Barnett y McKendrick - 2004 - Why Some Organizations More Competitive

Why Are Some Organizations More Competitive than Others?
Evidence from a Changing

Global Market
Author(s): William P. Barnett and David G. McKendrick
Source: Administrative Science Quarterly, Vol. 49, No. 4 (Dec., 2004), pp. 535-571
Published by: Johnson Graduate School of Management, Cornell University
Stable URL: http://www.jstor.org/stable/4131490
Accessed: 18/10/2010 22:36
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=cjohn.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Johnson Graduate School of Management, Cornell University is collaborating with JSTOR to digitize, preserve
and extend access to Administrative Science Quarterly.
http://www.jstor.org
Why Are Some

OrganizationsMore
CompetitiveThan
Others? Evidence from
a ChangingGlobal
Market
William P. Barnett
StanfordUniversity
David G. McKendrick
Universityof Durham
In this paper, we question the idea that large organizations

have advantages that make them particularlypotent rivals.
We argue that the ability of large organizations to ameliorate competitive constraints insulates them from an important source of organizational development and protects
them from being selected out if unfit. Consequently, we
predict that although large organizations are likely to do
well in technology contests, they also are likely to become
weak competitors over time compared with small organizations. We specify this prediction in an explicit model of
"Red Queen" competition, in which exposure to competition makes organizations both more viable and stronger
competitors. We find support for our ideas in empirical
estimates of the model obtained using data on hard disk
drive manufacturers. Large organizations led the technology race in this market yet failed to develop into stronger
competitors through Red Queen competition compared
with their small counterparts. We also find evidence that all
organizations in this market generated increasingly global
competition, regardless of the competitiveness of their
home markets. In these ways, our model elucidates important reasons why some organizations are stronger competitors and reveals how strategies that isolate organizations from competition may backfire.*
Variousorganizationaltheories regardcompetition to be a
centralforce shaping, and generated by, organizations.In
contrast to the field of economics, in which competition is
characterizedas a propertyof markets or market segments,
organizationtheories typicallyhighlightthe fact that organizations themselves differ in their competitiveness. For
instance, resource dependence and network researchers
have revealed importantdifferences among organizationsin
terms of their competitive positions (Pfeffer and Salancik,
1978; Burt, 1992; Podolnyand Stuart, 1995). At the interface
of organizationtheory and strategic management, central
importanceis given to organization-specificdifferences in
capabilitiesthat make them more or less competitive (Barney
and Zajac,1994). Similarly,organizationalecology research
often allows for competitive differences among forms of
organizations(Carrolland Hannan,2000). Throughsuch
research, we have made considerable progress in addressing
the question so often asked in casual discussions about business: Why are some organizationsmore competitive than
others?
Perhaps the single most importantcharacteristicdetermining
an organization'scompetitiveness is size, because various
sources of competitive advantage are known to co-varywith
@ 2004 by Johnson GraduateSchool,

CornellUniversity.
0001-8392/04/4904-0535/$3.00.
Thanksfor comments from GlennCarroll,

MikeHannan,Olga Khessina,Dan
Levinthal,Don Palmer,and TobyStuart.
We appreciatesupportfromthe AlfredP.
SloanFoundationand the Information
StorageIndustryCenter(to McKendrick)
and fromthe StanfordComputerIndustry
projectand the StanfordGraduateSchool
of Business (to Barnett).
size. Institutional economists have long argued that with

organizational size comes the opportunity to exploit economic
and technical advantages through the rationalization of production (Simon, 1945; Chandler, 1977), innovation (Galbraith,
1967), and transactions (Williamson, 1985). The strengths of
large organizations also are featured in sociological research.
Large, established organizations affect their environments
(Selznick, 1949), shaping other organizations (DiMaggio and
Powell, 1983) and reducing competitive threats (Pfeffer and
Salancik, 1978). Similarly, scholars have noted the politicaleconomic importance of large organizations as "consequen535/Administrative Science Quarterly,49 (2004): 535-571
tial actors" that affect state policies (Laumannand Knoke,

1987), that are favored by their connections to elite networks
(Mizruchi,1982; Mintzand Schwartz, 1985; Palmerand Barber, 2001), and that are rewardedfor being prestigious
(Podolny,1993). In this light, it is not surprisingthat studies
reveal an overwhelming survivaladvantage for large organizations (Carrolland Hannan,2000).
Althoughthe competitive advantages of large organizations
are well understood, these very advantages, when viewed
dynamically,can be seen to make largerorganizationsweaker competitors in the long run.To frame our arguments, we
draw a distinctionbetween two distinct logics of competition
prevalentin the literature:competition as contest and competition as constraint. Understood as a contest, competition
favors organizationsthat can remain up to date. In many contexts, such as in changing, global industries,this logic highlights the advantages of large, technicallysophisticated organizationsthat can stay on the cutting edge. By contrast,
understood as a constraint,competition stimulates organizational development and selects out weak competitors, further
intensifyingcompetition in a self-exciting process known as
the "Red Queen." Underthis second logic, the very
strengths that make large organizationsgood at managing
the constraints of competition in the short run may backfire
in the long run, insulatingthem from the Red Queen process
and so renderingthem less competitive as a result. Consequently,the answer to the question "Whyare some organizations stronger competitors?" hinges on which logic of competition guides one's analysis. Our purpose in this study is to
show that this difference is importanttheoreticallyand to
demonstrate this difference empiricallyin a study of the global harddisk drive market.
TWO LOGICSOF DYNAMICCOMPETITION
Manytheories include some logic of competition among
organizations.Two such logics appear across various organizationallyfocused treatments of the subject and are interesting because they have very different dynamic implications.
Some theories depict competition as a race or contest
among organizationsthat strive to surpass one another in different ways. These theories emphasize that in any given
organizationalcontext, ongoing changes in products, services, and technologies are typicaland often are the means
throughwhich organizationscompete. By contrast, other theories conceive of competition as a constrainton organizations, such as when price competition disciplines an organization to lower costs or improvequalityat a given cost. Each of
these logics has dynamic implications for organizational size.
Competition as Contest
Several theories conceive of competition among organizations, at least implicitly, as a race-like contest in which organizations are rewarded for remaining up to date, often with reference to technology or product-development races. Perhaps
the earliest theoretical treatment of such competition comes
from evolutionary economics, which elaborates a rationalchoice approach to this dynamic problem. The challenge from
536/ASQ, December 2004
Changing Global Market
an evolutionary-economicsperspective is to understandthe
incentive propertiesof the contest: Is it rationalto invest in
being first? Schumpeter's (1934, 1950) often-cited answer is
that competition renders investments in innovationirrational,
unless by being first one can enjoy a periodof monopoly-like
returnsthat (may) make innovationpay off (referredto as
"entrepreneurialrents"). As Nelson and Winter (1982) noted,
an organizationin a permanent monopoly position would not
have an incentive to disturbthis status quo by innovatingthe so-called "lazymonopolist" problem (see also Jewkes,
Sawers, and Stillerman,1969; Kamienand Schwartz, 1982).
Similarly,politicalanalyses of organizationalbehaviorreveal
that internalorganizationalpolitics might rationallylead organizationsto resist innovation(see Sch6n, 1967; Zald, 1970;
Frost and Egri,1991). But as long as competition is likelyto
materializeeventually (Fudenbergand Tirole,1985), Schumpeter's story holds in that the promise of a temporarymonopoly provides an incentive to invest in riskyinnovation.
Because such a monopoly position typicallycorresponds to
largerorganizationalsize, the implicationis that size
enhances an organization'schances of racingwell.
Temperingthe Schumpeterianhypothesis, other researchers
have argued or demonstrated that (typicallylarge)incumbent
organizationssometimes resist radicaltechnological changes,
preferringinstead to engage in innovationsthat buildon the
status quo (Menzel, 1960; Normann,1971; Tushmanand
Anderson, 1986; Dosi, 1988; Henderson and Clark,1990;
Banburyand Mitchell,1995; Christensen and Bower, 1996).
In some instances, radicaltechnological innovationsrequire
fundamentalorganizational-structural
transformations,which
in turn are likelyto be resisted, especially in large, complex
organizations(Hannanand Freeman, 1984; Carrolland Teo,
1996; D'Aunno,Tucci,and Alexander,2000; Hannan,P6los,
and Carroll,2003a, 2003b). Keeping in mind the problem of
structuralinertia,then, suggests that the role of organizational size in technological races depends on whether the
changes buildon the status quo: with the exception of radical
changes that destroy incumbent advantages, large organizations tend to race well along established, programmatictechnologicaltrajectories.
Schumpeter's argument has motivated much of the empirical
literatureon technology diffusion, in which early adoption of a
technology is claimed to provide largerbenefits than later
adoption, among survivors,that is (see Mansfield, 1961,
1968; Rogers, 1995). Some of the studies in this vein look at
the global disk drive industry, using part of the data we analyze here. Christensen, Suarez, and Utterback (1998) found
that disk drive manufacturers are more likely to survive if
they remain up to date technologically. Lerner's (1997) disk
drive study showed that organizations running just behind the
technological leader have been most likely to move up in the
race, reinforcing the importance of the technological-contest
perspective in this industry. Regarding organizational size,
Lerner found that larger disk drive manufacturers are especially likely to adopt new technologies, a finding consistent
with Schumpeter's ideas and with other work in the technol537/ASQ, December 2004
ogy diffusion literature(see Mansfield, 1963a; Rogers, 1995;

Swamidass, 2003).
Variousother theories have focused on differences among
organizationsin their abilityto innovate (Cohen and Levinthal,
1990; lansitiand Clark,1994; Teece, Pisano, and Shuen,
1997). The classic renderingof this argument also is by
Schumpeter (1950), who argued that large organizationshave
the capabilityto bringnew ideas to marketand the long-term
time horizonsto make this rational.Investigatingthis claim,
Mansfield (1963b) found that the largest firms are most likely
to account for a large partof an industry'ssuccessful innovations. Nelson and Winter (1982) added the possibilitythat
organizationsmay come to routinizethe innovationprocess
and, in so doing, rationalizethe timing of relativelyrapidproduct introductions.They concluded that this rationalization
favors large firms (see also Cohen and Klepper,1996; Klepper
and Simons, 2000). Similarly,Galbraith(1967) argued that
large organizationswere the innovativeengines in the U.S.
economy duringmuch of the twentieth century because
these organizationswere able to manage the innovation
process. Chandler(1962) emphasized the structuraladvantages of large organizationsas engines of innovation,in arguments that drew on the Carnegie School (Barnard,1948;
Simon, 1945), which depicted complex organizationsas especiallyeffective mechanisms for coordinatingand controlling
activities entailingmultipletime horizons. Cyert and March
(1963) noted that with increasingsize, organizationswere
able to buildinto their procedures mechanisms for innovation
and improvement. More recently, empiricalresearch supports
this idea, findingthat largerorganizationsare more likelyto
make relativelyprogrammaticchanges of many kinds (Haveman, 1992, 1993; Minkoff,1999; Greve, 1999; Havemanand
Nonnemaker,2000; Sorensen and Stuart,2000; Palmerand
Barber,2001; Henisz and Delios, 2001; Khessina, 2002;
Chuangand Baum, 2003). Overall,the picture painted by
these various literaturesfeatures large organizationsas particularlyadept at adaptive change in technology and product
races, at least when the changes involvedare incrementalor
programmed(Jewkes, Sawers, and Stillerman,1969; Abernathy and Utterback,1978).
Similararguments on the advantage of large scale also feature prominentlyin the internationalbusiness literature.Globalizationhas been viewed as an outcome of oligopolistic
competition, whereby the largest firms in concentrated industries drive foreign investment througha follow-the-leader
process (Knickerbocker,1973; Yuand Ito, 1988; Liand
Guisinger,1992). Such competition is seen as the norm
across industries and is reflected in international business
textbooks (Vernon, Wells, and Rangan, 1996: 55). The leading
theories in international business similarly underscore the
advantages of scale in global competition. Dunning's (1993)
"eclectic theory" of the multinational corporation (MNC), for
example, emphasizes that MNCs require certain "ownership
advantages" over purely domestic firms to stay ahead in
global markets (see also Hymer, 1960). In addition to its
intangible assets, the scale at which an MNC operates is a
central ownership advantage and provides economies in
538/ASO, December 2004
sourcing, production,research and development (R&D),

advertising,and administration.Both Porter's(1990) analysis
of how firms sustain advantages in internationalcompetition
and Chandler's(1990) comparativehistory of the expansion
of the modern corporationcomport with this view.
The racingmetaphor is widely used in analyses of global
competition. Time-to-marketpressures are severe in a range
of global industries, includingsemiconductors, harddisk
drives, and some segments of the clothing industry(Abernathy et al., 1999; McKendrick,Doner, and Haggard,2000;
Leachmanand Leachman,2004). Faster cycle times also
affect the R&Dstrategies of global firms. In a number of
industrialcontexts, evolutions in productand process technologies are occurringat such a rapidpace that many firms
have established foreign R&Dsites to accelerate the generation and acquisitionof new knowledge (Kummerle,1999;
Murtha,Lenway,and Hart,2001; Le Bas and Sierra,2002).
Others have emphasized the benefits to moving early in
establishing an internationalnetwork, which can lead to reputation, scale, and learningadvantages. Firmsthat move quickly into new segments can displace internationalleaders that
are slow to exploit the structuralchanges (e.g., Porter,1990).
In sum, there is widespread consensus that large organizations with global reach are thereby advantaged when competition takes on the qualities of a technology or product race.
From Competitive Constraints to the Red Queen
Sociologicaltheories typicallyregardcompetition not so
much as a race as a powerful constraintoperating on organizations. FollowingThompson (1967), resource dependence
theory places special emphasis on competition as a primary
source of uncertaintythat constrains organizations,their
structures, and their actions (Pfeffer and Salancik,1978).
More recent organizationtheories make reference to Simmel
(1955), whose observation that competition constrains actors
to work toward the benefit of others forms the basis of ecologicaland network theories of competition alike (Hannanand
Freeman, 1989; Burt, 1992). In organizationalecology, competition is thought to play an especially importantrole in
selecting for or against various forms of organizations
depending on whether they conform to environmental
requirements, both technical and institutional(Carrolland
Hannan,2000). The related variation-selection-retention
frameworkalso portrayscompetition as a constrainingforce
drivingselection among organizations(Aldrich,1999).
Research on network structures among organizations, similarly, regards competition as a primary constraint shaping the
actions and fates of organizations (Burt, 1987).
Global competition also can be understood as a constraint,
one that is shaped by powerful institutional forces (Ghoshal
and Westney, 1993). While the behaviors and structures of
firms generally reflect the practices and business models of
their home markets, firms engaged in global competition are
subject to quite different institutional environments (Rosenzweig and Singh, 1991; Kostova and Zaheer, 1999). All states
influence firm behaviors through coercive and normative
pressures, but states vary in their policies toward competi539/ASQ, December 2004
tion, labormarkets, corporategovernance, and the like (Lindblom, 1977; DiMaggioand Powell, 1983; DiMaggio, 1990;
Westney, 1993; Halland Soskice, 2001). Thus, states constitute powerful institutionalagents shaping how global competition affects firms.
Inan influentialarticle,VanValen(1973)
invokedthe Red Queen metaphorto
describe the coevolutionaryprocess in
which viabilityand competitiveness each
strengthenthe other.Strongercompetitors increase selection pressures, yielding
more fit survivors,which in turngenerate
strongercompetitionand so on in a selfacceleratingprocess of reciprocalcausality. The Red Queen refers to a character
from Lewis Carroll'sThroughthe Looking
Glass,to whom Alice comments that
althoughAlice is running,she does not
appearto be moving.The Red Queen
responds that in a fast-movingworld "it
takes all the runningyou can do, to keep
in the same place."As this image suggests, the relativepositions of playersin
Red Queen competitionmay be stable,
even thoughthe race is producing
absolute change for the system as a
whole.
By and large, research has examined competitive constraints

