did the markets for these new
eral past studies (Agarwal et al. 2004, Christensen
diameters grow faster than the markets for older
1993, King and Tucci 2002, Lerner 1997). These data
diameters,butintimethenewdiametersreplacedthe
were supplemented by information from company
older ones and rendered them obsolete.
news releases, scientific journals, books, and directo-
In addition to new markets, rapid technological
ries (e.g., the Directory of Corporate Affiliations and the
change due to numerous incremental and modular
International Directory of Company Histories). The final
innovations resulted in technology S-curves within
database contains the census of firms in the industry
during 1977–1997 and includes detailed information
on firm and industry characteristics.
Figure1 NumberofFirmsandSalesinDiskDriveIndustry
1.0 Variables
FirmSurvival. Thedependentvariableinourstudy,
es0.8 firm survival, is computed as a dummy variable that
u
d val0.6 takes a value of 1 if a firm survived to the following
e year (with acquisitions being treated as “censored”
z
ardi0.4 observations) and 0 otherwise.
d
n TimingofEntryIntoMarketMeasures:MarketPioneers
a
St0.2 Number of firms and Responders to Markets. Our market-pioneering and
Industry sales market-responding measures are based on new diam-
0
7890123456789012345678 eter introductions, which formed the basis on which
7778888888888999999999
9999999999999999999999
1111111111111111111111 new markets were formed. Our measures of the two
Year
variables are consistent with our dynamic conceptu-
Note. Thedottedverticallinesrepresentthetimingofthesuccessiveintro-
ductionofnewmarketswithintheindustryduringthesampleperiod. alization of pioneering and responding capabilities,
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
ManagementScience55(11),pp.1842–1860,©2009INFORMS 1849
and based on the idea that new product generations incumbentfirmenteredamarketafterthefirstyearof
present firms with multiple market-pioneering and the market’s introduction into the industry, divided
market-responding opportunities. Because the num- by the total number of new market introductions in
ber of such opportunities depends on a firm’s overall the industry since the year of entry for the firm for
time of entry into the industry as well as when the the time until time t. Mathematically, our measures
new markets emerged, our measures of the two vari- take these forms:
ablesvaryoverbothtimeandfirms.Wealsonotethat (cid:1)
t P (cid:2)t
in keeping with our theoretical focus, our timing of PM = (cid:1)k=Ei ik if D >0 and
entry variables relate to firm entry into the new mar- it t D k
kets rather than entry into the industry. k=Ei k k=Ei (1)
(cid:2)t
Regardless of whether they are incumbents in the =0 if D =0(cid:8)
k
industry (i.e., firms that existed in the industry prior
k=Ei
to the focal market introduction) or new entrants, (cid:1)
t R (cid:2)t
firms that entered in the first year of a new market RM = (cid:1)k=Ei ik if D >0 and
are identified as market pioneers in that product gen- it t D k
eration. Our measure of market pioneers as a cohort k=Ei k k=Ei (2)
(cid:2)t
of firms that creates new markets, and subsequently =0 if D =0(cid:8)
k
benefits from it, is consistent with earlier treatment
k=Ei
in the literature (Rosenbloom and Cusumano 1987).
