1. No category

When Are Technologies Disruptive? A Demand-Based View of the Emergence of Competition
Author(s): Ron Adner
Source: Strategic Management Journal, Vol. 23, No. 8 (Aug., 2002), pp. 667-688
Published by: John Wiley & Sons
Stable URL: http://www.jstor.org/stable/3094287
Accessed: 05/07/2010 14:37
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=jwiley.
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 [email protected].
John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Strategic
Management Journal.
http://www.jstor.org
Strategic ManagementJournal
Strat. Mgmt. J., 23: 667-688 (2002)
Published online 28 March 2002 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/smj.246
(
WHENARETECHNOLOGIES
DISRUPTIVE?A
DEMAND-BASEDVIEWOF THEEMERGENCEOF
COMPETITION
RON ADNER*
INSEAD, Fontainebleau Cedex, France
By identifyingthepossibilitythattechnologieswithinferiorperformance
can displaceestablished
incumbents,the notionof disruptivetechnologies,pioneeredby Christensen(1997), has had a
profoundeffecton the way in whichscholarsand managersapproachtechnologycompetition.
Whilethe phenomenonof disruptivetechnologieshas been well documented,the underlying
theoreticaldriversof technologydisruptionare less well understood.This article identifes
the demandconditionsthatenabledisruptivedynamics.By examininghow consumersevaluate
technologyand how this evaluationchangesas performanceimproves,it offersnew theoretical
insightintothe impactof thestructureof thedemandenvironment
on competitivedynamics.Two
new constructs preferenceoverlapand preferencesymmetry are introducedto characterize
the relationshipsamongthe preferencesof differentmarketsegments.The articlepresentsa
formalmodelthatexamineshowtheserelationships
leadto theemergenceof differentcompetitive
regimes.The model is analyzedusing computersimulation.The theorwand model results
hold implications
for understanding
the dynamicsof disruptivetechnologiesand suggestnew
indicatorsfor assessingdisruptivethreats.Copyright C) 2002 John Wiley & Sons, Ltd.
From S-curves (Foster, 1986), to technology
trajectories(Dosi, 1982) to punctuatedequilibria
(Tushman and Anderson 1986), the dominant
view in technology strategy has been that the
displacementof establishedfirmsandtechnologies
by new firms and technologies is driven by
the superiorperformanceoffered by newcomers
and establishedplayers' difficulties in matching
theirperformanceand capabilities.By identifying
the possibility that technologies with inferior
performancecan displaceestablishedincumbents,
the notion of disruptivetechnologies, pioneered
by Christensen (1997), has had a profound
effect on the way in which scholars and
managersalike approachtechnology competition
and has prompteda reassessmentof the ways in
Key words: disruptive technologies; demand heterogeneity; technology strategy; technology evolution
*Correspondence
to: R. Adner,INSEAD,Boulevardde Constance 77305 Fontainebleau
Cedex,France.
Copyright (C)2002 John Wiley & Sons, Ltd.
which firms approach technological threats and
opportunities.
While the phenomenon of disruptive technologies has been well documented, the underlying
theoretical drivers of technology disruption are less
well understood. Understandingthe conditions that
give rise to technology disruptions, however, is
fundamental to assessing the pervasiveness of the
phenomenon and for guiding strategic responses
to potentially disruptive threats. Indeed, without a
theoretical underpinning with which to answer the
question of sWhen are technologies disruptive?' it
is impossible to make ex ante distinctions between
disrup.tive technologies and technologies that are
merely inferior.
This article identifies the demand conditions that
enable disruptive dynamics. By examining how
consumers evaluate technology and how this evaluation changes as performance improves, it offers
new theoretical insight into the impact of the structure of the demand environment on competitive
ReceivedI December1999
Final revisionreceived16 November2001
668
R. Adner
dynamics. Two new constructs preference overlap and preference symmetry are introduced to
characterize the relationships among the preferences of different market segments. The article
presents a formal model that examines how these
relationships lead to the emergence of different
competitive regimes. The model is analyzed using
computer simulation. The theory and model results
hold implications for understanding the dynamics
of disruptive technologies and suggest new indicators for assessing disruptive threats.
TECHNOLOGYCOMPETITIONAND
DISRUPTIVETECHNOLOGIES
Explanations of technology competition outcomes
have tended to focus on the supply-side interactions of firms and technologies. At the technology
level, outcomes have often been attributed to the
exhaustion of the incumbent technology's development trajectory (Foster, 1986; Utterback and
Abernathy, 1975) or the entrant's outright superiority. At the firm level, the challenge of managing
displacement threats has often been attributed to
an unwillingness to cannibalize existing technology investments (Kamien and Schwartz, 1982);
organizational inertia (Hannan and Freeman, 1977;
Tushman and Romanelli, 1985); and the inability
to adopt the necessary skills needed to engage in
the new technology (Henderson and Clark, 1990;
Leonard-Barton, 1992).
Closer examination of technology competition,
however, reveals that technology transitions are
not necessarily due to the incumbent technology's inherent limits (Christensen, 1992; Cooper
and Schendel, 1976; Henderson, 1995), the new
technology' s ability to provide superior performance (Christensen, 1997; Levinthal, 1998), or
incumbents' inability to master new skills (Bower
and Christensen, 1995). While these factors are
important, numerous cases of innovative incumbents who did not suffer from these handicaps, yet
nonetheless mismanaged the challenge of technological transition (Smith and Cooper, 1988; Smith
and Alexander, 1988; Christensen and Rosenbloom, 1995), suggest the need for additional
explanations and argue that new insight can be
gained by considering the broader environment in
which technologies compete (Afuah and Bahram,
1995; Afuah, 2000).
Copyright (C)2002 John Wiley & Sons, Ltd.
Studies exploring the impact of market demand
on development strategies offer a complementary
set of explanations that highlight the influence
of consumer needs on technology development
at the level of technology projects (von Hippel,
1988; Lynn, Morone, and Paulson, 1996), business
strategy (Kim and Mauborgne, 1997; Day, 1990;
MacMillan and McGrath, 2000) and the broader
evolution of technological trajectories (Abernathy
and Clark, 1985; Malerba, 1985; Christensen,
1997; Sutton, 1998; Malerba et al., 1999; Tripsas,
2001; Adner and Levinthal, 2001).
The most influential expression of a demandside role in technology competition has been Christensen's examination of disruptive technologies.
Disruptive technologies are technologies that introduce a different performance package from mainstream technologies and are inferior to mainstream
technologies along the dimensions of performance
that are most important to mainstream customers.
As such, in their early development they only serve
niche segments that value their nonstandard performance attributes. Subsequently, further development raises the disruptive technology's performance on the focal mainstreamattributesto a level
sufficient to satisfy mainstream customers. While
improved, the performance of the disruptive technology remains inferior to the performance offered
by the established mainstream technology, which
itself is improving as well. Technology disruption
occurs when, despite its inferior performance on
focal attributes, the new technology displaces the
mainstream technology from the mainstream market. Christensen plots the performance-provided
and performance-demandedtrajectories for different technologies and market segments, and shows
that technology disruptions occur when these trajectories intersect. He documents these dynamics
in numerous contexts, including hard disk drives,
earthmoving equipment, retail stores and motor
controls (Christensen, 1997).'
' Christensen'smost prominentillustrationof disruptivetechnologies draws on researchin the hard disk drive industry.
Accordingly,the harddisk drive exampleis used to illustrate
the argumentsmade in this paper.Christensenobservedthat
new generationsof disk drivetechnologywere firstadoptedin
theyoffered.
nichemarketswhichvaluedthe newfunctionalities
For example73.5-inch hard disk drive technologyfound an
early home amongnotebookcomputerusers who appreciated
its reducedweight,size, and powerconsumption.Withfurther
(sustaining)development,3.5-inchharddrivesthenwenton to
disruptthe desktopmarket,replacingthe incumbent5.25-inch
technology.Thus,3.5-inchharddisk drivesdisplaced5.25-inch
Strat. Mgntt. J., 23: 667-688 (2002)
Whenare TechnologiesDisruptive?
Thedynamicsof disruptivetechnologiesarethus
charactenzedby three aspects: incumbenttechnologies that are displaced from the mainstream
themon
marketby technologiesthatunderperform
the performancedimensionsthat are most important to mainstreamconsumers;mainstreamconsumerswho shift theirpurchasesto productsbased
in the invading technology, even though those
productsoffer inferiorperformanceon key performance dimensions;and incumbentfirms that do
not react to disruptivetechnologies in a timely
manner.