as a force to be reckoned with at a given point in time. But
what happens, over time, when an organizationresponds to
competitive constraints?This response, in turn, represents a
furtherconstrainton the organization'srivals;in fact, a defining characteristicof competition is that one organization's
solution becomes its rivals'problem. The resulting increased
constraints, again in turn, are likelyto trigger responses
among rivals,again intensifyingcompetitive constraints on
the first organization,and so on. This escalating system of
reciprocalcausality,dubbed "Red Queen" competition by the
biologist VanValen(1973), has been found to strongly affect
rates of founding,growth, and failureamong organizations
(Barnettand Hansen, 1996; Barnettand Sorenson, 2002).1
Ouraim here is to see how organizationsshape this coevolutionaryprocess in unintendedways as they become large
and strategicallyformidable.
Among organizations,Red Queen evolution can come about
both throughorganizationallearningand by naturalselection.
To consider the role played by organizationallearning,we
buildon the model developed by Marchand his colleagues
(March,1988, 1994) and several of the basic assumptions in
that model. First,assume that people in organizations"satisfice" when confronted with the need to make decisions
(Marchand Simon, 1958). A so-called problemisticsearch for
alternativesis triggeredwhen performancefalls below some
aspirationlevel and is continued untilperformanceis considered satisfactory (Cyertand March, 1963). This search proceeds sequentially,presumably,stopping at the first satisfactory solution, ratherthan continuinguntilthe best possible
solution is found. Search also is assumed to remain "local,"
restrictedto solutions that are only incrementallydifferent
from currentpractice,and only moving to more distant possible solutions when no satisfactory local solutions are found
(Levinthaland March,1981). In this search process, learning
through imitationmight also occur (Mezias and Lant, 1994).
By these assumptions, organizationsadapt incrementallyin
an effort to maintainat least a minimum,satisfactory level of
performance.
Now consider a populationof competing organizations,each
behaving accordingto the satisficing model. In this context,
the organization-learning process does not end once a given
organization improves its performance by adopting some
new practice. Instead, the innovating organization, by improving its own performance, now has increased the intensity of
competition felt by the other organizations in the population.
At some point, this increased competitive intensity may
reduce performance in other organizations enough to trigger
search in these organizations. As each of these organizations
finds solutions that restore its performance, in turn, competition again increases for the rest of the population, again triggering the search for improvements. So learning and compe540/ASQ, December 2004
tition are linkedcausally,each accelerating the other in the

ongoing process of Red Queen evolution.
In addition,aspirationlevels might in fact change endogenously as partof Red Queen evolution. We know that aspiration levels among individualsadjust rapidly(Lant,1992) and
that often they are defined by social comparison in competitions that hinge on the relative positions of players (Herriott,
Levinthal,and March, 1985; Frankand Cook, 1995). If organizationalaspirationlevels are influenced by comparisons with
their competitors, as is often the case, then we might see
aspirationsratchetingupwardamong organizationsinvolved
in Red Queen evolution. Inthis way, makingaspirationlevels
endogenous is likelyto maintainthe process of Red Queen
evolution over time.
Red Queen evolution can also result purelythrough natural
selection among organizations.Assume that organizations
differwith respect to their fitnesses and that these differences remainedfixed over time. Assume also that competition culls from organizationsaccordingto these fitness levels,
so that as competition intensifies, the less fit organizations
are selected out (see Levinthal,1997). Among survivingorganizations,then, those that have faced very little priorcompetition would include both fit and unfit organizations.By contrast, fitness levels will generally be higher among
organizationsthat have survived considerable priorcompetition. Both through selection and learning,then, the process
of Red Queen evolution implies that organizationsare both
more viable, and more competitive, the more that they have
faced competition in the past.
The relativestabilityof players in Red Queen evolution
results from the mutuallyreinforcingincreases in each organization'sviabilityand the potency of each organization's
rivals.Our model of Red Queen evolution separates the two
parts of this dynamic, distinguishingbetween the organizational and the ecological implicationsof exposure to competition (see Barnett, 1997):
r(t) =
+ a(OkjEk)],
rj(t)*exp[P3Ej
where r.(t)is the failurerate for organizationj; r1(t)*is j's baseline failurerate as a function of its age t, its currentcompetitive context, and other observables; aoand 3 are coefficients
to be estimated; and E refers to the cumulative priorexposure to competition of organizationj or its currentrivalsk,
measured in organization-yearsso that, for instance, if organization j competed with four rivalsduringeach of its first and
second years of life, it would have a value of E = 8 duringits
thirdyear of life. If priorexposure to competition increases
organizationalviability,as implied by Red Queen evolution,
then we can expect 1 < O0.Meanwhile, the ecological consequences of Red Queen evolution materializeif a > 0, where
j's rivals'priorexposure to competition increases their competitive strength. Whether the twin consequences of Red
Queen evolution are offsetting, then, is treated by our model
as an empiricalquestion that depends on the exact magni541/ASQ, December 2004
tudes of oaand 3 as well as the observed values of E for any

given organizationand its rivals.
Organizational Size and Red Queen Evolution
Thus far we have assumed that an organizationresponds to
competitive constraints by searching for improvements, but
for large organizationsthere is an alternativeto the ongoing
process of Red Queen evolution. Organizationsthat have
attained positionaladvantage-market position, social prestige, centralityin social networks, politicalpower, and the
like-may attenuate or even eliminate the threat of competition from others. In fact, the modern field of strategic management in business education exists primarilyto investigate
and teach methods for findingsafety from the forces of competition, stimulated in large part by Porter's(1980) application
of industrialorganizationeconomics to the problem of competitive constraints. In sociology, similarideas appear in
Selznick's (1949) early work on cooptation, featuringlarge
organizationsof politicalimportancethat absorb interests and
avert threats. More recently, researchers have found large
organizationsto be especially capable of staking out and
defending their strategic position (Haveman, 1993; Barnett,
Greve, and Park,1994) and of maintaininginterlockswith
other importantactors (Konoet al., 1998). Yet precisely
because large organizationsare capable of avertingcompetitive constraints, they may be less susceptible to the Red
Queen process.
When faced, nonetheless, with competitive constraints, large
organizationsalso have distinct advantages in coping with
these pressures. Largeorganizationstypicallybuffer key
parts of the organizationfrom the external environment,
especially when they face complex and changing environments (Thompson, 1967; Meyer and Rowan, 1977). In light
of Red Queen evolution, this abilityimplies that key parts of
large organizationsmay remain insulated from competitive
threats and so may not recognize the imperativefor change.
Small organizations,by contrast, typicallyare less able to
insulate themselves from competitive threats. Theoryand
evidence also show that largerorganizationsare more capable of decoupling key activities from environmentalpressures, especially competition (Meyer and Rowan, 1977; Pfeffer and Salancik,1978), while small organizationsare known
to exhibit tight linkingbetween environmentalpressures and
internalprocesses (Hickson,Pugh, and Pheysey, 1969).
Moreover,large organizationsoften experience an increasing
isolationof their leadership (Michels, 1949; Gusfield, 1957)
and throughouttheir rankand file see a smaller proportionof
organizational members having contact with the external
environment (Blau, 1977). Overall, large organizations are less
directly affected by competition (Barron, 1999; Bothner,
2003) and so are less likely to conform to the predictions of
the Red Queen model.
Even when affected by competitive threats, however, large
organizations are likely to be less responsive than are small
organizations. As Hannan and Freeman (1984) argued in their
theory of structural inertia, with size come pressures for reliable behavior. Leadership in large organizations tends to
become less immediatelyresponsive to external demands

over time because of increased formalizationand rule-governed behavior(Weber, 1946; Bendix, 1956). Consequently,
large organizationstypicallyare designed to behave according
to established routines, continuingto behave in expected
ways ratherthan responding sensitively to performancefeedback (Greve,2003). Furthermore,when adjustment to performance feedback does occur, such adjustment is likelyto be
less profoundthan in small organizations.Change in a small
organizationentails obtainingcooperationfrom a smaller
number of people, groups, and other organizationalunits.
Largeorganizations,by contrast, are likelyto requirecooperation from proportionatelymore parties. Moreover,large organizationsare typicallymore complex than small organizations,
and with increasingcomplexity comes an exponential
increase in the numbers of ways that changes can be
blocked (Hannan,P61os,and Carroll,2003a, 2003b). Again,
these arguments point to large organizationsbeing less
responsive to the Red Queen process.
Finally,size per se often gives organizationstechnical advantages that can help them weather competitive constraints. In
general, it is well known that largerorganizationsare less
likelyto fail, other things being equal (Carrolland Hannan,
2000; McKendricket al., 2003). This may result from many
advantages of size, includingthe tendency for largerorganizations to behave more reliably(Haunschildand Sullivan,2002).
Largeorganizationsenjoy cost advantages in many industries, in which case competition favors largerorganizations
over smaller organizationswhen they occupy the same niche
(Carrolland Swaminathan,2000; Dobrev, Kim,and Carroll,
2002). Inthis light, it is not surprisingthat density-dependent
competition has been found to have a considerablyweaker
effect on organizationsas they grow large (Barron,1999).
Institutionaladvantages also help large organizationsto maintain themselves despite marketpressures. Meyer and Rowan
(1977) noted the survival-enhancinglegitimacyof organizational practices regardless of their marketefficiency, especially as organizationsbecome large and complex. DiMaggioand
Powell (1983) regardedthis idea as centralto two kinds of
isomorphism,observing that large organizationsbenefit others because of their legitimacy. Largeorganizationsalso are
advantaged in dealing with competition due to their typically
higher social status. In general, high social status eases pressures for conformitythat affect lower-status actors (Phillips
and Zuckerman,2001; Rao, Greve, and Davis, 2001; Zuckerman et al., 2003; Bothner, 2003). Relatively high social status
is known to help large organizations cope with competition
(Podolny, 1993).
As organizations grow large, we therefore expect that they
will become less disciplined by competition, and so less susceptible to the forces of Red Queen evolution. These arguments apply both to the learning and the selection components of our theory. Confronted by weaker competitive
discipline, large organizations are less likely to be stimulated
to search for solutions. Similarly, competition generates
selection processes that are more likely to eliminate small
organizations from the population if they fail to improve.
Largeorganizations,by contrast, enjoy advantages that make

it more likelythat they will continue to survive despite their
failureto respond to competition. So naturalselection reasoning also suggests that large organizationswill be less responsive to the stimulus of competition and consequently less
likelyto conform to the predictionsof the Red Queen model.
Operationally,these ideas implya Red Queen model with
separate parametersfor large and small organizations:
=
rj(t)
rj(t)*exp [PsEsj +
ILELk + sO(ISkwjESk)+ aL('Lk?jELk)],
where Es refers to organizationj's priorcompetition experienced at times when it was a small organization,and ELjis
j's priorcompetitive experience duringtimes when it was a
large organization.Similarly,Esk and ELkrepresent the prior
competitive experience of j's rivalsk, distinguishingbetween
competition experienced when these rivalswere small or
large organizations,respectively. If we are correct that organizations limitthe Red Queen process as they attainthe power
and stature that come with size, then we should see evidence of Red Queen evolution among small organizations
more stronglythan among large organizations.In terms of
our model, this would imply:
Hypothesis 1: Ps < 0 and Ps < PL, such that priorexposure to com-
is smallreducesits failureratemore
petitionwhen an organization
thandoes priorexposureto competitionwhen an organization
is
large.
Hypothesis 2: as > 0 and as > OL,such that a rival'sexposure to
competitionincreasesthe strengthof its rivalry,

especiallywhen this
exposurehappensto a smallrivalorganization.
To summarize,theories of competition that follow a racing
logic highlightthe strengths of large organizations,especially
in contexts like the harddisk drive market, in which technologicalchange is ongoing and relativelyprogrammaticand
global reach is especially important.Conceivingof competition as a constraint,by contrast, focuses our attention on the
dynamics of Red Queen competition. Followingthis logic, the
very strengths that make largerorganizationsable to manage
constraints also make them less susceptible to the survivaland competitiveness-enhancing consequences of the Red
Queen process. Overall,then, we expect to see large organizations do well when it comes to keeping up in the hard disk
drive technology race but to be less enhanced by exposure
to competitionthan are their smaller rivals. In global competition, however, Red Queen evolution implies more complex
dynamics.
Global Competition and Red Queen Evolution
Several scholars have argued that when firms from different
countries compete, cross-nationaldifferences among organizations become apparent,differences that remain unnoticed
when firms remainconfined within the boundariesof domestic markets (Anandand Kogut, 1997). The nationalcontextincludingpeople and their expertise, culture, social struc544/ASQ, December 2004
tures, industrialnetworks, and politicaland market institutions-arguably creates similaritiesin the strengths and
weaknesses shared by organizationsfrom the same country
(Porter,1990; Kogut, 1992; Nelson, 1993). Some have speculated that such cross-nationaldifferences might account for
much of the observed worldwide variationswe see in the
viabilityof firms (Chandler,1990), and Chesbrough (1999) has
found such differences to be importantin the harddisk drive
market.
Along these lines, and pertinentto our model, is Porter's
(1990) idea that competitive experiences in one's home country translate into stronger-lobal competition (see also Porter
and Sakakibara,2001). This possibilitycan be treated as an
extension of our model to the context of global competition.
If USFand aLF represent the competition generated by small
and large foreign rivals,respectively, as a function of their

own domestic competitive experience, then we would
expect:
Hypothesis 3: aSF> 0 and o SF>
LF,indicatingthat a foreign rival's
exposureto domesticcompetitionin its home countryincreasesthe

strengthof its globalcompetitiveness,especiallywhen this exposure happensto a smallforeignrival.
Alternatively,nationaldifferences in the competition generated by firms could increase over time among all firms, concurrent with largertrends in the globalizationof markets. While
global competition may initiallyreflect firms' nationalcharacteristics, over time, as competition increases, capabilitiesand
behaviorsamong firms may converge, thereby diminishing
nationaldifferences (McKendrick,2001). To allow for this possibility,we buildinto our model an alternativeformulationof
globalizationby distinguishingwhether an organization'srivals
are foreign or domestic (Zaheerand Mosakowski, 1997). At
one extreme, if a firm's domestic rivalsgenerate very strong
competition while potential rivalsfrom other countries generate none, then marketcompetition can be seen as strictly
domestic. At the other extreme, if the strength of competition generated by both domestic and foreign-based rivalsis
equally strong, then competition can be regardedas global.
At any point in time, the observed patternof competition in
an industrycould be described as more or less global
depending on how much it resembles one of these two
extreme cases. It is possible that competition becomes more
global for all organizationsover time, regardless of exposure
to competition, as markets develop (see Hannan, 1997). In
our model, we investigate this possibility as an alternative to
hypothesis 3.
METHOD
We investigated our ideas by estimating ecological models of
competition on data describing all manufacturers of hard disk
drives. Our data are more comprehensive than those analyzed in the prior research on the industry by Lerner (1997),
Christensen (1997), Christensen, Suarez, and Utterback
(1998), Chesbrough (1999), and King and Tucci (2002). While
their data are left-censored (using market research reports
that began in 1977), ours cover the entire organizational pop545/ASQ, December 2004
ulationdating back to the beginningof the industryin 1956.