Although only one firm can be considered a true pio- where E is the year of entry of the ith firm, t is the
i
neer on the basis of strict order of entry (Golder and currentyearofoperation,P isadummyvariablethat
ik
Tellis 1993), more than one firm may simultaneously indicates whether the ith firm was a market pioneer
engageinmarket-pioneeringactivitiesandexperience when a new market was introduced (P equals 1 if
ik
similartimelinesinundertakingnewmarketorienta- the firm introduced a drive for a new market within
tionstrategies,suchasscoutingopportunities,assess- the first year of the market’s introduction into the
ing uncertainty, and ramping up for production. industry, and 0 otherwise), R is a dummy variable
ik
Specifically, we adopt the Agarwal et al. (2004) that indicates whether an existing firm entered the
measure of pioneering markets (PM ), defined as the relevant market subsequent to the first year after its
it
number of times a firm introduces a drive in a new introduction(R equals1ifthefirmproducedadrive
ik
market within the first year of the market’s introduc- for the new market for the first time after the first
tion into the industry, divided by the total number of year of the market’s introduction into the industry,
newmarketsintroducedintheindustrysincetheyear and 0 otherwise), and D is a dummy variable that
k
of entry for the firm.2 In robustness checks, we also indicates whether there was a market introduction
examine whether our results are sensitive to a more during that year (D equals 1 if a market was intro-
k
stringent definition of pioneering as introducing the duced, and 0 otherwise). For each firm operating in
drive in the first quarter, rather than the first year. the industry at the time of a new market’s intro-
Between4%and7%ofthetotalentrantsineachmar- duction, the denominator of the variable is increased
ket are classified as pioneers based on the one-year by 1, and the numerator for PM (RM ) increases by
it it
window, and the range is between 2% and 7% if the 1 only if the firm was a market pioneer (follower)
more stringent definition of the first quarter is used. for that diameter. For firms that entered between
Firms change from being entrants into the industry two consecutive market introductions, both variables
to incumbents the first time a new market is intro- take the value of 0 until the year of the next market
duced in the industry after their entry. Our market- introduction. Tables A1–A3 and the related descrip-
respondingvariablecaptureswhetherincumbentfirms tions in the online technical appendix (provided in
that did not pioneer a market subsequently entered it the e-companion)3 provide detailed examples of how
during their life span. Responding to markets (RM ) both market-pioneering and market-responding variables
it
isthusanalogouslydefinedasthenumberoftimesan change over time based on new diameter introduc-
tions and firm decisions regarding entry into new
markets.Thebaselinegroupinouranalysisthuscon-
2Our choice of a one-year window in the operationalization of
sists of firms that have operations only in the market
themeasurewasprimarilydrivenbyconsiderationsofconsistency
thattheyenteredin;i.e.,thesefirmsneverenter,either
withthebulkofreceivedliterature.Amongthestudiesthatusea
“window”ratherthanastrictorderofentryscale,thespanranges early or late, new markets that emerge subsequent to
from the exact date that the first firm entered (e.g., Golder and their entry into the industry.
Tellis1993)toseveralyearsafterthefirstentryyear(e.g.,Agarwal
andGort2001).Themajorityofthestudies,however,useaone-year
window(e.g.,Bayusetal.1997,RobinsonandMin2002,Agarwal 3Anelectroniccompaniontothispaperisavailableaspartoftheon-
etal.2004). lineversionthatcanbefoundathttp://mansci.journal.informs.org/.
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
1850 ManagementScience55(11),pp.1842–1860,©2009INFORMS
Technological Capabilities. As indicated above, a crit- taking a value of 1 if a firm originated outside the
icaltechnologicalperformancemeasureofdiskdrives UnitedStatesand0otherwise.Twomeasuresareused
relates to areal density. Our measure of this variable for a firm’s product development capabilities: number
is based on the average of a firm’s diameter-specific of drives in a firm’s portfolio and the number of new
relative technological position for all the diameters drives introduced by the firm. Because our data are left
it produced in a particular year, using the following truncated until 1976, we control for any systematic
two-step procedure. The areal density (A) of the best effect for firms that entered prior to 1977 by coding
drive produced by firm i in diameter j in year t is the dummy variable incumbent76 as 1, and 0 other-
firstdividedbythehighestarealdensityinthatdiam- wise. Given our focus on firms entering a new mar-
eter available in the market that year to obtain the ket, we code the dummy variable incumbent in a prior
firm’sdiameter-specificrelativetechnologicalposition market as 1 for incumbents in the industry prior to
(TC (cid:12): the new market introduction, and 0 for new entrants
ijt
A (firms that entered since the last market was intro-
TC = ijt (cid:14) (3)
ijt max(cid:13)A (cid:12) duced).Thedummyvariablediversifiedfirmisequalto
ijt
1forfirmsthathadoperationsinotherindustries,and
As discussed in Agarwal et al. (2004), measuring a
is 0 for pure play disk drive manufacturers. Finally,
firm’s technological capabilities in comparison with
inrobustnesschecks,wereplacetheincumbentin1976
the best drive in that market circumvents problems
and firm entry year dummies with the entry year of the
related to cumulative and absolute increases in tech-
firm into the industry (imposing the limitation of 1976
nological know-how over time, because it is a relative
as the entry year for firms that entered prior to 1977).