Christensenexplainsthesedynamicsas resulting
fromthe interactionof resourcedependence(Pfeffer and Salancik,1978) and performanceoversupply. He argues that the resource dependenceof
incumbentson their most demandingcustomers
guides investmentstowardsenhancingfocal mainstreamperformancefeatures.Because the incumbent technology offers superiorperformanceon
these dimensions, incumbentfirms' investments
are directedtowardsextendingthe existing technology, rather than the (potentially) disruptive
technological opportunity.Incumbentshave an
additionalincentiveto ignoredisruptivetechnologies because, with their lower performance,they
appeal to the low-end, low-profitportion of the
mainstreammarket. In contrast, entrant firms,
whose decisionsare not constrainedby an existing
customerbase and whose technologyoffers inferior performanceon the focal mainstreamdimensions, are forcedto identifyconsumerswho value
the new featuresoffered by the new technology
and supportits furtherdevelopment.
Christensenintroducesthe idea of sperformance
oversupply'to explainthe mainstreamconsumers'
decision to adoptthe disruptivetechnologyin the
face of superiorincumbenttechnology.The principle of performanceoversupplystates that once
consumers'requirementsfor a specific functional
attnbuteare met, evaluationshifts to place greater
emphasison attributesthat were initially considered secondaryor tertiary.By this logic, when the
capacity providedby 5.25-inch hard disk drives
669
exceeded the requirementsof desktopconsumers,
these consumersbegan to place greaterweight on
attributessuch as size and weightin theirpurchasing decision and thus chose 3.5-inch hard drives
despite their lower capacity and higher cost per
megabyte.
This explanation of the drivers of technology disruptionleaves several issues unresolved.
Clearly the different functionalpackage offered
by the disruptivetechnology plays a role in its
adoption in the initial niche market. However,
the newfoundprominenceof previouslymarginal
attributesseems an incomplete explanationfor
adoptionin the mainstreammarket why would
desktopusers care about weight or size of internal hard disk drives? Similarly, much has been
made of the performanceand price/performance
disadvantagesof disruptivetechnologies relative
to theirmainstreamrivals,but the explicitroles of
price and performancein drivingdisruptionshave
Finally,while the promiremainedunderexplored.
nence of the performance-provided-performancedemandedrelationshiphighlightsthe criticalrole
of the demandenvironmentin shapingdisruptive
competition,the demand-sidefactorsthatdrivethe
emergenceof competitionremainlargely unstudied. For both researchersand managers,resolving
these issues is a necessaryconditionfor defining
the boundariesof disruptivetechnologies.
A DEMAND-BASEDVIEW OF
TECHNOLOGYCOMPETITION
This article develops a demand-basedview of
technology competitionthat resolves these open
questions surroundingdisruptivetechnologiesby
formallymodelingthe role of the demandenvironmentin shapingcompetitivedynamics.The structure of demandis characterizedby two elements
of the relationshipbetweenmarketsegmentpreferences: preference overlap and preference symmetty. Preferenceoverlaprefersto the extentto which
developmentactivity that is valued in one segmentis also valuedin anothersegment.Preference
drives from the mainstreamdesktopmarket,despite offering symmetryrefers to the symmetryof this overlap
lower storagecapacity,slower data access speeds, and being
the relative size of the functionalsshadows'that
basis.Further,thosefirms
morecostlyon a dollar-per-megabyte
that dominatedthe desktopmarketwith 5.25-inchtechnology segmentscast on each other. The articlepresents
a model that examineshow technologyimprovedid not managethe transitionto 3.5-inchtechnologyand were
displacedby entrantfirmsemployingthe new technology.Note
ments can blur the boundariesthat divide market
thatthecontextis internalharddiskdrives,notremovablefloppy
segmentandlead to differentcompetitiveenvirondisks,andthatcompatibilityanddisk portability(as opposedto
arenot significantconcernsforconsumers. ments. It shows that the extent to which demand
computerportability)
Copyright C) 2002 John Wiley & Sons, Ltd.
Strat. Mgmt. J., 23: 667-688 (2002)
670
R. Adner
heterogeneityis masked depends on consumers'
marginalutility from performanceimprovements,
which dictates their willingness to pay for product enhancements,andon the relationshipbetween
the functionalpreferencesof the differentmarket
segments.
The model is used to examine how preference overlap and preference symmetry interact
to affect the emergenceof competition.Controlling for supply-side asymmetriessuch as initial
resourceendowmentsand technologicalpotential,
the model examines the natureof three distinct
competitiveregimesthat arise undervaryingconfigurationsof demand:competitive isolation, in
which technologiesdo not interactthroughoutthe
courseof theirevolution;competitiveconvergence,
in which technologiesevolve to competehead-on
for the same consumer groups; and competitive
disruption,in which one technology cedes dominance of its home marketto its rival.
As technology development progresses, consumers' performancerequirementsare met, and
then exceeded, by their home technology. As
performance continues to surpass consumers'
requirements, consumers' willingness to pay
for improvementsdecreases, opening the door
for lower-priced,lower-performance(disruptive)
offers to capturethese consumers.As the overlap
between marketsegments' preferencesincreases,
firmshave greaterincentivesto enterrivals' markets. When preferenceoverlapis asymmetric,the
firm whose technologycasts a largerperformance
shadowon its rival's market,and whose technology is thereforerelevant to a larger numberof
consumers,has greaterincentiveto invade,trading
price for volume. The invadedfirm,confrontinga
smaller set of potentialusers, chooses to exploit
existing opportunitiesin the uncontestedportion
of its marketsegmentratherthan engage in price
competitionwith its rival.
The resultsof the analysishighlightthe relationshipsamongconsumerpreferencesandconsumers'
willingnessto pay for performanceimprovements
as key factors that give rise to technology disruptions. The analysis reveals the dynamics by
whichthe underlyingheterogeneityin marketpreferences,whichinitiallyactsto separatemarketsegments and attenuatecompetition,is progressively
maskedas technologyimprovementsexceed consumerrequirements.The results also identify the
increasingimportanceof unit price in determining consumers'choices as technologyperformance
Copyright C) 2002 John Wiley & Sons, Ltd.
progressesbeyond their requirements.This finding offersan alternativeto performanceoversupply
in explaining the consumer adoptionof disruptive technologies and points to price trajectories
as useful complementsto performancetrajectories
in identifying disruptivethreats. These findings
are supportedby data from the hard disk drive
industry.2
The remainderof the paper is structuredas
follows: first, a conceptual characterizationof
demandand demandstructureis presented.This
characterizationis then used to develop a formal model for examininghow demandstructure
influencescompetitiveinteractions.The model is
analyzedusing computersimulationand the simulationresults are presentedand interpreted.The
paper concludes with a discussion of the results
disruptive
andtheirimplicationsfor understanding
of
evolution
technologiesand, more broadly,the
competition.
DEMANDSTRUCTURE
the influenceof demandstructureon
To understand
the emergenceof competitionwe must characterize the demandenvironmentin whichtechnologies
compete. To do this we model the behaviorsof
consumersas individualdecision-makersand as
membersof marketsegments.Consumersarecharacterizedby the way they tradeoff performanceon
differentfunctionalattributes,their willingnessto
pay for performanceimprovements,and the minimum performancethresholdthat a productmust
reach if it is to be of value to a given consumer.
Marketsegmentsare composedof consumerswith
the sameperformancetrade-offsand are identified
by value trajectories,which characterizethe tradeoff. The focus of the analysis is the relationship
between marketsegment's value trajectoriesand
on the emergenceof competitiveisolation, convergenceand disruption.
2Adnerand Zemsky(2001) develop a formalgame theoretic
modelthatbuildson the analysisin this paperand generalizes
the breakdown
some of its results.They formallycharacterize
of competitiveisolationand examinethe additionaleffects of
costs, segmentsizes andnumberof firmsusingeach
asymmetric
technology.
Strat. Mgrrzt.J., 23: 667-688 (2002)
Whenare TechnologiesDisruptive?
671
Individual consumers: thresholds, utilities and
diminishing returns
even those with similar functional preferences.