The Appendixprovides details on our data and data-collection
methods. The harddisk drive market, described in figures 1
and 2 and table 1, is an ideal setting for our study. The definFigure 1. Harddisk drive manufacturers by home region.
55
N. America
50
45
40
-
35
S30
C.
.o 25
E
20
Japan
15
S. America
?
.
rope
10
tr
0
1955
1960
1965
.Europe'1980
1975
1970
.
1990
1985
1995
Asia
2000
Year
Figure 2. Entries and exits of hard disk drive manufacturers worldwide.
18
oi
16
Exits :I
14
:
Sga
"
10.
Entries
.o8 88Entries
E
2) 6
602
'-'
ell
I'
?
e
"-.
?I '
.r
"
I
;
\"
Year
1955
1960
1965
1970
54IAQDecmbe
2004
1975
1980
Year
1985
1990
1995
2000

Table1
Description of Disk Drive Manufacturers Worldwide 1956-1998 (N = 171)
No. of organizations
No. of largeorganizations
No. in med. or high capacity
% in med. or high capacity
No. of small organizations
No. of de novo organizations
No. of de alio organizations
No. of captive producers
No. of orgs. manuf. in Asia
No. of organizationsproducingfor sale*:
1.8-inchform factor
2.5-inchform factor
3.5-inchform factor
5.25-inchform factor
8-inchform factor
14-inchform factor
24-inchform factor
39-inchform factor
1956
1960
1965
1970
1975
1980
1985
1
0
0
0
1
0
0
0
1
1
0
2
1
0
0
1
1
1.0
0
2
1
0
8
3
2
.67
5
2
.40
1
7
5
0
29
4
3
.75
25
14
56
7
22
15
0
35
12
7
.58
23
6
.26
5
30
19
0
49
40
29
.73
9
7
.78
10
39
24
0
86
47
37
.79
39
18
.46
27
59
26
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
2
0
3
0
0
0
0
0
20
1
1
0
0
0
0
0
26
0
0
0
0
0
2
16
32
0
0
0
0
18
55
20
21
0
0
Over all years

1990
1995
1998
Min.
Max.
Mean
S.D.
No. of organizations
72
1
30
20
86
35.76
27.22
No. of largeorganizations
28
14
11
0
55
17.95
17.23
23
13
9
.82
.93
.82
No. of small organizations
44
16
9
0
49
17.81
14.36
18
4
10
.41
.44
.63
No. of de novo organizations
25
15
10
0
28
10.27
9.49
No. of de alio organizations
47
1
15
10
59
25.48
18.60
No. of captive producers
21
7
1
6
29
13.40
9.14
No. of orgs. manuf. in Asia
14
16
16
0
17
4.53
6.36
No. of organizationsproducingfor sale*:
1.8-inchform factor
0
2
6
0
9
.86
2.16
2.5-inchform factor
6
7
16
0
16
2.48
4.92
3.5-inchform factor
40
20
15
0
40
8.76
13.29
5.25-inchform factor
29
4
8
0
55
12.30
18.54
8-inchform factor
13
2
0
0
27
5.81
8.78
14-inchform factor
11
0
0
0
33
12.93
12.58
24-inchform factor
0
2
0
0
0
.18
.50
39-inchform factor
0
0
0
0
3
.51
.93
* These include only "non-captive"productofferings producedfor sale. In some
a
small
number
of
years, relatively
organizationsproducedonly "captively"for theirown use, in which case these non-captiveofferingsdo not sum to the
total numberof organization?.
ing characteristicof competition in the industryhas been the

ongoing race to deliver higher storage capacities on eversmaller devices at less cost and in less time (McKendrick,
Doner,and Haggard,2000).
Moreover,the industryhas become increasinglyglobalized,
both in terms of the nationalityof the firms and in the locus
of competition. Initially,most countries' harddisk drive markets were predominatelydomestic, with domestic computer
originalequipment manufacturers(OEMs)purchasingfrom
domestic harddisk drive firms. Japanese organizations
entered the harddisk drive market seven years after the
U.S., and Europetwelve years after. Brazilian,Taiwanese, and
Koreancompanies followed in the early and mid-1980s.

AlthoughAmericanfirms aggressively marketed drives worldwide early on, most firms kept their primaryfocus on domestic markets. Perhaps for this reason, organizationsin the
industryoften are identifiedaccordingto their countryof origin, and countryfactors have been found to be importantto
the formationof harddisk drive manufacturers(Chesbrough,
1999).
As time passed, however, the harddisk drive market became
increasinglyglobal. A watershed event in the computer
industrywas the debut of the IBMpersonal computer (PC)in
1981. The PC defined the dominantdesign in the microcomputer marketfor many years (Langlois,1992; Anderson,
1995). Inadditionto setting the standardfor what a desktop
computer should include, it featured an open architecture
that attractedthe entry not only of some of IBM'sestablished mainframeand minicomputerrivalsbut de novo startups that set out to manufactureIBM "clones." The same
open architecturethat attractedthe new clone manufacturers
also stimulated entry into peripheralequipment. While mainframe and minicomputermanufacturersmade many of their
own peripheralsand components, the assemblers of personal computers outsourced almost all of their production.
Competition as Contest in the Hard Disk Drive Market
Technologicalleadership in the harddisk drive market has
been based on a firm'sabilityto deliver higher-capacitydisk
drives. Capacityis determined by how many bits can be
stored on a square inch of disk, otherwise known as the
drive's areal density. Since IBMshipped the first movablehead disk drive in 1956, the industryhas undergone tremendous technological change. Until1991, areal density
increased at an annualrate of 30 percent but grew by an
astounding 60 percent per year from 1992 to 1997, a faster
rate of progress than semiconductors, and an amazing 125
percent in 1998, our last year of study. The average rate of
capacity increase was 30-40 percent each year between
1956 and 1991.
Rapidtechnological change has allowed the marketto develop into roughlythree segments. One segment is composed
of firms that offer the highest capacity drives. A second segment consists of firms that are the early leaders into a
"capacitypoint" in demand by the largest computer manufacturers,which typicallyexcludes the highest capacity
drives. The thirdsegment encompasses the technological
laggardsthat serve the secondary marketof second-tier
mainframeand minicomputermakers and the hundreds of
small and medium-sized microcomputer companies. Firms
late to market with a new drive thus suffer a severe revenue
penalty: if too late, they may find no customers and be
forced either to absorb their development costs and start
developing an even higher-capacity drive or to exit the industry. Even so, first-to-market innovators typically hold only the
slimmest of leads, as other manufacturers generally introduce comparable products within a relatively short time. The
likelihood of rapidly decreasing profitability over the life cycle
of any given product provides a strong incentive for firms to
innovate rapidly.But the result is extremely short product

cycles, estimated in 2003 to be from six to nine months.
Most firms have had trouble keeping up with such continuous productintroductions.
Not only have firms needed to keep up with changes in
capacity,they have needed to keep pace with the reductions
in the physical size of harddisk drives, from using disks with
diameters of 39, 31, and 24 inches to 14-, 8-, 5.25-, 3.5-, and
2.5-inch drives. These changes in "formfactor" have proven
problematicfor many drive manufacturers,and most firms
have not survived the transition.Forsome firms, the inability
to introducesmaller form factors was due to technological
reasons: scaling down components and getting designs to
work properlywere engineering challenges. In other cases,
as Christensen (1997) has pointed out, new form factors
sometimes served new markets, and managers of many
incumbentfirms were slow to recognize the necessity of
change.
As if these technological and marketingchallenges are not
dauntingenough, firms also have had to achieve economies
of scale in order to compete effectively on price. While being
first to marketwas in many ways critical,first-to-volumeproduction has often been more important.Many new entrants
have been among the first to introducea particularform factor or hit the highest capacity point. But computer vendors
seldom gave a second chance to firms that were unable to
ramp up to volume productioneffectively. As a result, survival in the disk drive industryhas increasinglyrequiredthe
marriageof technological prowess to manufacturingability.
The volatilityof this organizationalpopulationhas limitedthe
abilityof market leaders to control pricingor the length of
product life cycles. Periodicoversupplyand constant price
erosion are ways of life for disk drive producers. Disk drive
manufacturersmight attempt to avoid the pitfallsof the pricesensitive, high-volume,low end of the market by differentiating their products into higher-capacitysegments, such as
drives for file servers and network storage. But ongoing innovation has made it impossible for firms to sustain a product
differentiationstrategy for long periods. In this context, then,
racingwell as technologies advance has been crucialto organizationalsuccess. As the pattern in table 1 shows, large
organizationshave maintained,over most of the industry's
history,a clear advantage in terms of the proportionof organizationsproducingin the medium-to high-capacityranges.
Consistent with Lerner's (1997) analysis of this industry, our
data show that larger organizations hold a clear competitive
advantage, at least when we conceive of competition as a
contest for being technologically up to date.
Model Specification
and Estimation
We used the modeling framework of organizational ecology

to build our model of Red Queen competition because this
approach allows for explicit estimation of competition among
organizations. In particular,the so-called density-dependent
model allows each organization's viability to vary as a function of the number, or density, of other organizations in the
population (Carrolland Hannan, 2000). Failure rates are
thought to be non-monotonicallyrelated to density, falling

with initialincreases in density as legitimacy increases but
ultimatelyincreasingas competitive effects grow with crowding in the population(Hannanand Freeman, 1989; Hannan
and Carroll,1992). Furthermore,persistently higherfailure
rates are found for organizationsfacing greater density at the
time of their birth,the so-called density-delayeffect attributed to problems caused by buildingan organizationunder
adverse conditions (Carrolland Hannan,1989). Our model
includes these various density effects. For our purposes, the
effects of disaggregated and weighted densities are also estimated, because these allow for competition to depend on
both technologicaland geographicfactors.
The harddisk drive market's historyof technological differentiation suggests that we should operationalizedensity using
technology-overlapmeasures (see Carrolland Hannan,2000).
These overlapmeasures equal density counts if all organizations compete in all technological areas. Forany of j's rivals
that overlapj's technological domain only partially,however,
they contributeto j's overlap score only in proportionto their
degree of overlap.We specified the possible technological
domain of each organizationj to include relativelylow-, medium-, and high-capacitypositions in each of the various form
factors of harddisk drives that existed in a given year. We
defined these relative levels to account for changes over
time in absolute capacity,as explained in the Appendix. For
each firm in each year, we disaggregated the raw count of its
rivalsinto technology-overlapand non-overlapdensities. If
competition in the industrywas technologicallysegmented,
as we suspect, then failurerates will be drivenespecially by
overlapdensity as opposed to non-overlapdensity.
Implementingour model of the Red Queen, we tested for
the survivalimplicationsof each organizationj's priorexposure to competition. Foreach organizationin each year t, we
measured the sum of its domestic technology-overlapdensity score in each previous year from its birththroughyear t-1.
If priorexposure to competition enhanced survivalchances,
as we argue, then this variableshould reduce exit rates,
especially among small organizations.
Each organization'stechnology mix was measured using variables that sum the number of form factors (of different relative capacity levels) in which the firm had productofferings.
Forexample, a firmwith low-capacityproductofferings in
three differentform factors would have a measure of 3 for
"numberof low-capacityform factors produced." Ifthis firm
also produced medium-capacityproducts in four form factors
and high-capacity products in one form factor, then it would
also have a measure of 5 for "number of medium- and highcapacity form factors produced." These variables allowed us
to investigate whether and how being relatively ahead or
behind in the technology race affected survival rates.
Our main measure of organizational size is a categorical measure that distinguishes between large and small firms,
defined relatively for each year, and updated from year to
year. For years prior to 1976, this designation was made by
examining historical documents to identify major players in
the industryin each year. Forobservations after 1976, we