time-varying measure that reflects a firm’s competi-
Among industry-level controls, industry sales is the
tive positioning. We then average this measure over
logged value of all disk drive sales per year, and
all diameters (j = 1 to n) produced by the firm in
industry growth is the annual percentage change in
year t to obtain a measure of the firm’s average rela-
industrysales.Thetechnologicalfrontieroftheindus-
tive technological capability (TC ) in that year:
it try is measured as the highest areal density in the
(cid:1)
TC industry over all the diameters produced in a given
TC = j(cid:1) ijt(cid:14) (4)
it year. Growth in areal density, measured as the per-
j
centage change in highest areal density between any
In robustness checks, we use alternative measures two years, was used to control for the rapid changes
of technological capabilities. Specifically, instead of in areal density. To control for competitive density
using the average of the best drives across all diam- effects, we use both linear and quadratic specifica-
eters produced by a firm, we used (a) the areal den- tion for the number of firms in a year and include the
sity of the best drive in the latest diameter produced number of new entrants in the year. Finally, we include
by the firm (latest technological capabilities), (b) the dummies for firm entry year to control for differences
areal density of the best drive in the diameter prior in founding conditions.
to the last diameter introduced (recent technologi- Table 1 provides descriptive statistics and correla-
cal capabilities), and (c) the inverse of the coefficient tionsforallthevariables.Thecorrelationsdonotindi-
of variation in the areal density, measured as TC catethepresenceofmulticollinearity,andthevariance
it
divided by the standard deviation in the areal densi- inflation factors confirm its absence. In Table 2, we
tiesacrossthedrivesproducedbythefirm(variation- provide the count and average technological capabil-
adjusted technological capabilities).4 ities at time of entry into markets based on whether
Control Variables. Several firm-level and industry- the firms were entrants or incumbents in the indus-
level variables are included as controls. Among firm- try, and whether they were pioneers, responders, or
level controls, firm tenure is measured as the number did not enter subsequent markets after their initial
of years elapsed since a firm entered the disk drive market entry. On an average, 88% of the firms in the
industry; a squared term permits potential nonlinear industry survived to the following year. As shown
effects of tenure. We use the logged value, in millions in Table 1, the average value of pioneering across all
of dollars, of all disk drive sales per year as our mea- firm-year observations is 0.09, and of responding is
sure of firm sales. Foreign firm is a dummy variable 0.28.Amongthenewentrantsintheindustry,11firms
pioneered a new market, whereas 181 firms entered
4We note that, consistent with our other variable measures, after the first year of a market’s introduction. Among
our measure of technological capabilities is on an annual basis. theincumbents,therewere10instancesofpioneering,
Although we cannot conduct robustness checks for a measure 92 instances of responding, and 120 firms who did
basedonquarterlytechnologicalcapabilitiesduetomissingquar-
not enter a new product market. Thus, it is notable
terly observations, the correlation coefficient of technological capa-
that that the pioneers into new product generations
bilitiesmeasuredquarterly,andannuallyforfirmswhensuchdata
areavailable,is0.70. have been almost evenly split between entrants and
Francoetal.: ModeratingEffectsofTechnologicalCapabilitiesonMarketPioneering–FirmSurvivalRelationship
ManagementScience55(11),pp.1842–1860,©2009INFORMS 1851
Table2 NumberofEntriesandTechnologicalCapabilitiesatTimeof
0
2 1 EntryIntoNewMarkets
6
19 10(cid:1)5 Technologicalcapabilities
Numberof attimeofentry
18 1 0(cid:1)440(cid:1)14 entries (std.dev.)
Newentrantsintheindustrya
6 79
17 10(cid:1)0 0(cid:1)10(cid:1)0 Pioneersb 11 0.59(0.31)
− −−
Nonpioneers 181 0.38(0.27)
16 10(cid:1)740(cid:1)02 0(cid:1)180(cid:1)07 Incumbentsintheindustryc
− −− Pioneersb 10 0.48(0.27)
934 85 Marketresponders 92 0.41(0.17)
15 10(cid:1)90(cid:1)7−0(cid:1)0 −0(cid:1)1−0(cid:1)0 Noentryinsubsequentmarkets 120 0.38(0.28)
1157 61
14 10(cid:1)00(cid:1)00(cid:1)00(cid:1)0 0(cid:1)00(cid:1)0 aTheF-statisticfordifferencesamongthedifferenttypesoffirmsinthe
− − − newentrantgroup=1(cid:1)37(notsignificant).