Differencesin consumers'willingnessto pay may
drivenby differentialbudgetconstraints.They
be
The notion of thresholds,defined as critical peralso be driven by nonbudgetaryconsideramay
formance levels that must be met for an offerfor example, differences in the ability of
tions;
ing to become relevant to a decision set, is
to exploit the product. Such hetcustomer
the
well establishedin the social sciences (Granovetmay
stem from the customer's intererogeneity
ter, 1978; Vanan, 1978; David, 1969; McFadden,
1986), capabilities(Amit
(Barney,
resources
nal
1986; Meyer and Kahn, 1991). We distinguish
humancapital(Becker,
1993)
or
Schoemaker,
between two types of thresholds.A consumer's and
is able to
programmer
an
efficient
(e.g.,
1962)
functional thresholdspecifies the minimumlevel
system
computer
a
given
from
of performancebelow which a consumerwill not derivemorebenefit
Alternatively,
programmer).
inefficient
can
an
accepta productregardlessof its price. Functional than
thresholdsare determinedin partby inherenttask differencesin consumers'willingnessto pay may
requirementsand in partby context.Thus,a prod- stem from the scale at which the buyercan apply
uct that falls below one consumer's functional the product.A customerwho can applythe product
thresholdmay well be acceptableto anothercon- towardthe productionof a good that he can sell
to a largedownstreamcustomerbase will be willsumerwith a differentfunctionalthreshold.
As productperformanceimproves beyond the ing to pay more for the productthan a customer
functionalthreshold,a consumer'srelative pref- with a smallercustomerbase. Finally,differences
erences among the possible functionalattributes in willingnessto pay may reflect variationin the
impact the benefit she derives from the product. availabilityandpresenceof a substituteproductor
Functionalbenefit is thus a function of both the service. A consumerwho has previouslyinvested
product's objective performanceon each func- in a substitutegood will benefitfromthe new prodtional attnbute (e.g., speed, capacity, reliabil- uct only to the extent of the product'srelative
ity) and the consumer's trade-offs among these performanceimprovementover the existing subattributes.3For example, while a vehicle that can stitute.A similarconsumer,not in possessionof a
carry 50 passengers with a maximum speed of
substitute,will value the producton the basis of
30 miles per hour may satisfy the functional the absolutebenefitit provides.
thresholdsof both a private driver and a public
While consumers have a minimum threshold
driver,the latterwill derive greater for acceptableperformance,thereis no analogous
transportation
functionalbenefitfrom the offer. Thus, functional boundarythat specifies a maximumlimit to the
requirementsindicateinitial thresholdswhile rel- functionalperformancethat a consumerwould be
ative preferencesdictateconsumers'evaluationof
willing to accept.4At the same time, consumers
performanceimprovement.In this regard,Chris- face decreasing marginal utility from increases
tensen's trajectoriesof performancedemandedby in functionalitybeyondtheirrequirements(Meyer
differentmarket segments can be interpretedas andJohnson,1995). Correspondingly,
it is reasonplots of averagesegmentfunctionalthresholds.
able to assume that consumersshow a positive,
Net utilitythresholdsincorporateprice into the but decreasing,willingness to pay for improveconsumer'sdecisionfunction,specifyingthe high- ments beyond their requirements.Even if conest pricea consumerwill pay for a productthatjust sumers place little value on performancediffermeets her functionalthreshold.Here too, we may ences at sufficientlyhigh absolutelevels of funcexpect to find heterogeneityamong consumers, tionalitythey will still, all else being equal,choose
the more advancedproduct.5
follows a long traditionof work in
This conceptualization
marketing,decision science, and economics(Griliches,1961;
Lancaster,1979; Green and Wind, 1973; Trajtenberg,1990)
that suggests that consumershave relative preferencesfor
andthatconsumerchoicecan be usefully
productcharacteristics
conceivedas the maximizationof utilitymeasuredin termsof
thatareembodiedin theirproduct
the functionalcharacteristics
of preferencesfor goodsas beingderived
choices.Thetreatment
lies at the
from preferencesfor collectionsof characteristics
heartof establishedtechniquessuchas hedonicanalysis,conjoint
logit modelsof brandchoice.
analysis,andmultidimensional
3
Copyright C) 2002 John Wiley & Sons, Ltd.
Forthepulposesof thispaperincreasedfunctionalperformance
is treatedas a purelya positivefeature.
of technologyinnova5 Using a relatedmodelto studypatterns
tion,AdnerandLevinthal(2001) findthatcompetingfirmsmay
continueto enhanceproductperformancebeyondconsumers'
have little effect on conneeds,even whensuch enhancements
sumers'willingnessto pay,as a formof nonpricedifferentiation.
4
Strat. Mgmt. J., 23:
667-688(2002
672
R. Adner
Figure2 shows hypotheticalvalue trajectoriesfor
consumers in the market for desktop personal
computers(PCs),who valuestoragecapacitymuch
more than portability;consumersin the market
for personaldigital assistants(PDAs), who value
portabilitymuch more than storagecapacity;and
consumersin the marketfor notebookcomputers
(NCs), who value capacityandportabilityin equal
measure.
The preference overlap betweenthese segments
is the degreeof similaritybetweentheirfunctional
preferences,which is graphicallyreflectedin the
differencebetween their trajectories.The greater
the preferenceoverlap,the closer the value trajectories,andthe greaterthe segments'agreementon
As illustratedin
the level of productperforrnance.
derives a utilgroup
PC
the
Figure2(a), whereas
level of
a
utility
A
and
product
ity level of 3 from
a utilderives
group
PDA
the
1.4 fromproductB,
product
from
3
A
and
product
from
1.4
ity level of
B. As preferenceoverlapincreases,as illustratedin
Figure2(b) for the PC andNC groups,theseevaluationsconverge.Here,the PC group'sderivedutilities of 3 and 2.7 from productsA and C, respectively, are in greateragreementwith the utilities
of 1.6 and 3 derivedby the NC group.Preference
overlapis thus a measureof the extent to which
one market segment's satisfactionwith a given
product'sfunctionalityis indicativeof the satisfacutility
is the gradientof theCobb-Douglas
6 The valuetrajectory
tion of anothergroup the performanceshadow
curve. It is thus the vector that minimizesthe level of total
cast on a segment's value trajectoryby progress
functionalattainmentrequiredto attaina specifiedutilitylevel.
termsit is definedby the vectorwhichminimizes along its counterpart's
In Lancastrian
trajectory.Whenindividual
resourcesrequiredto attaina givenutilitylevel;
thecharacteristic
differentmarketsegments,
pursue
initially
firms
functionis equal
thatis, thevectoralongwhichthecompensating
the segments' preferenceoverlap is an indicator
to unity(Lancaster,1979, 1991).
Market segments: Value trajectories and the
structure of demand
Marketsegmentsare composedof consumerswith
similarfunctionalpreferences.Segment members
may have heterogeneousfunctionaland net utility thresholds.As a useful shorthandwe define
segments' value trajectories as characterizingthe
memberconsumers'relativepreferencesfor functional attributes.The relationshipbetween value
trajectoriesmaps the structureof demand. The
overlapbetweenvalue trajectoriesandthe symmetry of this overlapindicatesthe differentialmarket
incentivesfirmsface in choosing their innovation
activities, which in turn drive the emergenceand
evolution of competition.These relationshipsare
developedformallyin the next section.
As illustratedin Figure 1, the value trajectory
identifies the directionof propagationof a market segment'sindifferencecurvesas they progress
towardhigher utility levels.6 Consumers'evaluation of products'functionalperformanceis determined accordingto the products'projectiononto
the value trajectory-the fartherout on the value
trajectorythe projection,the greaterthe product's
utility to the consumer.
Considerthe following simplifiedexamplefrom
the market for information storage products.
\
.Q
\
X
/
Functional attribute X
Figure 1. Indifference curves and a value trajectory
Copyright (C)2002 John Wiley & Sons, Ltd.
Strat. Mgmt. J., 23: 667-688 (2002)
in
X\
\<
,AJpcX.
Upc =
1
=
=
_
1
Whenare Technologies Disruptive?
673
-o
@
/
o 4
en
\
owowo \
j
ow
/
\
,'
/
, '
\
\
\_
_
(a)
\_
nProduct
Upc=2
Value
Trajectory
B
Portability
|
0
PCValue
_/\Trajectory
3
:
<a)
Upc=3
/
k
|
\
\
3
.-'
45
_ry
\
\
,,,/
,
</Product
\\
C
e
I
Unc= 1
z
(b)
Figure 2.
Portability
(a) Small, symmetric preferences overlap. (b) Large, asymmetric preference overlap
of the ease with which the firmscan invade other
marketniches.
While the magnitude of preferences overlap
speaksto the absolutedegreeof preferencesimilarity,preference symmetry refersto the relativevalue
each segment places on performanceimprovements along anothersegment's value trajectory.