relied on the data source Disk/Trend.
Two kinds of controlvariableswere also included in our specification.One set of variablesallows the "carryingcapacity"
for harddisk drive manufacturersto vary over time, by
region, and by technologies. We included,for each firm,the
number of computer manufacturersin the United States that
made a computer correspondingto a form factor produced
by the harddisk drivefirm.The data on computer OEMs
come from various sources, as described in Barnett,Swanson, and Sorenson (2003). This variablereflects the market
for a given firm's disk drives and so can capture symbiosis
between these complementaryparts of the organizational
community.Calendaryear was includedto control for secular
trends in the carryingcapacity not otherwise captured in our
specification. Indicatorvariableswere included measuring
whether or not a given organizationin a given year was producing in a given form factor. Because organizationscould be
in any or all form factors at once, there is no omitted category among these indicators.Also, indicatorvariableswere
includedfor the region of the global economy in which each
organizationwas headquartered,with NorthAmerica serving
as the omitted category. This approachallowed us to investigate whether our hypothesized effects account for national
differences in failurerates.
Also includedwere some other organization-levelcontrol variables. Along with our large/smallcategorical measure of size,
we includeda measure of size in terms of the dollarvalue of
harddisk drive sales, but this measure was availableonly for
large organizationsand only after 1976. An indicatorvariable,
allowed to varyfrom year to year for each firm, was included
for firms engaged in any so-called "captive"productionof
disk drives for their own computers. Another indicatorvariable represented whether an organization(fromany country)
was manufacturingdisk drives in Asian facilities in a given
year, a low-cost productionstrategy pursued by selected
firms (McKendrick,Doner,and Haggard,2000). Following
Carrollet al. (1996), we also includedan indicatorvariableto
distinguishbetween de novo firms, those that entered the
industryas a start-up,and de alio entrants, who moved or
expanded into the industryfrom some other industry.Table2
describes our pooled annualobservations over the period of
the study.
We modeled the failurerate using a piecewise-exponential
specification for each organization's market tenure and estimated the model using the software package STATA.This is
an extremely flexible specification that allows failure rates to
change freely from period to period. We selected periods to
be as fine grained as possible while not being so short as to
prevent the estimation of statistically meaningful effects.
Although many studies have measured t in terms of organizational age, we measured t in terms of market tenure. For de
novo firms that were born as hard disk drive manufacturers,
age and tenure are the same. Firms with operations in other
markets, however, may have been born before they entered
the hard disk drive market. In these cases, market tenure differs from organizational age, and we used tenure rather than
Table2
Description of Pooled Annual Observations of Disk Drive Manufacturers Worldwide, 1956-1998*
Variable
Min.
0
Organization'smarkettenure
0
Calendaryear (1956 = 0)
Numberof low-capacityform factors producedby organization 0
0
Numberof medium-and high-capacityform factors produced
by organization
0
Hard-drivesales ($mil)by organization(largeonly)
0
Numberof computermanufacturersin the formfactors
producedby organization
0
Entriesof computermanufacturersin the form factors
producedby organization
0
Numberof rivalsworldwidein year of organization'sfounding
0
Numberof rivalsworldwide
0
(Numberof rivalsworldwide)2/1000
0
Numberof domestic rivals
0
Overlapwith domestic rivals(by form factorand capacity)
0
Overlapwith small domestic rivals(by form factorand
capacity)
Overlapwith largedomestic rivals(by form factorand capacity) 0
0
Non-overlapwith domestic rivals
0
Organization'scompetitiveexperience
0
Smallorganization'scompetitiveexperience
0
Largeorganization'scompetitive experience
0
Rivals'competitiveexperience, same region(experienceweighted overlap)
Small rivals'competitiveexperience, same region (experience- 0
weighted overlap)
Largerivals'competitiveexperience, same region(experience- 0
weighted overlap)
0
Overlapwith domestic rivalstimes calendaryear (1956 = 0)
0
Numberof foreign rivals
0
Overlapwith foreign rivals(by form factorand capacity)
0
Foreignrivals'competitiveexperience (experience-weighted
overlap)
0
Smallforeign rivals'competitiveexperience (experienceweighted overlap)
0
Largeforeign rivals'competitive experience (experienceweighted overlap)
0
Overlapwith foreign rivalstimes calendaryear (1956 = 0)
* The data include171 organizationsover 1,538 organization-years.
Max.
Mean
S D.
42
42
4
9
6.91
27.29
.73
1.23
7.04
7.99
.77
1.48
11979.1
2787
243.06
530
974.19
508
135
141
771
85
85
7.225
51
22
9
17
51
167.49
143
167.49
1098
42
55
3.543
20
5.65
1.82
3.82
15.11
31.41
10.57
20.84
270
25
22
2.352
15
4.79
2.01
3.76
12.85
37.52
17.20
33.03
285
292
64
69
873
205
233
160
34
10.65
402
141
20.59
9.39
438
624
107
126
1354
295
326
1334
313
291
616
85
46
1978
age in our analysis. We also allowed for the possibilitythat

the effects of industrytenure depend on organizationalsize
(Carrolland Hannan,2000) by estimating separate tenure
effects for large and small organizations.
RESULTS
Tables3a and 3b show estimates of baseline failurerate
models that includethe various controlvariablesand different
specifications of the density of organizations.Forcomparison, model 1 includes the controlvariablesbut no density
variables. Model 2 then includes density in the year of an
organization'sfoundingand a quadraticspecification of contemporaneous density. None of these density effects are statisticallysignificant,nor in model 3 does density have a statisticallysignificanteffect when specified without the
squared and density-at-birthterms. In model 4, however,
strong competitive results appear when worldwide density is
disaggregated to distinguishthe effects of domestic and foreign rivals,as well as different effects for domestic rivalsthat
overlap in productspace and those that do not. These disag552/ASQ, December 2004
gregated density terms show that competition has been

strong, but localized in the disk drive market, both geographically and technologically.Accordingto model 4, an organization faces strong competition from domestic rivalsthat have
products in the same form factor and capacity level, while
non-overlappingdomestic rivalshave a much smaller competitive effect. Foreignrivals,meanwhile, actuallyhave a mutualistic effect, lowering the failurerates of other organizations.
This pattern is consistent with models that reveal geographically localizedcompetition together with life-enhancinglegitimacy effects coming from increases in numbers of organizations over broadergeographic areas (Carrolland Hannan,
2000).
The models in tables 4a and 4b investigate Red Queen competition. Model 5 includes the effect of a firm's competitive
experience on its own exit rate, as well as the effect of its
rivals'competitive experience. Model 6 then distinguishes
between these experience effects accordingto whether
competition was experienced when an organizationwas
small or when it was large and also allows the effects of
domestic density to vary by the size of rivals.Supporting
hypothesis 1, there is strong evidence of Red Queen evoluTable3a
Baseline Models of Competition: Market Exit Rates among Disk Drive Manufacturers Worldwide, 1956-1998*
Model
Independent Variable
Numberof low-capacityform factors producedby organization
Numberof medium-and high-capacityform factors producedby
organization
Hard-drive
sales ($mil)by organization(largeonly)
(1)
(2)
(3)
(4)
-.3855
(.2347)
-.3456
(.2389)
-.27140
(.1579)
-.0032*
(.0018)
.2260
(.2106)
.3037
(.2635)
.1675
(.2875)
-.0010
(.0010)
.0015
(.0024)
-.0015
(.0065)
.0317
(.0357)
-.2617
(.3204)
-.3621
(.2385)
-.2700*
(.1573)
-.0031'
(.0017)
.2330
(.2093)
.3094
(.2602)
.1659
(.2766)
-.0006
(.0008)
.0009
(.0022)
-.4188*
(.2384)
-.2286
(.1599)
-.0032*
(.0018)
.1977
(.2070)
.3909
(.2681)
.3116
(.2830)
-.0001
(.0008)
.0009
(.0023)
-.2751*
(.1570)
-.0031*
Organizationwas founded de novo

Organizationhas captiveproduction
Organizationmanufacturesin Asia
Numberof computer manufacturersin the form factors producedby
the organization
Entriesof computer manufacturersin the form factors producedby the
organization
Numberof rivalsworldwidein year of organization'sfounding
(.0017)
.2265
(.2089)
.3010
(.2607)
.1507
(.2751)
-.0003
(.0006)
.0007
(.0022)
Numberof rivalsworldwide
(Numberof rivalsworldwide)2
.0027
(.0048)

Log likelihood
Degrees of freedom
0
p < .10;
-106.20
31
p < .05.
-105.69
34
-106.04
32
.0686"
(.0237)
.0318"
(.0123)
-.0215"
(.0085)
-98.81
34
* Standarderrorsare in parentheses. Each model also includes markettenure, calendar

year, form factor,and region
effects, as shown in table 3b. The data cover 1,538 organization-years,171 organizations,and 157 exits.
Table3b
Estimated Market Tenure, Calendar Year,Form Factor, and Region Effects from the Models in Table 3a*
Model
Organizationof markettenure 0-1 year
Organizationof markettenure 1-3 years
Organizationof markettenure 20+ years
Largeorganizationof markettenure 0-3 years
Largeorganizationof markettenure 20+ years
Japanese organization
Eastern-European
organization
Western-Europeanorganization
South Americanorganization
Asian (otherthan Japanese) organization
Organizationproduces 1.8-inchform factor
Organizationproduces 8-inchform factor
Organizationproduces 14-inch(orabove) form factor
0
(1)
(2)
-5.168"
(.7406)
-3.370"
(.5797)
-3.165"
(.5965)
-3.034"
(.6226)
-2.679"
(.6633)
-2.398"
(.7821)
-2.225"
(.7384)
-2.768"
(1.027)
-1.830"
(.5087)
-1.634"
(.4889)
-1.069
(.9609)
.0609"
(.0177)
-.4071
(.2752)
-.9793
(.6235)
-.3682
(.2954)
-.9883"
(.3528)
.0403
(.3932)
-.3618
(.6586)
-.1377
(.5648)
.6095
(.5354)
.4692
(.5220)
.2491
(.4265)
.1017
(.3751)
-5.860"
(1.065)
-4.039"
(.9425)
-3.824"
(.9415)
-3.687"
(.9606)
-3.349"
(1.005)
-3.184"
(1.188)
-2.305"
(.7446)
-2.824"
(1.030)
-1.899"
(.5152)
-1.692"
(.4933)
-.9682
(.9704)
.0584"
(.0216)
-.4200
(.2760)
-1.026*
(.6291)
-.3648
(.3037)
-1.001"
(.3561)
.0435
(.3958)
-.1215
(.7020)
.0998
(.6112)
.9048
(.6083)
.7520
(.5943)
.2933
(.4338)
.1558
(.3823)
(3)
(4)
-5.297" -6.110"
(.7805) (.8324)
-3.488" -4.280"
(.6229) (.6831)
-3.283" -4.039"
(.6384) (.6948)
-3.138" -3.875"
(.6560) (.7050)
-2.789" -3.477"
(.6987) (.7477)
-2.529" -3.095"
(.8217) (.8804)
-2.261" -2.380"
(.7414) (.7418)
-2.799" -2.968"
(1.029) (1.033)
-1.860" -1.947"
(.5118) (.5133)
-1.653" -1.659"
(.4899) (.4903)
-1.031 -1.250
(.9625) (.9574)
.0595" .0684"
(.0180) (.0192)
-.4085
.5735
(.2757) (.4122)
-1.001
.9475
(.6244) (.8334)
-.3494 1.109"
(.2972) (.4934)
-.9857" .5866
(.3523) (.5534)
.0523 1.766"
(.3937) (.5975)
-.2710 -.6952
(.6775) (.7045)
-.0525 -.2470
(.5846) (.5971)
.6988
.2575
(.5571) (.6036)
.5298
.1573
(.5318) (.5674)
.2441
.0834
(.4269) (.4278)
.1192 -.0446
(.3765) (.3859)
"
p < .10;
p < .05.
* Standarderrorsare in parentheses. Forregioneffects, the U.S. is the omitted category
tion revealed in these models, but only due to competition

experienced by organizationswhen they are small. We do
not find differences in failurerates due to experience in other
industries(the de novo/de alio comparison).Conditionalon
survival,firms that faced more competition in the past are
less likelyto fail, but only if they were small when they experienced this priorcompetition.
These effects are described in figure 3, based on the estimates from model 14, the preferredspecification from the
most complete models. Figure3 shows the combined
effects of an organization'smarkettenure, size, and competi554/ASQ, December 2004
tive experience on its own exit rate, controllingfor all the

other variablesin model 14. The solid lines show how the
exit rate changed with markettenure for organizationsthat
faced no domestic competition, as would be the case, for
instance, if an organizationwere technologicallydifferentiated
from its domestic rivals.Among these "monopolists," exit
rates increased with markettenure for both large and small
firms. Thus organizationsthat faced no competition suffered
a powerful liabilityof aging. Among small firms, however,
another patternwas possible, as indicatedby the dotted line.
Small firms that experienced the mean level of rivalry
Table4a
Models of Red Queen Competition: Market Exit Rates among Disk Drive Manufacturers Worldwide,
1956-1998*
Model
Numberof low-capacityform factors producedby organization
Numberof medium-and high-capacityform factors producedby
organization
Harddrivesales ($mil)by organization(largeonly)
Organizationhas captive production
Numberof computer manufacturersin the form factors producedby
the organization
Entriesof computer manufacturersin the form factors producedby
the organization
Organization'scompetitiveexperience
(5)
(6)
(7)
(8)
-.3791
(.2364)
-.2204
(.1594)
-.0036"
(.0018)
.1641
(.2090)
.3176
(.2684)
.3642
(.2839)
.0004
(.0009)
.0003
(.0024)
-.0063?
(.0035)
-.4252*
(.2383)
-.2014
(.1621)
-.00370
(.0019)
.0641
(.2149)
.2279
(.2709)
.1519
(.3202)
-.0001
(.0009)
.0008
(.0023)
-.4029?
(.2391)
-.2326
(.1606)
-.0035*
(.0018)
.0881
(.2132)
.2189
(.2727)
.1400
(.3149)
-.0001
(.0009)
.0006
(.0023)
-.4256*
(.2386)
-.1959
(.1623)
-.0037"
(.0019)
.0718
(.2154)
.2230
(.2709)
.1493
(.3194)
-.0003
(.0009)
.0009
(.0023)
-.0116"
(.0050)
-.0029
(.0043)
-.0127* -.0121m
(.0050) (.0050)
-.0023 -.0032
(.0042) (.0043)
.0291
(.0657)
.1199"
(.0468)
-.2397 -.2048
(.2278) (.2279)
.4259" .1753
(.1623) (.2284)
.0041*
(.0023)
-.0018"
(.0008)
.0031
(.0026)
-.0018
(.0014)
.0107
.0092
(.0080) (.0084)
-.0126" -.0018
(.0058) (.0093)
.0138
.0221
(.0140) (.0147)
-.0182" -.0196".
(.0087) (.0089)
-91.67 -90.46
39
41
Largeorganization'scompetitiveexperience
.1143"
(.0401)
Overlapwith small domestic rivals

Overlapwith large domestic rivals(by form factor and capacity)
Domestic rivals'competitiveexperience (experience-weightedoverlap)
Smalldomestic rivals'competitiveexperience (experience-weighted
overlap)
Largedomestic rivals'competitiveexperience (experienceweighted overlap)
Overlapwith small domestic rivalsx calendaryear (1956 = 0)
-.0009
(.0007)
Overlapwith large domestic rivalsx calendaryear (1956 = 0)

.0312"
.0239?
(.0123)
(.0136)
-.0210"
-.0187"
(.0085)
(.0086)
-95.76
-91.07
36
39
Log likelihood
Degrees of freedom
"
p < .10; p < .05.
* Standarderrorsare in parentheses. Each model also includes markettenure, calendar
year, form factor,and region
effects as shown in table 4b. The data cover 1,538 organization-years,171 organizations,and 157 exits.
Table4b
Estimated MarketTenure, Calendar Year,Form Factor, and Region Effects from the Models in Table 4a*
Model
Eastern-European
organization
Organizationproduces 8-inchform factor
Organizationproduces 14-inch(orabove) form factor
(5)
(6)
-6.593"
(.8721)
-4.670"
(.7205)
-4.339"
(.7251)
-4.030"
(.7334)
-3.466"
(.7861)
-2.955"
(.9516)
-2.481"
(.7445)
-3.018"
(1.034)
-1.976"
(.5115)
-1.686"
(.4947)
-1.175
(.9794)
.0890"
(.0217)
.4072
(.4224)
.3456
(.8778)
.6799
(.5276)
.0174
(.6044)
1.347"
(.6234)
-1.006
(.7201)
-.5487
(.6121)
.0908
(.6125)
-.1011
(.5805)
.0105
(.4255)
-.0861
(.3918)
-6.442"
(.8962)
-4.431 "
(.7518)
-4.076"
(.7579)
-3.804"
(.7640)
-3.245"
(.8112)
-2.934"
(.9590)
-2.526"
(.7457)
-3.081 "
(1.034)
-2.150"
(.5149)
-1.766"
(.5027)
-.8907
(1.008)
.0935"
(.0218)
.0758
(.4511)
.1446
(.8933)
.3276
(.5539)
-.2328
(.6146)
1.081*
(.6381)
-.9478
(.7160)
-.2898
(.6178)
.2963
(.6146)
.1106
(.5821)
.0494
(.4375)
-.2016
(.3991)
" < .05.