13 1 0(cid:1)060(cid:1)810(cid:1)740(cid:1)590(cid:1)09 0(cid:1)140(cid:1)01 bTheF-statisticfordifferencesamongpioneersthatareentrantsversus
−−− incumbents=1(cid:1)36(notsignificant).
7 7 818 48 cTheF-statisticfordifferencesamongthedifferenttypesoffirmsinthe
12 1−0(cid:1)4 −0(cid:1)00(cid:1)60(cid:1)50(cid:1)5−0(cid:1)0 −0(cid:1)1−0(cid:1)0 incumbentgroup=1(cid:1)62(notsignificant).
29 87485 34
11 10(cid:1)50(cid:1)4 0(cid:1)10(cid:1)30(cid:1)30(cid:1)50(cid:1)1 0(cid:1)10(cid:1)0
− −−− incumbents. Moreover, not only are the technological
10 10(cid:1)240(cid:1)140(cid:1)1 0(cid:1)050(cid:1)070(cid:1)060(cid:1)120(cid:1)25 0(cid:1)360(cid:1)08 capabilities of these two groups similar at the time of
− − entry (Table 2), they also evolve the same way over
6739 29 44 17
9 10(cid:1)30(cid:1)20(cid:1)10(cid:1)0 0(cid:1)00(cid:1)00(cid:1)10(cid:1)10(cid:1)1 0(cid:1)20(cid:1)0 the first five years subsequent to entry into the new
− − −−−− −−
market (Figure 3). Across all firms, the average firm
63342 5671 9
8 10(cid:1)10(cid:1)50(cid:1)00(cid:1)00(cid:1)0 0(cid:1)00(cid:1)00(cid:1)00(cid:1)00(cid:1)3 0(cid:1)30(cid:1)1 technological capabilities in a year is 0.38. The first
− −
quartile value is 0.16, whereas the value at the third
939372 6 936 61
7 10(cid:1)10(cid:1)00(cid:1)10(cid:1)10(cid:1)00(cid:1)1 0(cid:1)10(cid:1)10(cid:1)00(cid:1)10(cid:1)1 0(cid:1)20(cid:1)1 quartile is 0.57. The top 10% of the firms had techno-
− −−−
7982396 53214 92 logical capabilities higher than 0.8.5
6 10(cid:1)10(cid:1)30(cid:1)10(cid:1)40(cid:1)20(cid:1)10(cid:1)2 0(cid:1)00(cid:1)20(cid:1)20(cid:1)20(cid:1)3 0(cid:1)50(cid:1)2
− − −−−
96655139 26469 3 Estimation
5 10(cid:1)80(cid:1)20(cid:1)4−0(cid:1)20(cid:1)50(cid:1)3−0(cid:1)20(cid:1)2 0(cid:1)1−0(cid:1)2−0(cid:1)2−0(cid:1)20(cid:1)4 0(cid:1)60(cid:1)2 We use hazard rate methodology to test our hypothe-
662 681 986 85 ses relating to the probability of a firm surviving in
4 10(cid:1)30(cid:1)20(cid:1)10(cid:1)1−0(cid:1)20(cid:1)10(cid:1)0−0(cid:1)10(cid:1)1 0(cid:1)1−0(cid:1)1−0(cid:1)0−0(cid:1)00(cid:1)2 0(cid:1)30(cid:1)2 a given year. Several discrete and continuous time
3 10(cid:1)210(cid:1)450(cid:1)30(cid:1)180(cid:1)37−0(cid:1)140(cid:1)340(cid:1)15−0(cid:1)090(cid:1)13 0(cid:1)08−0(cid:1)08−0(cid:1)06−0(cid:1)080(cid:1)79 0(cid:1)430(cid:1)1 05level. mraotedsel(sAallriesonava1i9la9b5)l;e wfoer uthsee eastfiimrmat-iyoenarofobhsaezravrad-
2 10(cid:1)10(cid:1)120(cid:1)070(cid:1)080(cid:1)020(cid:1)060(cid:1)040(cid:1)080(cid:1)090(cid:1)080(cid:1)06 0(cid:1)040(cid:1)10(cid:1)10(cid:1)110(cid:1)2 0(cid:1)10(cid:1)04 he0. tion structure with a complementary log-log speci-
1 10(cid:1)030(cid:1)08−0(cid:1)1100(cid:1)040(cid:1)240(cid:1)05−0(cid:1)04−0(cid:1)08−0(cid:1)13−0(cid:1)080(cid:1)04− 0(cid:1)030(cid:1)050(cid:1)050(cid:1)050(cid:1)07 0(cid:1)130(cid:1)08 antatt fioncalytioann.nuBaelclayu,stehisthfeordmautalatoinonsualrlvoiwvasluasretourpedcoavteedr
− − − − c
fi continuous-timehazardratesfromdiscrete-timedata.