When preferenceoverlap is symmetricthe utility which one group derives from a given level
of performancealong the others' trajectoryis the
same for both groups.In the case of the PC and
PDA groups,each derivesa functionalutilitylevel
Copyright C) 2002 John Wiley & Sons, Ltd.
of 3 when the other drives a level of 1.4. When
preferencesare not symmetric,a product positioned at a given distance along one segment's
valuetrajectoryprovidesa differentlevel of utility
to membersof the other segments than a product positionedat the same distancealongthe other
segment'svalue trajectoryprovidesto membersof
the first segment. In the case of the PC and NC
groups,for a utility level of 3 along its counterpart'svalue trajectory,the PC groupwill derivea
utilityof 2.7, while for a utilitylevel of 3 alongthe
PC group'svalue trajectorythe NC groupderives
Strat. Mgmt. J., 23: 667-688 (2002)
674
R. Adner
a utility of only 1.6. For firms initially pursuing
different market segments, the symmetry of preference overlap plays a critical role in structuringtheir
differential incentives for pursuing other segments.
Conceptualizing demand according to consumers' requirements and preferences suggests
a demand landscape that coevolves with technology development: as consumers' requirements
are exceeded, they derive positive but decreasing
marginal utility from furtherperformance improvements. This satisfaction leads to the need for
greater and greater performance increases to support a given increase in utility, which is reflected
in a decreasing willingness to pay for a given level
of performance improvement. Stated differently, as
performance exceeds requirements, price increases
in importance. As developed below, these changes
in consumers' valuation of performance improvements lead to changes in firms' innovation incentives, which in turn affect consumers' decisions in
future periods.
The model assumes repeat purchasing behavior
such that the entire population of consumers
considers making a single unit purchase in every
period. Consumer preferences are stable over
time, but, consumer purchase decisions may vary
as product offerings change over time. The
model assumes that there are no switching costs,
consumption externalities, capacity constraints,
economies of scale, or price discrimination.
Consumers are assumed to have a one period
decision horizon and to be well informed, with
perfect information about product performance.8
Given a choice among a set of acceptable products,
a consumer will select the product that meets her
thresholds, improves her utility over her previous
purchase, and maximizes her utility in the present
period. In every period the entire population of
consumers is exposed to the available products and
each consumer selects her own best choice.
Consumer choice
MODEL STRUCTURE
The model follows in the Lancastrian tradition of
conceptualizing the market space along attribute
dimensions. It extends previous applications of
characteristics demand models by examining how
the relationship between the value trajectories of
discrete market segments affects firms' development choices throughout a technology's evolution
and how these factors affect the emergence of competitive dynamics. The model is used to examine
the behavior of two firms pursuing independent
technology initiatives in a market with two consumer segments.
The model structure has two basic components:
a characterization of consumers and consumer
preferences, which comprises the market, and a
mechanism by which products move through this
market space. The market space is defined by two
functional dimensions and by a price dimension.7
In every period, single-product firms make development decisions that affect the location of their
product technology in this space; consumers, in
turn, respond to the product offerings by purchasing either a unit quantity of the product or making
no purchase at all.
7 The model is presented using two functional dimensions, but
can be extended in a straightforward manner to incorporate
higher-dimensional spaces.
Copyright (O2002 John Wiley & Sons, Ltd.
In the model individual consumers are characterized by their threshold requirements and their relative preferences for improvements beyond these
requirements. Each individual, i, has a net utility threshold, UiO,which a product must meet to
be considered for purchase. An individual derives
utility from a given product offering according to
the functional benefit she derives from the product
and from its price. Functional benefit is determined
by the functionality of the product in excess of the
consumer's functional threshold requirements, Fio,
and by the consumer's relative preference for each
The model makes the simplifying assumption that consumers
are perfectly informed regarding product performance. Because
the interest is in the qualitative pattern of behavior, consumer
uncertaintywould be a relevant factor if it were to affect firms'
development decisions in a nonsystematic way. Given, ex aslte,
no compelling reason to bias consumers' assessment of product
performancein either the positive or negative direction, the error
in the assessment would have to be modeled as a symmetric
distribution around the true value. If consumers are assumed to
be risk neutral, and the errorrandom, the expected actions of the
populations would mirrorthe fully informed case. If consumers
are assumed to be uniformly risk averse (risk seeking), their
functional requirements would shift up (down) to compensate
for the uncertainty. Such a shift, while affecting the absolute
values associated with the observed outcomes, would not affect
the qualitative nature of the results. The more complex case of
heterogeneous risk preferences is beyond the scope of the current
8
.
.
c ,xscusslon.
Strat. Mgmt. J., 23: 667-688 (2002)
Whenare Technologies Disruptive?
dimensionof functionalperformance,y.9 The consumer's trade-offbetween functionalbenefit and
.
.
pRlCe 1S (X.
The marketspace is definedby two functional
dimensions,X and Y. Bij is the functionalbenefit
derivedby consumer,i, with functionalthreshold
requirements
of FiXand Fi,.,fromproductj, which
offers functionalityFj, with componentsFjXand
FJx(Fjx-FiX)Y
Bij = Bi(Fj) =
(Fjy _ Fiy)1
y
+ 1 if (FjX-Fi X),
Ootherwise
(1)
whereO < y < 1.
The utility, Uij, that consumeri derives from
productj is specified as a Cobb-Douglasutility
functionwhich tradesoff productprice and functionalityl°:
Uij = Ui(Fj, Pj) = (Bij)a(l/Pj)l-
(Bij = 1), is:
P = (U )1/(a-l)
(3)
Pij,
the maximum price a consumer would be
willing to pay for a product that exceeds his
functionalityrequirements,is thus:
Pij
=
Pio(Bij)a/(l
)
(4)
As discussed above, consumersvalue functionality improvementsbeyond their thresholdrequirements, AUij/8Bij
> O, but at a decreasingrate,
9Fio is a vectorwhich specifiesconsumeri's minimumfunctionalattainment
requiredon eachfunctionaldimension.Graphically, it is the positionalong an individual'svalue trajectory
thata productmustpass to becomedecisionrelevant.
° Note that [logU]/( I-a) = a/( 1-a) log B-log P. Hence
the model is a monotonicvarianton the utility functionof
standardverticaldifferentiationmodels U = KB-P
(Tirole,
1988). The currentmodel differsfrom the standardmodel in
that it incorporatesmultiplefunctionaldimensions,minimum
thresholdsforfunctionality
anddiminishingreturnsto functional
improvements.
Copyright C) 2002 John Wiley & Sons, Ltd.
82Uij/8Bi2j< O. This valuation is reflected in
consumers' decreasing willingness to pay for
improvements,(0Pij/0Bii ' °, 02Pii/0B,2i' °),
when a < 0.5.
Demand structure
In the modelthe essentialaspectsof demandstructure derive from the relationshipbetween consumers' relative functionalpreferences,y. Consumers belong to one of two market segments.
All membersof a marketsegment, m, have the
same relativefunctionalpreferencesYm.
The value
of Ymspecifies a marketsegment's value trajectory. Graphically,for two functionaldimensions,
the degree measureof the value trajectoryis 90o
(1 - y).
For two market segments A and B, whose
derivedutilityfromtwo functionalattributesX and
Y is describedby:
(2)
where O < a < 1.
Consumerswill rejectany productthatdoes not
meet both their functionaland net utility thresholds, requiringboth Bij ' 1, and Uij > Uio. Thus
Pio, the price a consumerwill be willing to pay for
a productthatjust meets his functionalthreshold,
675
UA=
(5a)
(FX)YA (FY) 1-YA
UB= (FX)YB ( FY)
(5b)
YB
whereO<yA,yB<l.
Preferenceoverlapis definedas:
Preference overlap = 1-|
YA-YBI
(6)
The greater the preference overlap, the greater the
segments' agreement on the rank ordering of alternatives. When preference overlap is zero, the segments' preferences are entirely divergent, so that
progress along one segment's value trajectory has
no impact on the other segment's assessment of
the product's functional benefit. When preference
overlap is unity, the value trajectories coincide and
the two segments have identical functional preferences, essentially behaving as a single segment.