<
p.10;S
p
* Standarderrorsare in parentheses. Forregioneffects, the U.S. is the omitted category.
(7)
(8)
-5.989" -6.236"
(.8757) (.9012)
-3.986" -4.196"
(.7379) (.7624)
-3.625" -3.835"
(.7459) (.7697)
-3.333" -3.551 "
(.7544) (.7791)
-2.771 " -2.979"
(.8053) (.8276)
-2.466" -2.611"
(.9734) (.9840)
-2.509" -2.571"
(.7438) (.7467)
-3.101 " -3.120"
(1.034) (1.034)
-2.190" -2.192"
(.5170) (.5171)
-1.874" -1.829"
(.5030) (.5048)
-.9360
-1.082
(.9974) (1.013)
.0866" .0881"
(.0216) (.0220)
-.0722 -.0230
(.4571) (.4621)
-.0188
-1.563
(.8943) (.9046)
.2759
.0881
(.5651) (.5802)
-.5748 -.4698
(.6516) (.6697)
.6999
.9871
(.6681) (.6935)
-.9181 -.9419
(.7073) (.7168)
-.3212 -.2097
(.6123) (.6237)
.4129
.4737
(.6196) (.6336)
.2043
.1261
(.5772) (.5879)
-.0201
.0997
(.4385) (.4443)
-.1409 -.0676
(.4252) (.4307)
observed for a given tenure period saw their exit rates initially increase with markettenure but ultimatelyfall as their
competitive-experienceeffect dominated. Especiallyinteresting is the comparisonwith large organizations.Small firms
that survive competition ultimatelyend up less likelyto fail
than their large counterparts,as the consequences of Red
Queen evolution ultimatelymore than offset the liabilityof
smallness.
Turningto the rivalryside of the Red Queen model, a comparisonof models 5 and 6 shows stronger competition from
rivalsthat experienced more competition in the past, but

Figure 3. Effects of age and competitive experience on the exit rate (based on the estimates in model 14).
0.14
Small Monopolist
0.12
2,
0.1
x
" 0.08
rS0.06
Large Monopolist
------------------
7-----
S0.04
Small Firm With Mean Number of Rivals per Year

0.02
0
5
10
15
20
25
35
30
40
45
MarketTenure
Figure 4. Rivals' effects on an organization's exit rate (based on the estimates in model 14).
3.5
3-
2 2.5-
Effect of a Small Rival
CX
Co
_. 1.5
Baseline Rate
0.5
Effect of a Large Rival
25
50
75
100
125
150
175
200
225
250
275
300
Rivals' Competitive Experience
again, this holds only if the priorcompetition was experienced while the rivalwas small, in support of hypothesis 2.
Among rivalswith no competitive experience, however, we
only find evidence of competition from large rivals. Figure4
illustratesthe implicationsof these rivalryeffects combined.
The figure shows how the competitive intensity of an organization's rivalsvariedaccordingto the rivals'priorexposure to
competition (againfrom model 14). The plot combines the
effect of density, the overlapwith domestic rivals,and the
December 2004
557/ASQO,
effect of the competitive experience of these rivalson an

organization'sexit rate. Among rivalswith no competitive
experience, only the density effect operates, so large rivals
generated stronger competition than did small rivals.Having
to compete against one large firmthat was previouslya
monopolist raised an organization'sfailurerate by about 24
percent, while competing against a small firm that had no
competitive experience generated no competition that we
could detect. This pattern reverses with competitive experience, however, so that an organization'sfailurerate increased
as its (small)rivals'competitive experience increased. This
was a strong effect, implyingincreases in the failurerate of
up to 328 percent over the observed range of this variable;
by far the strongest competition in the market came from
rivalsthat had survived competition when they were small.
Ourfindings also show an unexpected but slight decrease in
the competition generated by large rivalsas they experienced
more competition.
Figure5 illustratesthe trade-off impliedby these results.
Everyplotted point in this figure represents an organizationyear observation in the data. The points are positioned to
show observed pairingsof the competitive experience of
each organizationas of each year (on the verticalaxis) versus
the observed competitive experience of the organization's
rivalsin that year (on the horizontalaxis). All experience
effects here are limitedto those accumulated when these
organizationswere small. The diagonalline indicates the
threshold of equalityat which the advantages of experiencing
competition are exactly offset by the competitive strength of
one's rivalsbecause of their experience (accordingto the
Figure 5. Organization's vs. rivals' competitive experience in the disk drive market (line indicates threshold of
equality between organization's and rivals' strength).
160
140
*
AD
S-120
120
0.
X
LU
a 100
E 80-
cc**
S*.N
ca
40
0
Organization'sStrength Dominates
Rivals' Strength Dgminates

*
--
0
0
50
100
150
200
Rivals' Competitive Experience

250
300
estimates of model 14). Forobservations on this line, then,

the advantages and disadvantages of Red Queen competition
exactly offset each other. Organizationsabove this threshold
are more likelyto survive due to Red Queen competition, and
organizationsbelow the threshold face a greater net hazard
due to Red Queen competition.
Models 7 and 8 investigate whether the increase in domestic
rivalrythat we are attributingto Red Queen competition
might, instead, be due simply to increasing competitiveness
among all organizationsover calendartime, instead of
increasingcompetitiveness due to priorexposure to competition. Model 7 suggests that competitive intensity among
domestic rivalsdid increase with calendartime, but this
effect vanishes once the Red Queen effect is included in
model 8. Neither is the Red Queen effect significant in model
8, but this appears to be due to colinearity,as the coefficients of the rivals'competitive experience terms fall only
slightlyfrom model 6 to model 8, but the standarderrors
increase dramatically.By comparison, the calendartime x
domestic rivalryinteractionterm sees its coefficient fall by an
order of magnitude between models 7 and 8. Thus it appears
that domestic rivalryincreased due to exposure to competition, ratherthan simply with the passage of calendartime.
A different story emerges for foreign competition. Tables 5a
and 5b investigate competition from foreign rivalsin greater
detail. Model 9 is similarto model 8, except that the density
of rivalsfrom other regions of the world is specified as a
technology-overlapdensity, and the domestic competitive
experience of foreign rivalsis included in orderto test
hypothesis 3. Model 10 is the same as model 9, except that
it also includes interactionsbetween foreign rivals(small and
large)and historicaltime to test for the alternativeto hypothesis 3, that increasingglobal competition occurred marketwide over time regardless of exposure to competition. A
comparison of models 9 and 10 clearlyfails to support
hypothesis 3. Globalcompetition increased in strength over
historicaltime-generated by small and large rivalsalikeand once these effects are included,we fail to find evidence
that exposure to domestic competition made organizations
stronger global competitors. Models 11 through 14 confirm
this conclusion, showing the same patternof results in specifications that also demonstrate that the strength of global
competition did not appear to hinge on whether rivalswere
large or small.
Although we fail to find support for hypothesis 3, we do find
strong evidence that competition became increasingly global
in the hard disk drive market as calendar time passed. Using
the most parsimonious specification of global competition,
model 14, figure 6 illustrates the way that competition
became more global over time in the hard disk drive market.
The plot reflects the negative effect of technology overlap
with foreign rivals, together with the positive effect of technology overlap interacted with calendar time. In the early
days, the existence of foreign rivals made a firm more likely
to survive, lowering failure rates by about 30 percent per
rival, for instance, back in 1965. This survival-enhancing
effect is consistent with the argument that geographically
Table5a
Models of Global Competition: Market Exit Rates among Disk Drive Manufacturers Worldwide, 1956-1998*
Model
(9)
(10)
(11)
(12)
(13)
(14)
Numberof low-capacityform factors producedby

-.3114
-.3762
-.2982
-.3720
-.3705
-.3724
(.2368) (.2423) (.2343) (.2380) (.2364)
(.2364)
organization
Numberof medium-and high-capacityform factors -.2066
-.2201
-.1943
-.1916
-.1881
-.1879
(.1651) (.1654) (.1616) (.1623) (.1639)
(.1620)
producedby organizations
Harddrivesales ($mil)by organization
-.00360 -.0037" -.0035* -.0037" -.0037" -.0037"
(.0019) (.0019) (.0018) (.0018) (.0018)
(.0018)
.1057
.1381
.1047
.1378
.1371
.1371
(.2160) (.2153) (.2160) (.2154) (.2152)
(.2154)
.2436
.2469
.3188
.3180
.3192
.3180
Organizationhas captive production
(.2659) (.2689) (.2659) (.2686) (.2686)
(.2687)
.1167
.2181
.1304
.2156
.2182
.2159
(.3236) (.3216) (.3206) (.3204) (.3192)
(.3197)
Numberof computermanufacturersin the form
-.0017* -.0019* -.0015* -.0019" -.0019" -.0019"
factors producedby the organization
(.0009) (.0010) (.0009) (.0009) (.0009)
(.0009)
Entriesof computermanufacturersin the form
.0030
.0024
.0028
.0024
.0025
.0024
factors producedby the organization
(.0024) (.0024) (.0023) (.0023) (.0023)
(.0023)
-.0108" -.0119" -.0108" -.0119" -.0119" -.0119"
(.0051) (.0051) (.0051) (.0051) (.0051)
(.0051)
-.0023
-.0022
-.0036
-.0035
-.0035
-.0035
Largeorganization'scompetitiveexperience
(.0043) (.0044) (.0043) (.0044) (.0044)
(.0044)
.0334
.0325
.0483
.0479
.0481
.0479
Overlapwith small domestic rivals(by form factor
and capacity)
(.0667) (.0689) (.0665) (.0689) (.0688)
(.0689)
.1451
.2071" .1424"
.2127"
.2080"
.2128"
Overlapwith largedomestic rivals(by form factor
and capacity)
(.0506) (.0600) (.0501) (.0565) (.0587)
(.0560)
Smalldomestic rivals'competitiveexperience
.0056"
.0040'
.00390
.0055" .00390
.0039?
(.0023) (.0023) (.0022) (.0023) (.0023)
(.0023)
(experience-weightedoverlap)
-.0016" -.0025" -.0016" -.0025" -.0025W -.0025"
Largedomestic rivals'competitiveexperience
(.0008) (.0009) (.0008) (.0009) (.0009)
(.0009)
.02200
.0188
.0199
.0192
.0198
.0228?
(.0130) (.0138) (.0129) (.0132) (.0131)
(.0129)
-.0988" -.4904" -.0938" -.4786" -.4831" -.4795"
Overlapwith foreignrivals
(.0364) (.1455) (.0338) (.1398) (.1272)
(.1266)
.0019"
.0001
Foreignrivals'competitiveexperience (experience(.0008) (.0010)
weighted overlap)
Smallforeign rivals'competitiveexperience
.0025
.0001
(.0018) (.0023)
.0017" -.0001
Largeforeign rivals'competitiveexperience
(.0008) (.0013)
.0150"
.0150"
Overlapwith foreignrivalsx calendaryear (1956 =
0)
(.0052)
(.0041)
.0152"
.0149"
Overlapwith small foreign rivalsx calendaryear
(1956 = 0)
(.0053)
(.0041)
.0181"
.0154"
Overlapwith largeforeign rivalsx calendaryear
(1956 = 0)
(.0059)
(.0043)
-89.26
-84.89
-89.34
-84.93
-84.89
-84.93
Log likelihood
41
43
41
40
40
40
Degrees of freedom
0
p < .10;
p < .05.
* Standarderrorsare in parentheses. Each model also includes markettenure, calendaryear, form factor,and region
effects as shown in table 5b. The data cover 1,538 organization-years,171 organizations,and 157 exits.
separated organizationscontributeto legitimatingone another even as they are too distant to generate noticeable competition. Over time, however, the patternchanges, with foreign rivalsgenerating increasingcompetition. By the end of
the 1980s, organizationsare actuallydrivingup the failure
rates of rivalsin other countries. And by the end of the study
period,the competitive effect of foreign rivalshas grown to
be almost indistinguishablefrom that generated by domestic
rivals.This can be seen by comparingthe far rightendpoint
of the plot in figure 6, which illustratesrecent foreign compe560/ASQ, December 2004
Estimated Market Tenure, Calendar Year,Form Factor, and Region Effects from the Models in Table 5a*
Model
(9)
(10)
(11)
(12)
-6.024" -5.725" -5.997" -5.764"