2 0 8 ni
Max 1111411,68116(cid:1)311117(cid:1)10(cid:1)32(cid:2)637(cid:1)5 0(cid:1)9836,883161 11973 aresig Araltsioo,nsuofchtiames-pveacriyfiicnagtiocnovaallroiawtess.foFruerathsieerrmionrceo,rpwoe-
6(cid:1) control for firm-level unobserved heterogeneity by
2 613 3 0
Min 000000−18(cid:1)400014(cid:1)7−0(cid:1)13(cid:1)7 −705(cid:1)72562500 00 ove(cid:1)0(cid:1) urosbinugstatroanadlotemrn-eaftfievcetshsapzeacridficartaitoen.6esOtiumrarteiosnulttsecahre-
b
Std.dev. 0(cid:1)330(cid:1)290(cid:1)430(cid:1)275(cid:1)76156(cid:1)0111(cid:1)260(cid:1)380(cid:1)420(cid:1)460(cid:1)670(cid:1)1434(cid:1)56 46(cid:1)9516(cid:1)861(cid:2)897(cid:1)214(cid:1)480(cid:1)49 14(cid:1)024(cid:1)73 ationsa nwiiqsueeesx,pinocnleundtiinagl mCooxdeplsro.portionalhazardandpiece-
el
an 0(cid:1)880(cid:1)090(cid:1)280(cid:1)386(cid:1)039(cid:1)576(cid:1)960(cid:1)180(cid:1)230(cid:1)316(cid:1)340(cid:1)142(cid:1)87 6(cid:1)142(cid:1)845(cid:1)068(cid:1)010(cid:1)4 9(cid:1)432(cid:1)17 corr
Table1DescriptiveStatistics VariableMe 1.Survival2.Pioneeringmarkets3.Respondingtomarkets4.Technologicalcapabilities5.Firmtenure26.Firmtenure67.Firmsales8.Diversifiedfirm9.Foreignfirm10.Incumbentin1976dummy11.Industrysales112.Industrygrowth13.Highestarealdensityin19industry14.Growthinhigharealdensity−15.Numberoffirms6216.Numberoffirms4(cid:2)2117.Numberofentrants18.Incumbentinpriorsubmarket19.Numberofdrives20.Numberofnewdrivesintroduced Notes.N=1(cid:2)199observations.All ficatopbur56eeanTFveanrcspmcicehchuxceakroenolebatnitdonscnigiddlmll-deiooiieittnsistesgifetitefeico.ribeboschlinceyibnanawtaculesoladchtcmelufiioaodeosoxorpngneredeuaiecdosebtasoatb-lipifoslelsmsioffcuetmfaniaarefterfipvdthceseeatcaasess(itbnnseil(eiwoenioAlkslenitteitteltsiivlTshelmaiietaassvhaloarbastoitetnshliioeesloefihsatd1i1frba9cw)mdldl9t,feaeii5oi-stolcn)yhetlih.forndseinttwaibhhncortuieae.holntotTetvuigigonhaseisngteerxcthiaaiaoswseolnontff.difimcctfiaaehIxtrnepreiemrdnccadepo-hbpefirdseniorptrlfuosiarmfotitmeltisreoiectsadvgmstettsiianifccinbrocetaaiygoesr-ll