Preference symmetry, the extent to which one
segment' s preferences project onto the others'
preferences, is defined as:
Preference symmetry-10.5-YAI-10
5-YBI
(7)
When the preference symmetry measure is zero,
preferences are symmetric and unit progress on
each segment's value trajectory yields equal functional benefit to consumers in the other segment.ll
} Implicitin this discussionis the assumptionthat the distance between indifferencecurves is cardinal,which allows
Strat. Mgmt. J., 23: 667-688 (2002)
676
R. Adner
Whenthe preferencesymmetrymeasureis positive
(negative),the projectionof segmentB's valuetrajectory on segment A's value trajectoryis greater
(less) than A's is on B 's, and a unit advance
along B's value trajectorywill yield greaterfunctional benefits to consumersin segment A than
vice versa.
functionalrequirementsof all consumersin the
market.As discussedbelow, the allocationof effort
along functional dimensions, A Fr and A F,., is
determinedaccording to a local search of the
marketopportunitiespresentedto the firm.
A productinnovationaffects productj as:
Fj,+, = (FjX, + AFx), (FJY,+ AFY)
Technologydevelopment
In the model, technology initiatives are characterized by their performanceon each functional
dimensionand by a productioncost. Technologies
are developedby firms. Firmsintroduceproducts
to the marketon the basis of theirtechnologypositions, with the goal of maximizingprofitsin the
currentperiod.Eachfirmis initiallyendowedwith
a technology that has initial functionaland cost
characteristics.In the course of the simulation,
firms develop their technologies in response to
the opportunitiesthey perceivein the marketplace.
Technology development is locally constrained,
and thereforepathdependent,but globally unconstrainedregardingthe absolutelimit of technology
progress.As describedbelow, in everyperiodfirms
can engage in productinnovation,process innovation, or can choose to forgo the opportunity
to innovate. Following the characterizationused
in previous analyticalmodels (Cohen and Klepper, 1996; Klepper, 1996), the effects of product
and process innovationsare reflectedin changes
in product functional performanceand in product cost.
Productinnovationenhancesperformancealong
the functional dimensions by a fixed Euclidean
distance in market space, FPr°d,and leads to a
In the spirit
fixed productioncost increase, CPr°d.
relatively
is
FPr°d
approach,
of an evolutionary
small in comparisonto the range of functional
performancedemandedin the market(maxFiOminFio), such thatnumerousproductdevelopment
attemptsare requiredif a firm is to satisfy the
for comparisons regarding positions and advances along valuation trajectories. While unconventional in classical demand
theory, Lancaster speaks of the desirability of being able to
make quantitative utility comparisons across goods (Lancaster,
1991: 158-159). Because his closed-form models do not allow
for the simultaneous consideration of final utilities for multiple
consumers, he opts for a 'second best' approach which proxy's
resource content for derived utility. The simulation methodology
used to examine the current model allows, in the spirit of Lancaster's stated intent, for the explicit consideration of cardinal
utilities, and such utilities are therefore used.
Copyright (D 2002 John Wiley & Sons, Ltd.
where [(S FX)2 + (S Fy)2]l2 =
Cn
j,t+l
=
tjst
+
(8a)
I
IFPri
n
t
prod
(8b)
The choice of relative allocation of improvement
determines the technology's development trajectory. While this trajectory can be altered in every
period, the firm is charged a development cost
against profits that is proportional to the shift in
trajectory.
Process innovation leaves product functionality
unchanged while lowering the cost of production
by a constant percentage, Ac.. Thus, a process
innovation affects product j as:
Fj,+, = Fj,
(9a)
Cj,,+l = Cj,,(1-Ac)
(9b)
Firmscan pursueone innovationperperiod.A firm
can pursueeither innovationmode in any period
and for as many periodsas desired.Further,there
is no uncertaintyas to the successof an innovation
attempt.'2Firmsarechargedan innovationcost, I,
in every periodin which they choose to innovate.
In the model,firmschooseinnovativeactivityon
search.Firms
the basis of local, profit-maximizing
searchtheir local marketenvironmentsto predict
consumerreactionto changes that are attainable
througha single productor process innovation.
The model assumesthat firms are fully informed
regardingconsumers'responsesto pricingandthat
firms can, given their product'sperformanceand
productioncosts, determinethe price point which
will yield greatestperiodprofit.Firms are unable
to evaluatepotentialconsumerdemandfor changes
'' Clearly, eliminating uncertainty regarding innovative outcomes is a strong simplification. However, as was the case with
consumer uncertainty, because there is no justification, ex ante,
to bias expectations, the addition of an error term would not
affect the qualitative behaviors that are of concern here.
Strat. Mgmt. J., 23: 667-688 (2002)
Whenare TechnologiesDisruptive?
677
The first proceduresof the simulationinitialize
the populationof consumersandthe initialcharacteristicsof the producttechnologies.The consumer
populationis initializedby specifying each consumer'sminimumfunctionalityandutilityrequirements.The marketsegmentsare specifiedby their
value trajectories.The range of consumers'minimum functionalthresholdsis such that approximately 20 productinnovationattemptsare necessary to span the distance between the minimum
requirementsof the least and most demanding
customerin the market.Consumersare randomly
drawnfroma uniformdistributionalongthis range.
Consumerpreferencesand requirementsremain
constant throughoutthe simulation, but buying
behaviorchangesas firms'developmentactivities
change the producttechnologies' characteristics.
Eachfirm'stechnologyis initiallypositionedalong
the valuetrajectoryof one of the marketsegments.
The specificationof value trajectoriesand initial
product technologies is discussed in the results
section.
After initialization,the following sequence of
events is repeateduntilthe marketdynamicsreach
a steady state.l5First, each firm engages in local
searchto deterrninethe profitexpectationsfor each
possible developmentaction. Second, the firms
commit to the activity that will yield the highest
expected profit. Third, every consumerindependently evaluates the available product offerings
and decides on purchaseas above. Finally, marModel analysis
ket outcomes are tallied and firms realize their
results
actualmarketpay-offs.In the representative
The model is used to examine the behavior of
space
is
seeded
with
below
the
market
discussed
two firmspursuingindependenttechnologyinitiadistributed
between
who
are
evenly
250
consumers
tives in a marketwith two consumersegments.
The analysisexploreshow changingthe preference two marketsegments.Individualconsumers'funcoverlapandpreferencesymmetryof the segment's tionalrequirements,IFiolareuniforrnlydistributed
valuetrajectoriesaffectsthe emergenceandevolu- along the range from 5 to 25. Each firm's techtion of competition.The model is analyzedusing nology is initiallyseeded along a segment'svalue
trajectorywith lFjl set at 7 and initial production
a computersimulationprogrammedin Pascal.
cost 1.7. These settingsprovidefor an initial isolationbetweenthe technologies.In the absenceof
some distance between their initial positions the
marketresearch,
13 To the extentthatfirmsengagein additional
firms face identicalmarketlandscapesand therethese effortswould be reflectedin a searchradiusthat would
Such
extendbeyondtheirimmediatedevelopmentopportunities.
fore always engage in convergentcompetition.
marketresearchwouldaffectbehaviorwhenthe relativeprefer= 1,
For all presented results, cr= 0.2, FPr°d
ences of moredistantconsumersdifferfromthe preferencesof
changing
parameter
=
0.1,
Ac
=
0.05.
While
cPr°d
theirlocal counterparts.
valuesaffectsthe absolutemagnitudesandratesof
searchspeaksto invest14 Thismodelof local,profit-maximizing
that result beyond a single developmentaction.l3
Firms are unable to predicttheir rival's development activity.Firmsbecome awareof theirrival's
activitiesthroughtheirreflectionin consumerpurchasingdecisions-consumers who wereexpected
to purchasebut did not. Firms' expectationsof
marketresponseis thus deterrninedby theirproduct's functionalstate,the magnitudeof innovation
thatcan be executedin a single period,the firm's
pricingdecision,and the productoffers of the previous period.'4
In the evaluationof potentialproductinnovation, the firm must determinethe allocationof its
developmentefforts among the functionaldimensions anddeterminea pricepointat which to offer
the productto the market.The firm has a profit
expectationfor each possible action that is determinedby its productioncosts, developmentcosts,
and predictedmarketfeedback.Similarly,the firm
has a profitexpectationfor potentialprocessinnovation by predictingmarketresponseto different
price points given its reduced productioncosts.
The firmcommitsto the innovationactivitythatis
expectedto yield the highest profitsfor the ensuing period.If the expectedprofitsfrominnovation
are not greaterthan the realizedprofitsfrom the
previousperiod,the firmwill forgo innovationfor
the period.
rather
mentsthatare drivenby immediatemarketopportunities
thanthose drivenby visionary,long-termgoals. Thus, it does
not speakto activitieswith very long-terminvestmenthorizons
R&Dor visionarytechnologybetsmade
suchas pharmaceutical
return.
withoutexpectationof any short-term
Copyright
t 2002JohnWiley& Sons,Ltd.