(.9018) (.8715) (.8981) (.8631)
-4.002" -3.682" -3.980" -3.719"
(.7573) (.7186) (.7538) (.7105)
-3.662" -3.332" -3.646" -3.376"
(.7608) (.7284) (.7582) (.7160)
-3.417" -3.058" -3.384" -3.106"
(.7683) (.7356) (.7625) (.7190)
-2.869" -2.466" -2.855" -2.506"
(.8072) (.7721) (.8057) (.7622)
-2.581 " -2.127" -2.592" -2.165"
(.9314) (.9144) (.9287) (.9051)
-2.483" -2.528" -2.472" -2.532"
(.7464) (.7488) (.7458) (.7481)
-3.075" -3.065" -3.068" -3.062"
(1.033) (1.035) (1.033) (1.035)
-2.154" -2.089" -2.168" -2.081 "
(.5170)
(.5197) (.5161) (.5186)
-1.826"
Largeorganizationof markettenure 10-20 years -1.833"
(.5002) -1.713"=
(.5020) (.4993) -1.715"=
(.5018)
-.8838
-.8665
-.8604
-.8733
(1.002)
(1.010)
(.9972) (1.009)
.0608"
.0467" .0611"
.0470"
(.0237) (.0232) (.0236) (.0232)
-.3222
-.2302
-.3425
-.2111
(.4032)
(.4094) (.4003) (.4035)
-.7030
-.5290
-.7365
-.5112
Eastern-European
organization
(.7844) (.7971) (.7802) (.7927)
-.1962
-.2275
-.2463
-.1759
(.5410) (.5600) (.5270) (.5275)
-1.013
-.7999
-.7334
(.6288) (.6659) -1.050?
(.6221) (.6256)
.4281
.3429
.3826
.3886
(.6745) (.6791) (.6635) (.6574)
-.0243
.0274 -.0352
.0365
(.7302) (.7382) (.7291) (.7369)
.1450
.4605
.0616
.4958
(.6488) (.6655) (.6135) (.6342)
.6689
.5907
.6100
.6388
(.6039) (.6286) (.5855) (.5957)
.8339
1.080*
.7630
1.1110
(.6014) (.6234) (.5758) (.5972)
.1178
.4550
.0831
Organizationproduces 8-inchformfactor
.4667
(.4491) (.4688) (.4398) (.4642)
.4609 -.0175
Organizationproduces 14-inch(orabove) form factor .0028
.4647
(.4056) (.4345) (.4019) (.4339)
"
Sp < .10; p < .05.
* Standarderrorsare in
parentheses. Forregioneffects, the U.S. is the omitted category.
(13)
-5.726"
(.8720)
-3.684"
(.7191)
-3.338"
(.7274)
-3.061 "
(.7355)
-2.471 "
(.7712)
-2.138"
(.9080)
-2.524"
(.7479)
-3.061 "
(1.034)
-2.091"
(.5195)
(14)
-1.714"=
(.5004)
-.8655
(1.007)
.0464"
(.0230)
-.2314
(.4094)
-.5327
(.7961)
-.2292
(.5607)
-.8063
(.6508)
.3352
(.6755)
.0408
(.7220)
.4504
(.6540)
.5889
(.6245)
-5.765"
(.8618)
-3.719"
(.7095)
-3.377"
(.7154)
-3.106"
(.7184)
-2.506"
(.7620)
-2.165"
(.9052)
-2.533"
(.7474)
-3.062"
(1.034)
-2.081 "
(.5182)
-1.715"*
(.5009)
-.8730
(1.009)
.0470"
(.0229)
-.2106
(.4019)
-.5106
(.7916)
-.1748
(.5220)
-.7301
(.5856)
.3907
(.6433)
.0342
(.7211)
.4969
(.6301)
.6394
(.5945)
1.078?
(.6106)
.4459
(.4619)
.4535
(.4292)
1.111?
(.5970)
.4681
(.4547)
.4659
(.4272)
tition, to the y-intercepts in figure 4, which illustratethe

strength of domestic competition. These results show that
by the end of the 1990s, competition in the harddisk drive
market had become global.
Lookingback over tables 3b, 4b, and 5b, it is interesting to
see how the effects of nationalregions change across specifications. These nationalregion effects change dramatically
from model 3 to model 4, as domestic and foreign competition are specified separately. Nationaleffects continue to
change as we elaborate our model by includingthe effects of
exposure to competition in table 4. Once we fully specify the
Figure 6. The development of global competition among disk drive manufacturers (based on the estimates of
model 14).
1.2
1.1
Baseline Rate
Effect of One Additional Foreign Rival
0.9
0.8
0.7
0.6
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
Year
model in table 5, allowing for the globalizationof competition,

we no longer find statistical evidence of nationalregion
effects. Neither do we find evidence that manufacturingin
Asia affected failurerates. Apparently,at least to the extent
that our models can detect them, nationaldifferences in
organizationalfailurerates in the harddisk drive market
appearto be due to differences in the competitive landscape.
Fullyspecifying the competitive landscape also was important to obtainingunbiased estimates of productstrategy
effects. In particular,some of the models in tables 3a and 4a
show marginallysignificantdifferences in the failurerate
depending on the capacity levels of firms' productofferings.
These differences no longer appear once the various competitive effects are fully specified in table 5a. Meanwhile, we initiallysee no apparentdifferences in failurerates due to firms'
choices of form factors (in tables 3b and 4b), but significantly
greater failurerates appear for the 5.25-inch form factor once
competition is fully modeled in table 5b. Inthese ways, we
were not able to correctlyrecognize the survivalimplications
of productstrategies untilwe fully modeled the different
sources of competition to which organizationswere exposed
due to these strategies.
Finally,we find strong community ecology effects. On the
one hand, we see clear symbiosis between the populationof
computer manufacturersand disk drive manufacturers:failure
rates fall among harddisk drive firms with increases in the
populationof computer firms for whom they manufactured.
But we do not find that this symbiosis was "internalized"by
organizationsthat verticallyintegrated between the harddisk
drive and computer industries, in that captive producersdid
not enjoy greater survivalchances; in fact, the "captive"variable has a positive, though non-significanteffect on the fail562/ASQ, December 2004
ure rate. It appears, then, that evidence of symbiosis

between computer manufacturingand disk drive manufacturing could be found only at the populationlevel.
DISCUSSIONAND CONCLUSION
Why are some organizationsmore competitive than others?
As we have demonstrated, the answer to this question
hinges on the logic one chooses to guide the study of competition. Understood as a contest, competition in the disk
drive industryhas clearlyfavored organizationswith the ability to orchestrate continuous productdevelopment over time,
in short, the largest harddisk drive firms that remained disproportionatelyat the cutting edge of technology. By contrast, we come up with the opposite answer when competition is understood as a powerful source of constraintthat
drives organizationaldevelopment and selection. Guided by
this logic, we found that organizationsthat have been
exposed to competition are less likelyto fail and that they
generate stronger competition-so-called Red Queen competition-but that these benefits accrue only when organizations are small. Largeorganizations,well known to have
advantages that help them cope with competitive constraints, appear to be unresponsive to the Red Queen
process.
Regardingglobal competition, organizationsdo not appear to
projectstronger competition globallyas a result of their
domestic competitive experience. Instead, global competition
among disk drive manufacturersincreased over calendar
time, with all organizationsgenerating increasinglystrong
global competition. Many scholars, but most notably Porter
(1990), have emphasized the continuingimportanceof homecountryeffects even as firms engage in global competition.
Ourfindings suggest, however, that at least in the harddisk
drive market,competition became increasinglyglobal over
calendartime regardless of the competitiveness of one's
home country.Consequently, country-specificadvantages
have faded in importanceas time has passed. Similarly,
although many have noted country-specificdifferences in failure rates among disk drive manufacturers(e.g., Chesbrough,
1999), we found that these differences vanish when specified in models that allow for Red Queen development and
global competition.
Ourfindings suggest that neither development nor obsolescence is built into the process of organizationsaging. Rather,
whether and how organizations develop over time hinges on
whether they are exposed to competition. Except for large
organizations, those that have endured competition are likely
to be better adapted and may even appear to have moved
down a learning curve. By contrast, organizations that have
remained isolated from competition will not have been
spurred by Red Queen evolution and consequently are more
likely to exhibit liabilities of senescence and obsolescence as
their ages increase (Barron, West, and Hannan, 1994). In neither case, however, are these the inevitable outcomes of
aging. We found instead that change in survival chances over
time depends on whether organizations have been exposed
to competition.
Our results also reveal a possible unintended consequence of

technological differentiation.A large literaturein strategic
management emphasizes the fact that organizationsperform
better and have better life chances when they are differentiated on various dimensions. Organizationalecology research,
too, demonstrates that competition is localized on various
dimensions (McPherson, 1983; Hannanand Freeman, 1989;
Baum and Mezias, 1992; Baum and Singh, 1994; Podolny,
Stuart,and Hannan,1996; Carrolland Hannan,2000). Our
findings suggest that organizationspay a hidden cost for
such differentiation.Isolationfrom competition has obvious
current-timebenefits but also has the less-obvious downside
that it deprives an organizationof the engine of development.
The more that change is driven by Red Queen evolution, the
more will there be a trade-offbetween positionaladvantage
that isolates an organizationfrom competition and the advantages that come from the development of new capabilities.
Inthis way, positional-and capabilities-basedadvantage may
be inversely related, with those in safe places enjoying their
positions but suffering over time as they fall behind those
who remain in the race.
These results speak, as well, to the growing literatureon
"coevolution"among organizations.Earlyattempts at linking
organizationsto their environment in a system of reciprocal
causalitydid not advance beyond speculative frameworks
(Emeryand Trist,1965; Terreberry,1968). More recently, reciprocalcausality has been addressed insofaras organizations
refer to one another in a process of social comparison
(White, 1981) or in strategies of imitation(Mezias and Lant,
1994). Some research describes such reciprocalprocesses as
"coevolution,"with reference by analogy to developments in
biology (e.g., Baum and Singh, 1994). The strength of
research in this vein is that we now know of various regulatory,technical, and organizationalprocesses that run in parallel
and that affect one another as they develop over time (see
various papers in Baum and McKelvey,1999, for example).
The downside, thus far, is that research on coevolution often
fails to produce falsifiablehypotheses. Our model, though it
features reciprocalcausality,is sufficientlyexplicit that it
makes falsifiablepredictions.We hope that other work in the
general area of coevolution will take a similarlyexplicit modeling approach.
To conclude, we think it is useful to consider the dynamic
implicationsof the prevailingsociological view of competition
as a constrainingforce. Seen as partof an ongoing, coevolutionarydynamic, such competition may be managed by large,
formidable organizations, but this may not work to their
advantage. Intentions notwithstanding, the ultimate consequences of organizations' efforts to strategize around, directly manage, or otherwise control the incidence and impact of
competition may be to hinder their own development, or at
least to allow organizations to survive that would otherwise
have been selected out. Informed by this possibility, our
analyses of competition will be able to reveal this potentially
important source of competitive weakness among our most
formidable organizations.

REFERENCES
Abernathy, F. H., J. T. Dunlop,
J. H. Hammond, and D. Weil
1999 A Stitch in Time:Lean Retailof
ing and the Transformation
Manufacturing-Lessons
from the Appareland Textile
Industries.New York:Oxford
UniversityPress.
Abernathy, F. H., and J. Utterback
1978 "Patternsof industrialinnovation." TechnologyReview,80
40-47.
Aldrich, H. E.
1999 OrganizationsEvolving.Thousand Oaks, CA:Sage.
Anand, J., and B. Kogut
1997 "Technologicalcapabilitiesof
countries, firm rivalryand foreign directinvestment."Journal of InternationalBusiness
Studies, 28: 445-465.
Anderson, P.
1995 "Microcomputermanufacturers." In G. R. Carrolland
M. T. Hannan(eds.), Organizations in Industry:Strategy,
Structureand Selection:
37-58. New York:OxfordUniversity Press.
Banbury,C. M., and W. Mitchell
1995 "Theeffect of introducing
importantincrementalinnovations on marketshare and
business survival."Strategic
ManagementJournal,16
(Summer):S161-S182.
Barnard,C. I.
1948 The Functionsof the Executive. Cambridge,MA:Harvard
UniversityPress.
Barnett,W. R
1997 "Thedynamicsof competitive
intensity."AdministrativeScience Quarterly,42: 128-160.
Barnett, W. R, A. N. Swanson,
and 0. Sorenson
2003 "Asymmetricselection
among organizations."Industrialand CorporateChange,
12: 673-695.
Barney, J. B., and E. J. Zajac
1994 "Competitiveorganizational
behavior:Towardan organizationally-basedtheory of competitive advantage."Strategic
ManagementJournal,15:
5-9.
Barron, D. N.
1999 "Thestructuringof organizationalpopulations."American
SociologicalReview, 64:
421-445.
Barron, D. N., E. West, and M. T.
Hannan
1994 "Atime to grow and a time
to die: Growthand mortality
of credit unions in New York
City,1914-1990." American
Journalof Sociology, 100:
381-421.
Baum, J. A. C., and W. McKelvey
(eds.)
1999 Variationsin Organizational
Science. ThousandOaks, CA:
Sage.
Baum, J. A. C., and S. J. Mezias
1992 "Localizedcompetitionand
organizationalfailurein the
Manhattanhotel industry,
1898-1990." Administrative
Science Quarterly,37:
580-604.
Baum, J. A. C., and J. V. Singh
niches and
1994 "Organizational
the dynamicsof organizational mortality."AmericanJournal of Sociology, 100:
346-380.
Barnett,W. P., H. R. Greve, and

D. Y. Park
1994 "Anevolutionarymodel of
organizationalperformance."
StrategicManagementJournal, 15: 11-28.
Bendix, R.
1956 Workand Authorityin Industry:Ideologies of Management in the Course of Industrialization.Berkeley,CA:
Universityof CaliforniaPress.
Barnett,W. P., and M. T. Hansen

1996 "The Red Queen in organizationalevolution."Strategic
139-157.
Blau, P. M.
1977 Inequalityand Heterogeneity:
A PrimitiveTheoryof Social
Structure.New York:Free
Press.
Barnett,W. P., and 0. Sorenson

2002 "The Red Queen in organizationalcreationand development." Industrialand Corporate Change, 11: 289-325.
Bothner, M. S.
2003 "Competitionand social influence: The diffusionof the
sixth-generationprocessor in
the globalcomputer industry."AmericanJournalof
Sociology, 108: 1175-1210.
Burt, R. S.
1987 "Socialcontagion and innovation: Cohesion versus structuralequivalence."American
1287-1335.
1992 StructuralHoles: The Social
Structureof Competition.
Cambridge,MA:HarvardUniversity Press.
Carroll,G. R., L. Bigelow, M. D.
Seidel, and L. Tsai
1996 "Thefates of de novo and de
alio producersin the American automobileindustry,
1885-1982." Strategic Management Journal,17:
117-137.
Carroll,G. R., and M. T. Hannan
1989 "Densitydelay in the evolution of organizationalpopulations: A model and five empirical tests." Administrative
411-430.
2000 The Demographyof Corporations and Industries.Princeton, NJ: PrincetonUniversity
Press.
Carroll,G. R., and A.
Swaminathan
2000 "Whythe microbrewery
movement? Organizational
dynamics of resource partitioning in the Americanbrewing industryafter prohibition."
AmericanJournalof Sociology, 106: 715-762.
Carroll,G. R., and A. C. Y.Teo
1996 "Creativeself-destruction
among organizations:An
empiricalstudy of technical
innovationand organizational
failurein the Americanautomobile industry,1885-1982."
Industrialand Corporate
Change, 6: 619-644.
Chandler,A. D., Jr.
1962 Strategyand Structure:Chapters in the Historyof the
IndustrialEnterprise.Cambridge, MA:MITPress.
1977 The VisibleHand:The ManagerialRevolutionin American Business. Cambridge,
MA:HarvardUniversityPress.
1990 Scale and Scope: The Dynamics of IndustrialCapitalism.
Cambridge,MA:HarvardUniversity Press.
Chesbrough, H.
1999 "Theorganizationalimpactof
technologicalchange: A comparativetheory of national
institutionalfactors." Industrial and CorporateChange,8:
447-485.
DiMaggio, P.J., and W. W. Powell

1983 "Theironcage revisited:Institutionalisomorphismand collective rationalityin organizationalfields." American
147-160.
Christensen, C. M.
1997 The Innovator'sDilemma:
When New Technologies
Cause GreatFirmsto Fail.
Cambridge,MA:Harvard
Business School Press.
Dobrev, S. D., T. Y. Kim, and G. R.

Carroll
2002 "Theevolutionof organizationalniches: U.S. automobile
manufacturers,1885-1981."
AdministrativeScience Quarterly,47: 233-264.
G.
Dosi,
1988 "Sources, procedures,and
microeconomiceffects of
innovation."Journalof Economic Literature,26:
1120-1171.
Christensen, C. M., and J. L.

Bower
1996 "Customerpower, strategic
investment, and the failureof
leadingfirms."StrategicManagement Journal,17:
197-218.
Christensen, C. M., F. F. Suarez,
and J. M. Utterback
1998 "Strategiesfor survivalin
fast-changingindustries."
ManagementScience, 44:
S207-S220.
Chuang, Y.T., and J. A. C. Baum
2003 "It'sall in the name: Failureinducedlearningin multiunit
chains."AdministrativeScience Quarterly,48: 33-59.
Cohen, W., and S. Klepper
1996 "Firmsize and the natureof
innovationwithin industries:
The case of process and
productR&D."Review of
Economicsand Statistics, 78:
223-243.
Dunning, J. H.
1993 MultinationalEnterprisesand
the GlobalEconomy.Menlo
Park,CA:Addison-Wesley.
Emery, F. E., and E. L. Trist
1965 "Thecausal texture of organizationalenvironments."
HumanRelations,18: 21-32.
Frank,R. H., and P.J. Cook
1995 The Winner-Take-All
Society.
New York:Free Press.
Frost, P.J., and C. P. Egri
1991 "Thepoliticalprocess of innovation."In L. L. Cummings
and B. M. Staw (eds.),
Research in Organizational
Behavior,13: 87-128. Greenwich, CT:JAIPress.
Cohen, W., and D. Levinthal

1990 "Absorptivecapacity:A new
perspective on learningand
innovation."Administrative
128-152.
Fudenberg, D., and J. Tirole

1985 "Pre-emptionand rent equalizationin the adoptionof new
technology."Review of Economic Studies, 52: 383-401.
Cyert, R. M., and J. G. March

1963 A BehavioralTheoryof the
Firm.EngelwoodCliffs,NJ:
Prentice-Hall.
Galbraith,J. K.
1967 The New IndustrialState.
Boston: HoughtonMifflin.
D'Aunno, T., M. Succi, and J. A.

Alexander
2000 "Therole of institutionaland
marketforces in divergent
organizationalchange."
DiMaggio, P.J.
1990 "Cultural
aspects of economic organization."In R. Friedlandand A. F. Robertson
(eds.), Beyond the Marketplace: RethinkingEconomy
and Society: 113-136. New
York:Aldinede Gruyter.
Ghoshal, S., and D. E. Westney

(eds.)
1993 OrganizationTheoryand the
Multinational
Corporation.
London:St. MartinsPress.
Greve, H. R.
1999 "Theeffect of change on performance:Inertiaand regression towardthe mean."
2003 Organizational
Learningfrom
PerformanceFeedback:A
BehavioralPerspective on
Innovationand Change. Cambridge:CambridgeUniversity
Press.
Gusfield, J. R.
1957 "Theproblemof generations
in an organizationalstructure." Social Forces, 35:
323-330.
Hall, P.A., and D. Soskice (eds.)
2001 Varietiesof Capitalism-The
InstitutionalFoundationsof
ComparativeAdvantage.New
York:OxfordUniversityPress.
Hannan, M. T.
1997 "Inertia,density, and the
structureof organizational
populations:Entriesin European automobileindustries,
1886-1981."
Organization
Studies, 18: 193-228.

Hannan, M. T., and G. R. Carroll
1992 Dynamicsof Organizational
Populations:Density, Legitimation,and Competition.
New York:OxfordUniversity
Press.
Hannan, M. T., and J. Freeman
1984 "Structuralinertiaand organizationalchange." American
149-164.
1989 OrganizationalEcology.Cambridge, MA:HarvardUniversity Press.
Hannan, M. T., L. P6los, and G. R.
Carroll
2003a "Cascadingorganizational
change." OrganizationScience, 14: 463-482.
2003b"Thefog of change: Opacity
and asperityin organizations."
Haunschild, P. R., and B. N.
Sullivan
2002 "Learningfrom complexity:
effects of prioraccidents and
incidents on airlines'learning." AdministrativeScience
Quarterly,47: 609-643.
Haveman, H. A.
1992 "Between a rockand a hard
place: Organizational
change
and performanceunderconditions of fundamentalenvironmentaltransformation."
size and
change: Diversificationin the
savings and loan industry
after deregulation."AdministrativeScience Quarterly,38:
20-50.

Haveman, H. A., and L.
Nonnemaker
2000 "Competitionin multiplegeographicmarkets:The impact
on growth and marketentry."
Henderson, R. M., and K. B. Clark
1990 "Architectural
innovation:The
reconfigurationof existing
producttechnologies and the
failureof established firms."
Henisz, W. J., and A. Delios
2001 "Uncertainty,imitationand
plant location:Japanese
multinationalcorporations,
1990-96." AdministrativeScience Quarterly,46: 443-475.
Herriott,S. R., D. Levinthal,and
J. G. March
1985 "Learningfrom experience in
organizations."AmericanEconomic Review, 75: 298-302.
Hickson, D. J., D. S. Pugh, and
D. C. Pheysey
1969 "Operationstechnology and
organizationstructure:An
empiricalreappraisal."AdministrativeScience Quarterly,
14: 378-397.
Hymer,S.
1960 "Theinternationaloperations
of nationalfirms." Unpublished Ph.D.Thesis, Massachusetts Instituteof Technology.
HTW(High Technology West)
1970 MagneticDisk Memories. Los
Angeles: HTW,November.
lansiti, M., and K. B. Clark
1994 "Integrationand dynamic
capability:Evidencefrom
productdevelopment in automobiles and mainframecomputers." Industrialand Corporate Change,3: 557-605.
Jewkes, J., D. Sawers, and R.
Stillerman
1969 The Sources of Invention,2d
ed. New York:Macmillan.
Kamien, K., and N. Schwartz
1982 MarketStructureand Innovation. Cambridge,MA:CambridgeUniversityPress.
Khessina, O. M.
2002 "Effectsof entry mode and
incumbencystatus on the
rates of firmproductinnovation in the worldwideoptical
disk driveindustry,
1983-1999." Report2002-01,
InformationStorage Industry
Center,Universityof California, San Diego.
King, A. A., and C. L. Tucci

2002 "Incumbententry into new
marketniches: The role of
experience and managerial
choice in the creationof
dynamiccapabilities."Management Science, 48:
171-186.
Klepper,S., and K. L. Simons
2000 "Dominanceby birthright:
Entryof priorradioproducers
and competitive ramifications
in the U.S. television receiver
industry."Strategic Management Journal,21: 997-1016.
Knickerbocker,F.
1973 OligopolisticReactionand
MultinationalEnterprise.
Boston: GraduateSchool of
Business Administration,Harvard University.
Kogut, B.
1992 "Nationalorganizingprinciples of work and the erstwhile dominanceof the
Americanmultinationalcorporation."Industrialand Corporate Change, 1: 285-326.
Kono, C., D. Palmer, R. Friedland,
and M. Zafonte
1998 "Lost in space: The geography of corporateinterlocking
directorates."AmericanSociological Review, 103:
863-911.
Kostova, T., and S. Zaheer
legitimacy
underconditionsof complexity: The case of the multinationalenterprise."Academy
of Management Review, 24:
64-81.
Kummerle, W.
1999 "Thedriversof foreign direct
investment into researchand
development:An empirical
investigation."Journalof
InternationalBusiness Studies, 30: 1-24.
Langlois, R. N.
1992 "Externaleconomies and economic progress:The case of
the microcomputerindustry."
Business HistoryReview, 66:
1-50.
Lant,T. K.
1992 "Aspirationlevel adaptation:
An empiricalexploration."
ManagementScience, 38:
623-644.
Laumann, E. O., and D. Knoke
1987 The Organizational
State:
Social Choice in NationalPolicy Domains. Madison,WI:
Universityof Wisconsin
Press.
Leachman, R. C., and C. H.

Leachman
2004 "Globalization
of semiconductors: Do real men have fabs,
or virtualfabs?" In M. Kenney
(ed.), with R. Florida,Locating
GlobalAdvantage:Industry
Dynamicsin the International
Economy:203-231. Stanford,
CA:StanfordUniversity
Press.
Le Bas, C., and C. Sierra
2002 "Locationversus home country advantages in R&Dactivities: Some furtherresults on
multinationals'locational
strategies." Research Policy,
31: 589-609.
Lerner,J.
1997 "Anempiricalexplorationof a
technology race." RAND
Journalof Economics, 28:
228-247.
Levinthal, D. A.
1997 "Adaptationon rugged landscapes." ManagementScience, 43: 934-950.
Levinthal, D. A., and J. G. March
1981 "A model of adaptiveorganizationalsearch." Journalof
EconomicBehaviorand Organization,2: 307-333.
Li,J. T., and S. Guisinger
1992 "Theglobalizationof service
multinationalsin the 'triad'
nations:Japan, Europe,and
NorthAmerica."Journalof
InternationalBusiness Studies, 23: 675-696.
Lindblom, C.
1977 Politicsand Markets:The
World'sPolitical-Economic
Systems. New York:Basic
Books.
Mansfield, E.
1961 "Technicalchange and the
rate of imitation."Econometrica,29: 741-766.
1963a"Thespeed of response of
firms to new techniques."
QuarterlyJournalof Economics, 77: 290-311.
1963b"Size of firm, marketstructure and innovation."Journal
of PoliticalEconomy,71:
556-576.
1968 IndustrialResearch and TechnologicalInnovation:An
EconometricAnalysis. New
York:W.W. Norton.
March, J. G.
1988 Decisions and Organizations.
Cambridge,MA:Basil Blackwell.
1994 A Primeron Decision Making:
How Decisions Happen.New
York:Free Press.
March, J. G., and H. A. Simon

1958 Organizations.New York:
Wiley.
McKendrick,D. G.
2001 "Globalstrategy and population level learning:The case
of harddisk drives."Strategic
301-317.
McKendrick,D. G., R. F. Doner,
and S. Haggard
2000 FromSiliconValleyto Singapore: Locationand Competitive Advantagein the Hard
Disk DriveIndustry.Stanford,
Press.
McKendrick,D. G., J. Jaffee, G. R.
Carroll,and 0. M. Khessina
2003 "Inthe bud? Disk arrayproducers as a (possibly)emerform."
gent organizational
McPherson, J. M.
1983 "Anecology of affiliation."
AmericanSociological
Review,48: 519-535.
MDS (Modern Data Systems)
1972 Disk/DrumMemories Update.
Framingham,MA:MDS.
Menzel, H.
1960 "Innovation,integration,and
A survey of physimarginality:
cians." AmericanSociological
Review, 25: 704-713.
Mezias, S. J., and T. K. Lant
1994 "Mimeticlearningand the
evolutionof organizational
populations."InJ. A. C.
Baumand J. V. Singh (eds.),
EvolutionaryDynamicsof
Organizations:179-198. New
Meyer, J. W., and B. Rowan
1977 "Institutionalized
organizations: Formalstructureas
myth and ceremony."American Journalof Sociology,83:
340-363.
Michels, R.
1949 PoliticalParties.Trans.by E.
and C. Paul.(Originally
published in 1915.) Glencoe, IL:
Free Press.
Minkoff,D. C.
1999 "Bendingwith the wind:
change in
Organizational
Americanwomen's and
minorityorganizations."
AmericanJournalof Sociology, 104: 1666-1703.
Mintz, B., and M. Schwartz
1985 The Power Structureof American Business. Chicago:University of ChicagoPress.
Mizruchi,M.
1982 The AmericanCorporateNetwork: 1904-1974. Beverly
Hills,CA:Sage.
Murtha,T. P., S. A. Lenway, and
J. A. Hart
2001 ManagingNew IndustryCreation:GlobalKnowledgeFormationand Entrepreneurship
in HighTechnology.Stanford,
Press.
Nelson, R. R. (ed.)
1993 NationalInnovationSystems:
A ComparativeAnalysis. New
Nelson, R. R., and S. G. Winter
1982 An Evolutionary
Theoryof
EconomicChange. Cambridge, MA:BelknapPress of
HarvardUniversity.
Normann, R.
innovative1971 "Organizational
ness: Productvariationand
reorientation."Administrative
203-215.
Palmer, D., and B. Barber
2001 "Challengers,elites, and owning families:A social class
theory of corporateacquisitions in the 1960s." AdministrativeScience Quarterly,46:
87-120.
Pfeffer,J., and G. R. Salancik
1978 The ExternalControlof Organizations.New York:Harper
and Row.
Phillips, D. J., and E. W.
Zuckerman
2001 "Middle-statusconformity:
Theoreticalrestatement and
empiricaldemonstrationin
two markets."American
379-429.
Podolny, J. M.
1993 "Astatus-based model of
marketcompetition."American Journalof Sociology,98:
829-872.
Podolny, J. M., and T. E. Stuart
1995 "A role-basedecology of
technologicalchange." American Journalof Sociology, 100:
1224-1260.
Podolny, J. M., T. E. Stuart, and
M. T. Hannan
1996 "Networks,knowledge, and
niches: Competitionin the
worldwidesemiconductor
industry,1984-1991." American Journalof Sociology, 102:
659-689.
Porter, M. E.
1980 CompetitiveStrategy.New
York:Free Press.
1990 The CompetitiveAdvantage
of Nations. New York:Free
Press.
Porter, M. E., and M. Sakakibara
2001 "Competingat home to win
abroad:Evidencefrom Japanese industry."Review of Economics and Statistics, 83:
310-322.
Rao, H., H. R. Greve, and G. F.
Davis
2001 "Fool'sgold: Social proof in
the initiationand abandonment of coverage by Wall
Street analysts."Administrative Science Quarterly,46:
502-526.
Rogers, E. M.
1995 Diffusionof Innovations,5th
ed. New York:Free Press.
Rosenzweig, R, and J. Singh
environments
and the multinationalenterprise." Academy of Management Review, 16: 340-361.
Sch6n, D. A.
1967 Inventionand the Evolutionof
Ideas. London:Tavistock.
Schumpeter, J. A.
1934 Theoryof EconomicDevelopment: An Inquiryinto Profits,
Capital,Credit,Interest,and
the Business Cycle. Cambridge, MA:HarvardUniversity Press.
1950 Capitalism,Socialism,and
Democracy,3rd ed. New
York:Harperand Row.
Selznick, R
1949 TVAand the Grass Roots.
Berkeley,CA:Universityof
CaliforniaPress.
Simmel, G.
1955 Conflictand Web of Group
Affiliations.Trans.by K. H.
Wolff and R. Bendix.(Originally publishedin 1922.) New
York:Free Press.
Simon, H. A.
1945 AdministrativeBehavior.New
York:Macmillan.
Serensen, J. B., and T. E. Stuart
2000 "Aging,obsolescence, and
organizationalinnovation."
Swamidass, P. M.
2003 "Modelingthe adoptionof
manufacturingtechnology
innovationsby small US manufacturers:A longitudinal
investigation."Research Policy, 32: 351-366.