15Steady state is defined as having been reached when combined sales for both firms remain unchanged for 15 consecutive
periods.
Strat. Mgmt. J., 23: 667-688 (2002
678
R. Adner
behaviors,the qualitativepatternsof competition
are robust. Similarly, the qualitativeresults are
robust with regard to the stochastic seeding of
the consumerpopulation.The resultsarepresented
in terms of the development activities of two
firms, Firm 1 and Firm2, that act in a market
spacewithtwo consumersegments,segmentA and
segmentB. At the startof the simulation,Firm l's
technology position is aligned with segment A's
value trajectoryand Firm 2's technologyposition
is alignedwith segment B's value trajectory.
RESULTS
The simulation explores how the structureof
demand influences the emergence and extent of
competitionover consumersegmentsin the market. On the supply side we are interestedin the
technologies' functionaldevelopmentand pricing
decisions, while on the demandside we are interested in the degreeof firms'penetrationinto market segments,in termsof both satisfactionof consumers' requirementsand consumers'actualpurchasingdecisions.
Figure3 presentsa stylizedsummaryof the relationships between preferenceoverlap, preference
symmetryand the competitiveregimes to which
they lead. Holding one value trajectoryfixed, the
figure maps the competitive regimes that arise
as a second value trajectoryis varied.I6 Three
qualitativelydistinctdynamicsemerge duringthe
courseof the analysis.Underdemandconditionsof
low preferenceoverlap,the developmentdynamics lead to competitive isolation, a purepartitioning of the marketbetween the technologies,such
thateach focuses exclusively on its own segment.
As preferenceoverlap increases,isolation breaks
down and the developmentdynamicslead to the
emergence of two distinct classes of competition. Whensegmentpreferencesare symmetricwe
observe competitive convergence, in which each
technology'sdevelopmentis directedat expanding
its appealnot only in its own home marketbut in
its rival's as well. When segmentpreferencesare
asymmetricwe observecompetitive disruption, in
which one firm maintainsdominanceof its home
marketwhile displacingits rival from the rival's
market.
As a technology'sperformanceadvances,it surpasses the requirementsof a growing numberof
16 The objective of the model is to provide a qualitative representation of the competitive patterns of interest rather than a
quantitative replication of particular settings. In this spirit, the
results are presented in terms of three representative demand
configurations and the dynamics. Because the precise (numerical) location of the boundariesbetween the regimes is dependent
on specific parametervalues as well as the random seeding of
the consumer population it is not a focus of this analysis. The
interested reader is referredto Adner and Zemsky (2001)^which
presents closed-form expressions for the breakdownof isolation.
I (Fixed
trajectory)
L
|convergence/
>
/
D
/
|
(
I
,o
L
/
I
a)
D
I
disru
/
/
/
Functional attribute X
Figure 3. Preference relationships and competitive regimes: holding the preferences of one segment fixed (value
trajectory shown in bold), different competitive regimes arise as the relative preferences of the second segment are
varied
Copyright C) 2002 John Wiley & Sons. Ltd.
Strat. Mgmt. J., 23: 667-688 (2002)
x
X
x
Whenare Technologies Disruptive?
679
value trajectory.Becauseof the wide differencein
relativepreferences,technologydevelopmentfollowing one segment's value trajectorydoes not
lead to significantimprovementalong the other
segment's criteria.Firms' developmentand pricing decisions are thus substantiallydetermined
by the requirementsof their home segments.As
developmentproceeds, and functionalattainment
increases,the firms are able to satisfy the functionalrequirementsof the lower end of the foreign
marketsegment.By this point,however,the appeal
of pursuing these customers is limited by two
factors: (1) its rival offers a higher-functionality
product;and (2) to reach these customers,given
the functionalityof the existing alternative,the
firm would need to lower prices well below the
level it is able to charge its existing customers.
Thus, while consumersin the foreignsegmentare
functionallyvisible, theirpricerequirementsmake
themunattractiveto the firmgiven its profitopportunities in its home market.As a result, neither
firmregistersany sales to consumersoutsideof its
home segment(Figure4).
consumers in the home market and, with sufficient preference overlap, it begins to satisfy and
surpass the requirements of consumers in the foreign market. When preferences are asymmetric, the
firm whose home market casts the larger functional
shadow faces greater marginal incentives to pursue consumers outside its home market because its
offer appeals to a larger number of consumers. As
the invading firm pursues consumers at the low end
of its rival's segment with low-priced offerings,
the invaded firm can either defend its position at
the low end through price reductions or focus on
its own high-end consumers with higher price and
performance offers. From the invaded segment's
perspective the appeal of the invading technology,
which offers neither higher performance nor higher
price/performance value, is due to its lower unit
price.
Demand structure and competitive isolation
Competitive isolation, in which each technology
is sold only in its home market, is characteristic when the preference overlap is small. Figure 4
shows representative simulation results for the
low overlap case in which the two market segments have functional preference parameters YA=
0.1 and YB= 0.9. This case characterizes the
PDA-desktop segments for hard disk drives in
the stylized example above. Initially, technology
initiatives satisfy the functional requirements of
only their home segment and development proceeds along a path that matches that segment's
Demand structure and technology competition
As the overlapbetweenvaluetrajectoriesincreases,
the satisfactionof its home segment's functional
demandsleads each firm towardsearliersatisfaction of the functionalrequirementsof consumers
in the foreign segment. Firms face incentives to
pursue higher-endconsumerswithin their home
segment and, simultaneously,are temptedby the
160 140
t
,'
'
120
,,
n
S
,
80
xx
xx
xxx
xxx
xxx
xxx
xxxxxx
xxxxx
xxx
xxx
xxxxxx
xxx
xxxxx
xxx
xx
x
--'
100
xxx
xX
t
x
A
en
- Jxxx
60
40
xxx
t
,x
20
,
x
O
wB_TTswgwwW^.XsWBPggsswlPg"""
10
0
20
40
30
50
60
70
Period
|- - - Firm 1 sales to segment A
1 sales to segment B
|-Firm
Figure 4.
Copyright t 2002 John Wiley & Sons, Ltd.
g Firm 2 sales to segment A |
x Firm 2 sales to segment B |
Isolation dynamics (YA= 0-9, YB= 0.1), sales by segment
Strat. Mgmt. J., 23: 667-688 (2002)
680
R. Adner
presence of a new set of consumers. Because
demandfor functionalitymaturesearlierat lower
functional levels, lower-end consumers provide
the entry point for foreign technologies to capture market share with lower prices. Competition emerges when the volume of relevantconsumersin the foreignsegmentis sufficientto offset
the price reductionsrequiredto attractthese consumers.
Competitive
convergence
Whenpreferenceoverlapis symmetric,both firms
face broadlysimilar marketlandscapes.FigureS
shows representativeresults for market conditions with YA= 0.6 and YB= 0.4. With sufficient
preferenceoverlap and technology development,
one firm finds the marginalconsumerin the foreign segmentthat has the effect of redirectingits
developmentefforts and pricing policies towards
140120
100
invasion.l7The transitionto convergenceis evident in FigureS(a), which shows the sales of each
firmto each segment.Firm2 is firstto penetrateits
rival's segment.Key to Firm2's ability to attract
consumersfrom segmentA, given the presenceof
Firm l's functionallysuperiorproduct(from the
perspectiveof all segment A consumers),is the
consumers'decreasingmarginalutility from performanceimprovementswhich is reflectedin their
decreasingwillingness to pay for improvements
beyondthe minimumrequirements.By period 10,
the performanceprovidedby both firmsmorethan
of theirlowestdoublesthe thresholdrequirements
to pay,
willingness
Decreasing
consumers.
end
whicheffectivelyincreasesthe importanceof price
differentiationas functionalityimproves, allows
17 In the simulationthe specific locationof consumersin the
thefirmwill be first
foreachrun,determines
space,idiosyncratic
to redirectits activitiesto penetrateits rival's segment.Thus,
while the dynamicsare identicalacross all runs, the specific
identityof the firmvariesby run.
xxx>
,
<
\
r
n
:
,/,C
X,80
40
20
O
XI
t|a
-A^,,
,,
>
^.
.
Period
Firm 1 sales to segment B x Firm 2 sales to segment B |
(a) |--- Firm 1 sales to segment A n Firm 2 sales to segment A
4.5
4
3.5
3
2.5
E
@
._
o
2
sL
L
1.5
0.5
o
0
20
10
30
50
40
60
70
80
90
Period
(b)
|-
- Firm 1 performance
x Firm 2 performance
Firm 1 price a Firm 2 price |
by firm
Figure5. Convergencedynamics(YA= 0.6, YB= 0.4): (a) firmsales by segment;(b) priceandperformance
Copyright C) 2002 John Wiley & Sons, Ltd.
Strat. Mgmt. J., 23: 667-688 (2002)
w-...........................
..........
,
r
Whenare Technologies Disruprive?
Firm 2's functionallyinferiorproductto be perceived as the bettervalue by lower-endconsumers
in segment A, whose requirementsare satisfied
by both alternatives.FigureS(b) shows the firms'
price and performancepositionsover time.'8
Firm 2's redirection of innovation efforts
towardspriceaffectsthe innovationincentivesthat
Firm 1 faces. Firm 1 is faced with the option
of matching its rival's price cuts, or holding
back from following suit, exploitingthe presence
of high-end consumerswho value its functional
superiority(along their value trajectory)and who
will support its higher prices. Note that with
every developmentstep, Firm l's productoffer
becomes relevantto a largerset of consumersin
segmentB. Thus, even with a focus on its home
segment's value trajectory,Firm 1's incentives
is measuredas products'
performance
18 In all figuresfunctional
Euclideandistancefromthe origin,IFjl.
681
to pursue segment B customers increase. The
resultingdynamicof competitiveconvergence,in
which both firms pursue each other's markets,
is quite stark in the later periods shown in
FigureS(a).'9
Competitivedisruption
Whenthe preferenceoverlapbetweensegmentsis
not symmetric,progressalong the differentvalue
trajectoriesleads firms to view differentdemand
landscapes.Figure6 shows representativeresults
for marketconditionswith YA= 0.9 and YB= 0-519 The 'trivial' case of convergent competition is when
heterogeneityin eitherthe demandor supplyenvironmentsis
eliminated.Thus,underconditionsof high preferenceoverlap,
the technologiescompete for what increasinglyresemblesa
single marketsegment,giving rise to competitiveconvergence.
Similarly,when the initial positions of the technologiesare
sufficientlyclose, the distinctionbetweenthe two technologies
disappears,also givingrise to competitiveconvergence.
140
120
;"
100
80
j
t
--------
=-_^
60
x
40
qPd
w
20
X
Hn
xXx
oi
"
O
20
0
40
60
Period
80
(a) | - - - in segment A by Firm 1 n in segment A by Firm 2
100
120
in segment B by Firm 1 x in segment B by Firm 2 |
8
. .o
E
o
L
Period
(b)
|
Firm1 performancexFirm2 performance
Firm1 price o Firm2 price|
Figure 6. Disruptiondynamics (YA= 0-9, YB= 0.5): (a) consumers with satisfied functional requirementsby segment;
(b) price and performanceby firm; (c) firm sales by segment; (d) price/performanceand price along segment A's value
trajectory
Copyright (g32002 John Wiley & Sons, Ltd.
Strat. Mgmt. J., 23: 667-688 (2002)
I
682
,
--
t
--.P
\\\
..
nmnS
X
$
R. Adner
-
v
140
120
100
;
|
t
:
*
lN
80
ax
-
60
v
a
-
l
cn
y
40
:
I
"
.- @
/
r
,
|
rcr
i
20
0
20
40
60
80
100
120
o
Period
I- - - Firm 1 saies to segment A n Firm 2 saies to segment A |
Firm 1 sales to segment B x Firm 2 sales to segment B
(c)
0.8
* 40
0.7
*35
0.6
-30
0.5
*25
0.4
, 20
0.3
.=
15
Q
_
<
sn
o
-
O
n
L
0.2
* 10
0.1
-5
20
40
60
80
100
o
L
o
120
Period
Firm1 price/performance
| - - - Firm 1 performance in segment A
I
(d)
n
x
Firm 2 price/performance
l
Firm 2 performance in segment A
Figure6. continued
This case characterizesthe desktopcomputer(segment A)-notebook computer(segment B) markets for harddisk drives discussed above. Under
such asymmetricalpreferenceconditions,progress
along the value trajectoryof segment B leads to
faster naturalprogressalong the value trajectory
of segment A than vice versa.
Figure6(a) shows the numberof consumersin
each segment whose functionalrequirementsare
satisfied by each firm's technology. In period 6,
Firm2's productbecomes functionallyrelevantto
the segmentA consumerwiththe lowest functional
requirements.As shown in Figure6(b), Firm 2
gains its first customerfrom segment A in period
23. This firstsale outsideits home segmentscorrespondsto a price decrease,evidentin Figure6(c),
which shows the price and absoluteperformance,
Fj, of each firm's offering.
Figure6(d) graphs the price/performanceand
performancecurves for each firm's offering as
Copyright (D 2002 John Wiley & Sons, Ltd.
measuredalong segmentA's value trajectory.Significantly,it shows that at the time that Firm2 is
makinginroadsinto segmentA, its productoffers
lower performanceand a worseprice/performance
level thandoes Firm l's product.20
The dynamics
behindthis disruptionare furtherexaminedin the
Discussionsection.
While progress along segment A's value trajectory does eventually allow Firm 1 to satisfy
segment B's functionalrequirements(period 98
in Figure6a), it lags Firm 2 and is effectively
precludedfrom pursuingthe segment.Because of
the preferenceasymmetriesbetween segments A
and B, Firm 1 is fighting an uneven battle for
marketshare. At equal rates of innovation,following segment A's value trajectorywill attract
20 Firm 2's price reduction ultimately provides for a more attractive price/performance offer (still at a lower absolute performance level) in period 74 and beyond but, as seen in the figures,
this transition does not affect its market performance.
Strat. Mgmt. J.. 23: 667-688 (2002)
Whenare Technologies Disruptive?
fewer consumersfrom segmentB thanvice versa.
Further,as Firm2 begins to addressthe functional
requirementsof segmentA, Firm 1 faces pressure
to justify its price to its existing high-end customersand respondsto this competitivethreatby
focusing on providingincreasinglevels of performance. Firm 1 is thus driven furtherand further
upmarketby a combinationof marketincentives
and competitivethreats.
MODELS,SIMULATIONS,AND
REALITY
Drawingon Christensen'srich inductiveresearch,
this paper presents a model that strives to offer
a parsimoniousexplanationfor the dynamics of
technology disruptions.The intent of the model
is to focus on the essential demand-sidedrivers
that shape the emergenceof competition,particularlycompetitivedisruption.As such, the model
does not seek to disputethe importanceof factors
such as resourceallocation,organizationdesign,
or resourcedependence.Rather,by isolating the
effects of demand structureon competition,the
model should aid future research to more precisely considerthe impactof such complementary
factors.
The value of formalizing the model is that
it brings to light implicit links and assumptions
that lie unexplored in the inductive work (cf.
Nelson and Winter,1982; Saloner 1991; Malerba
et al., 1999). Having identifiedwhat he believes
are the key features driving the phenomenon,
the modeler is forced to make explicit the
relationships between these elements. It was
throughthis process of explicationthat many of
the most interestinglinks in the present model
were revealed. For example, in approachingthe
model, it was evidentthat consumers'reservation
prices and productquality improvementswould
be importantfactorsto include. In characterizing
the evolutionary rules of the model, however,
it became clear that these two factors must
be linked how does reservationprice change
with quality improvement? Recognizing the
implicationsof consumers' decreasing marginal
returnsfrom performanceimprovementsfor their
willingness to pay for new productsturnedout
to be one of the most powerful realizations
of the model. Similarly, building the model
required characterizingthe utility functions of
Copyright t 2002 John Wiley & Sons, Ltd.
683
different market segments according to their
relativepreferencesfor functionalattributes.This
posed the novel requirementof relatingsegment
preferencesto each other, and resulted in the
creation of the grammarof preferenceoverlap
and preferencesymmetryto characterizedemand
structure.
The model is a focusingtool thathelps interpret
andorganizethe richnessof empiricalobservation.
In turn, the model's structureand results raise
furtherquestionsfor empiricalexploration e.g.,
what is the actual 'shape' of decreasingmarginal
returns;how does the way firms and consumers
perceive preferenceoverlapaffect their definition
of industrystructureand industryboundaries;are
pricereductionssubjectto decreasingreturnsin the
way functionalimprovementsare;can firmsaffect
the onset of decreasingreturnsto theircustomers?