Teece, D. J., G. Pisano, and A.
Shuen
1997 "Dynamiccapabilitiesand
strategic management."
StrategicManagementJournal, 18: 509-534.
Vernon, R., L.T. Wells, Jr., and S.

Rangan
1996 The Managerin the InternationalEconomy,7th ed.
EnglewoodCliffs, NJ: Prentice Hall.
Terreberry,S.
1968 "Theevolutionof organizationalenvironments."AdministrativeScience Quarterly,
12: 590-613.
Weber, M.
1946 FromMax Weber: Essays in
Sociology.Trans.and ed. by
H. H. Gerthand C. W. Mills.
New York:OxfordUniversity
Press.
Thompson, J. D.
1967 Organizationsin Action. New
York:McGraw-Hill.
Tushman, M. L., and P.Anderson
1986 "Technologicaldiscontinuities
and organizationalenvironments." AdministrativeScience Quarterly,31: 439-465.
Van Valen, L.
1973 "A new evolutionarylaw."
Evolutionary
Theory,1: 1-30.
Westney, D. E.
1993 "Institutionalization
theory
and the multinationalcorporation." In S. Ghoshaland D. E.
Westney (eds.), Organization
Theoryand the Multinational
Corporation:53-76. London:
St. MartinsPress.
White, H.
1981 "Wheredo marketscome
from?"AmericanJournalof
Sociology,87: 517-547.
Williamson, O. E.
1985 The EconomicInstitutionsof
Capitalism.New York:Free
Press.
Yu, C. M., and K. Ito

1988 "Oligopolisticreactionand
foreign direct investment: The
case of the U.S. tire and textile industry."Journalof InternationalBusiness Studies, 19:
449-460.
Zaheer, S., and E. Mosakowski
1997 "Thedynamics of the liability
of foreignness: A global study
of survivalin financialservices." Strategic Management Journal,18: 439-464.
Zald, M. N.
1970 OrganizationalChange:The
PoliticalEconomy of the
YMCA.Chicago:Universityof
ChicagoPress.
Zuckerman, E. W., T. Y. Kim, K.
Ukanwa, and J. von Rittmann
2003 "Robustidentities or nonentities? Typecastingin the feature-filmlabormarket."American Journalof Sociology,
108: 1018-1074.
APPENDIX:Data on the Global Hard Disk Drive Market

Dataon every organizationever to manufacturea harddisk drivewere gathered from marketresearch reports,publiclyavailablefinancialinformation,
industryparticipants,and an extensive search of the business press. From
these disparatesources, we compiled the life historyof each disk drivecompany.These histories cover marketentryand exit dates, sales, presence in
each year in a given "formfactor"(the physicalsize of a disk drive),products' technical specifications,acquisitionhistory,and nationalityfor each
company that made a harddisk drivesince the first known manufactureof a
disk in 1956 through1998. The resultingdatabase includes 171 organizations
that manufacturedharddisk drives at any time or place over the period.
Nearlyall organizations(155) exited the industryby the end of 1998. The
data cover 1,538 organization-years.
The primarysource of data for this study is the Disk/TrendReport:RigidDisk
Drives, publishedannuallysince 1977 by Disk/Trend,Inc., a marketresearch
company in MountainView, California.These reportstrackevery known
company that made harddisk drives, list detailed productspecifications and
shipment dates, and publishrevenue and unit shipment informationby form
factor and capacityrange. The reportsdo not, however, list the date of first
entry for many companies that produceddisk drives before 1976. Inseveral
instances, companies operatingin 1976 shipped theirfirst disk drives in the
1960s, but these productswere no longer in productionwhen the first
Disk/TrendReportwas publishedand so "firstshipment" dates could not be
determined.We contacted survivingcompanies for information,spoke with
retiredengineers involvedin the development of a company'sfirst disk drive
for both survivingand formerdisk drive producersand often received copies
of writtentechnicaland marketinformationin their possession, reviewed
company documents in the files of Disk/Trend,Inc., and researched company histories in books, financialreports,and the business press. Reports by
two marketresearchfirms in particularwere basic sources of informationon
productspecificationsand shipment dates of firms operatingduringthe
1960s and early 1970s (HTW,1970; MDS, 1972). These searches resulted in
earlierentry dates than would otherwise be inferredfrom informationin
Disk/Trendfor 20 of 32 firms makingdisk drives as of January1, 1976. An
additional17 companies made disk drives but exited the industrybefore
Disk/Trendwas first published.Of these 37 firms (20 plus 17), precise dates
of first shipment for 15 companies were unavailable;instead, entry dates
were set at three months after first productannouncement.
Lifeevents were coded so that name changes and reorganizationswere not

counted as entries and exits. A numberof disk drivecompanies changed
their names or were acquiredduringthe periodof study. Name changes
sometimes occurredwhen a parentfirmannounceda new strategy. In these
cases, the disk driveoperationsmay have been combined with some other
corporateactivity.These were not coded as marketexits. Forinstance, England'sData RecordingInstrumentsput its disk driveoperationsinto a newly
formed peripheralsoperationnamed Data RecordingEquipmentin 1978, and
its name was changed again in 1982 to NewburyDataafter it was combined
with other operations.Over time these incarnationswere treated as the
same entity,so its reorganizationswere not counted as entries or exits. Nor
did we code as exits any acquisitionsof disk drivefirms by non-diskdrive
companies if the disk driveoperationswere left largelyintact. Forexample,
PerkinElmeracquiredWangco, a disk drive maker.Wangco then continued
to make disk drives as a PerkinElmersubsidiary.The acquisitionwas thus
not coded as an exit (or a new entry).By contrast, if a companywas liquidated and the liquidatedassets formed the basis of a new company,then the
two events were coded as an exit and an entry.
To investigate the technology race in this industry,we coded for each firm
whether it shipped a productin a particularform factorand capacity range in
any given year.The bulkof shipments were non-captivemarkettransactions,
and our analysis of the technology positions of each firmwas restrictedto
these shipments. We did includefirms in the survivalanalysis, however,
even if they only engaged in captive production.We reliedon Disk/Trend
technology classificationsfor all 1976-1998 observations. Disk/Trendcreated
productgroup categories by drivetype (cartridgedisk drives, disk pack
drives,fixed disk drives)and by capacityrange (30-60 megabytes, 100-300
megabytes, etc.). Because of the dramaticimprovementsin disk drive
capacityover the years, these categories evolved over time. Forinstance, in
1976, Disk/Trendcreated nine productgroups, includinga groupconsisting
of firms shippingdisk drives of less than 12 megabytes and one consisting
of firms shippingdisk drives greater than 200 megabytes. In 1996, nine
productgroups were also used, but this time the smallest capacity range for
a productgroupwas less than 500 megabytes and the largest was for more
than 20 gigabytes. We coded whether a given productgroup (capacityrange)
representeda relatively"low-," "medium-,"or "high-"capacitydisk drivefor
that form factor in that year.Thus, the technologicalposition of a firm in a
given form factorwas made relativeto that of other firms in the same form
factor (otherfirms shipping5.25-inchdrives, for example)and not across
form factors, and a disk drivethat was a high-capacityproductin one year
could become a medium-or low-capacityproductin the next year or two.
This approachallowed us to specify in our models the precise technological
domainof each organizationin each year. Forinstance, an organizationoffering both 3.5-inchand 5.25-inchproductsin a given year could have been in
any of six technology segments: low-, medium-,and high-capacitywithinthe
3.5-inchform factor and low-, medium-,and high-capacitywithinthe 5.25inch form factor. If this organizationwere in all six of these segments, then a
rivalthat overlapsthis organizationin only one of these six segments would
contribute1/6 to the organization'stechnology-overlapdensity score. By contrast, a rivalthat offered productsin all six of these areas would contribute
one to the organization'stechnology-overlapscore. At the extremes, if an
organizationproducedin the same form factors and capacitylevels as all
other manufacturers,then its technology-overlapdensity would equal simply
the numberof other organizations.If an organizationdid not overlapwith any
other manufacturers,however, then its technology-overlapdensity would
equal 0, reflectingits complete technologicaldifferentiation.
Althoughthe Disk/TrendReportis an incrediblyrichand exhaustive source of
informationon products,revenues, and shipment volumes, its use requires
some care. When a disk drivecompany acquiresanother,Disk/Trendlists the
combined productsunderthe name of the acquiringfirm, however,
Disk/Trendnotes the shipment dates of productsthat were originallydeveloped by the acquiredfirmas the dates of first shipment, not dates of first
shipment by the acquiringfirm. Knowingthis, we took care not to code a
companyas shippinga particulardisk driveearlierthan it actuallyhad. Moreover, many productswere announcedthat never shipped, and sometimes
the same productwas reportedlyshipped on two differentdates depending
on the year of the Disk/TrendReport.A couple of companies, such as Memorex Telex,were listed as disk drivecompanies when in fact they only resold
drives made by other manufacturers.Ina few cases it was not clear that a
company actuallyever shipped a product.Forexample, Disk/TrendReport

would describe a new company and list its products,all of which were
scheduled to be shipped sometime after the publicationdate of Disk/Trend
Report.Thatcompany would not appearin the next year's report,however,
and it was unclearwhether the firm in fact shipped the products,only to exit
before Disk/Trendwas publishedthe followingyear, or else never shipped.
We checked each of the more than 10,000 productslisted in the Disk/Trend
Reports(includingthe same productappearingin more than one year)for
such anomalies. When we encoutered problems,we contacted Disk/Trend,
Inc.staff, who always kindlyhelped us to clarify.
We also had to classify productsand capacity ranges for 1956-1975, priorto
the publicationof Disk/Trend.This requiredcreatingproductgroups that mirroredthe technologicalstate of the industry.We started by extending the
1976 productgroup classificationsback in time untilthey seemed to collide
with the realityof the productsbeing offered. This resulted in a gradual
reductionin the numberof productgroups in the 14-inchform factorand the
creationof productgroups for largerform factors. First,as was the practice
in Disk/Trend,we groupedtogether similarlysized form factors. Thus, 39inch, 34-inch,and 31-inch productswere groupedtogether as one form factor;28-inch,26-inch,and 24-inch productswere similarlygrouped;and 16inch, 14-inchand 12-inchproductswere classified as a single form factor.
(Forreportingpurposes, Disk/Trend,for example, grouped 10.5-inchdrives
with the 14-inchform factor,9-inchdrives with 8-inch,3.9-inch drives with
3.5-inch,3-inch drives with 2.5-inch,and 1.3-inchdrives with 1.8-inch.)Only
a few new capacity ranges were created for the 1956-1975 period.This procedure affected 282 firm-yearobservations.

Barnett y McKendrick - 2004 - Why Some Organizations More Competitive

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Barnett y McKendrick - 2004 - Why Some Organizations More Competitive

Uploaded by

Copyright:

Available Formats

Why Are Some Organizations More Competitive than Others?

Evidence from a Changing

Why Are Some

In this paper, we question the idea that large organizations

@ 2004 by Johnson GraduateSchool,

Thanksfor comments from GlennCarroll,

size. Institutional economists have long argued that with

tial actors" that affect state policies (Laumannand Knoke,

Changing Global Market

ogy diffusion literature(see Mansfield, 1963a; Rogers, 1995;

Changing Global Market

sourcing, production,research and development (R&D),

By and large, research has examined competitive constraints

Changing Global Market

tition are linkedcausally,each accelerating the other in the

tudes of oaand 3 as well as the observed values of E for any

Changing Global Market

become less immediatelyresponsive to external demands

Largeorganizations,by contrast, enjoy advantages that make

ILELk + sO(ISkwjESk)+ aL('Lk?jELk)],

competitionincreasesthe strengthof its rivalry,

Changing Global Market

and large foreign rivals,respectively, as a function of their

LF,indicatingthat a foreign rival's

exposureto domesticcompetitionin its home countryincreasesthe

ulationdating back to the beginningof the industryin 1956.

Changing Global Market

Over all years

ing characteristicof competition in the industryhas been the

Koreancompanies followed in the early and mid-1980s.

Changing Global Market

innovate rapidly.But the result is extremely short product

We used the modeling framework of organizational ecology

thought to be non-monotonicallyrelated to density, falling

Changing Global Market

the industryin each year. Forobservations after 1976, we

age in our analysis. We also allowed for the possibilitythat

Changing Global Market

gregated density terms show that competition has been

Organizationwas founded de novo

Non-overlapwith domestic rivals

* Standarderrorsare in parentheses. Each model also includes markettenure, calendar

* Standarderrorsare in parentheses. Forregioneffects, the U.S. is the omitted category

tion revealed in these models, but only due to competition

Changing Global Market

tive experience on its own exit rate, controllingfor all the

Overlapwith small domestic rivals

Overlapwith large domestic rivalsx calendaryear (1956 = 0)

Numberof foreign rivals

" < .05.

Changing Global Market

Small Firm With Mean Number of Rivals per Year

Effect of a Small Rival

Effect of a Large Rival

Rivals' Competitive Experience

effect of the competitive experience of these rivalson an

Rivals' Strength Dgminates

Rivals' Competitive Experience

Changing Global Market

estimates of model 14). Forobservations on this line, then,

Numberof low-capacityform factors producedby

Changing Global Market

Organizationof markettenure 0-1 year