Even as they provide some logical closure on
puzzlesraisedby earlierobservations,the model's
implicationsprovideimpetusfor furtherempirical
.
.
.
nvestlgatlon.
The model is, by definition, a simplification.
In the presentmodel, the simplifyingchoices are
driven by the goal of creating a baseline model
that isolates the influenceof demandstructureon
the emergenceof competition.While factorssuch
as asymmetricfirm capabilities,asymmetricmarket segment sizes, and economies of scale could
be incorporatedinto the model structure,they
have been excluded from the present discussion
to reduce the complexity involved in interpreting driversof the model results.2'The structure
of the currentmodel does not lend itself to the
explorationof firm-leveldecisions, such as decisions to investin alternativetechnologyopportunities. The incorporationof factorssuch as resource
allocationand resourcedependence,which operate at the intrafirmlevel, would requirea different modeling approachthat pays explicit attention to the internaldynamics of decision making and cognition. While unexaminedhere, the
relationshipsbetween market size, organization
structure,creditallocation,and managerialincentives certainlyoffer a rich context for modeling
explorations.22
'' See Adner and Zemsky (2001) for a game theoretic examination of these relationships.
22 See Sastry (1997) and Gavetti and Levinthal (2000) for illustrative examples of how simulation models can be applied to
explore intraorganizationalchange processes.
Strat. Mgmt. J., 23: 667-688 (2002)
684
R. Adner
DISCUSSION
The model developed in this article explores the
influence of the structureof consumer demand
on technologyrivalry.By examiningheterogeneity in consumerrequirementsandpreferences,and
by relatingthese factors across marketsegments,
the model offers a basic grammarfor characterizing the structureof demand and offers a new
perspectiveon the emergenceof competition.As
technologiesdevelopthey increasinglysatisfyconsumers within their target marketsegments and
may also become relevantto consumersin other
segments.Categorizingdemandstructureaccording to preferenceoverlapand symmetryhighlights
the differentialmarketincentives that firms face
as their technologies advance. When consumers
face diminishingmarginalreturnsto performance
improvements,technologiesthat offer lower relative performanceat lower price become increasingly attractive.In the face of continuedperformance improvements,this dynamicmay lead to a
blurringof the market'sunderlyingheterogeneity
which, in turn,influencesthe emergenceof different competitiveregimes.
The model sheds light on the importantphenomenonof disruptivetechnologies(Christensen,
1997). ConsiderChristensen'sexplanationof why
desktopcomputerusersoptedto purchase3.5-inch
hard disk drives when they offered lower capacvaluethantheir
ity at a worse dollar-per-megabyte
5.25-inchrivals:
Why did the 3.5 inch drive so decisively conquer
the desktop PC market?A standardeconomic guess
might be that the 3.5-inch format represented a
more cost-effective architecture:If there were no
longer any meaningful differentiationbetween two
types of products (both had adequate capacity),
price competition would intensify. This was not
the case here, however. Indeed, computer makers had to pay, on average, 20 percent more per
megabyte to use 3.5-inch drives, and yet they still
[italics in original] flocked to the product. Moreover, computer manufacturersopted for the costlier
drive while facing fierce price competition in their
own product markets. Why?
Performance oversupply triggered a change in the
basis of competition. Once demand for capacity
was satiated, other attributes[size; power consumption], whose performancehad not yet satisfied market demands, came to be more highly valued ...
(Christensen, 1997: 166-167)
But why should desktop users, who had never
before been concernedwith issues such as power
Copyright t 2002 John Wiley & Sons, Ltd.
consumption,andfor whomenergycosts represent
a negligible expense, suddenlychoose to give up
capacityfor reducedenergyuse?The answeris not
immediatelyapparent.
The currentmodel offers an alternativeexplanation that the desktop users were not choosing 3.5-inch hard disk drives due to their new
attributes,but ratherfor their lower price their
lower price as measured not on a dollar per
megabyte basis but ratheron the basis of unit
price. Thatis, consumerswithsufXcientlysatisfied
functional requirementsare more concernedwith
in absoluteprice than with diffierences
diJjoerences
points.
in price/performance
In supportof this proposition,considerFigure7,
which shows the volume weighted averageprice
of harddisk drives from 1984 to 1990, the years
spanningthe critical substitutionperiod for 3.5inch for 5.25-inchdrives.Throughoutthis period,
the unit price of the 3.5-inch drives is below that
of 5.25-inchdrives.
Viewed through a demand-based lens, the
technology dynamics observed in the hard disk
industrycan be interpretedas follows: preference
asymmetriesare fundamentalto shaping firms'
incentives in a way that leads to the technology
dynamics observed in the hard disk drive
industry. Following the argumentsmade here,
the critical factor driving the displacementof
3.5-inch for 5.25-inch hard drives was not the
latent desire of desktop users for smaller, more
energy-efficientdisk drives; rather, it was that
along with these other features, notebook users
valued improvementsin capacity the attribute
consideredmostcriticalby desktopusers.Because
the evaluation criteria of desktop users were
subsumedby the criteriaof notebookusers, the
developersof 3.5-inch technology,following the
notebooksegmentvalue trajectory,quite naturally
began to satisfy users in the desktopsegment.As
such,facinghighervolumesin the broadermarket,
the 3.5-inchfirmshad greaterincentivesto pursue
a low price strategy.
Having achieved sufficientstorage capacity to
meet the minimumrequirementsof some low-end
desktopusers, 3.5-inch technologyfirmsstill had
to contendwith a 5.25-inchtechnologythatoffered
consumers greater absolute capacity at a better
ratio.The currentanalysissugprice/performance
gests thatthe essentialaspectof consumerchoice
which allows for disruptive displacementmay
be consumers' decreasing marginalutility from
Strat. Mgmt. J., 23: 667-688 (2002)
-
400
,,
,,
-
>
'
'
'
"
-
-
w
'
'
-
--
-
-
-
'
'
Whenare TechnologiesDisruptive?
685
1000
900
800
/
700
600
.o
cL
500
300
200
100
O
1985
1984
are fromDisk/Trend Report
Data
1986
198{18
1987
1989
1990
Year
|- - - 3.5" $/unit
5.25" $/unit|
Figure7. Averageunit pricesfor 3.5 inch and5.25 inch harddisk drives, 1984-l990
performanceimprovementsbeyond their requirements,ratherthan a new found appreciationfor
previously marginal attributes.This decreasing
marginalutility translatesinto a decreasingwillingnessto pay for improvements.Thus a critical
factorin consumers'decisions,once theirrequirements are met, is absolute price while 5.25inchtechnologywas cheaperin price/performance
terms,because the absoluteperformanceof 5.25inch driveswas muchhigherthan3.5-inchdrives,
5.25-inch drives were more expensive on a unit
firms should organize for marketentry (Moore,
1991).
At early stages of technology development,
before performanceis sufficient to satisfy conthe issue of priceis irrelevant.
sumerrequirements,
At later stages, when consumers' performance
requirementsare well satisfiedand their willingness to pay for additionalperformanceimprovementshas diminished,performancegainslose their
efficacy as competitive actions. It is during the
middlepartof developmentthatprice/performance
is a criticalfactor.Technologydisruptionsaremore
basis.23
The logic of decreasingmarginalutility com- likely in this laterstage, in which consumersmay
plements the notion of performanceoversupply. be willing to accept a worse price/performance
offering if its absolute price is sufficientlylow.
It presents an even more fundamentalaspect of
consumerchoice in that it accountsfor changing This dynamic, driven by consumers'decreasing
behaviorin the absenceof the introductionof new marginalreturnsfromperformanceimprovements,
attributes.Further,this logic suggeststhatinforma- speaks to the increasingimportanceof price as
tion aboutthe state of demandcan be ascertained technologiessurpassconsumers'requirements.
not only by observing the relative valuationof
This logic informs the response to the quesattributes,but also by observingconsumers'will- tion of 'When are technologies disruptive?'and
ingnessto pay for the totalproductat it evolves and helps distinguish between inferior technologies
improves.It also suggests that the driversof the and disruptivethreats.Christensensuggests that
disruptivetechnology's adoption differ between disruptivethreatscan be recognizedby identifyits initial home segment and the mainstream;a ing the point of intersectionbetweenperformance
differencewhich may have implicationsfor how providedby a new technology and performance
demandedby the existing consumers.This advice
can now be furtherrefined:first, the degree of
23 Christensen states that incumbents have higher cost structures
than entering rivals so that they need higher margins to survive.
preferenceoverlapbetweenthe new technology's
In the context of the model, this speaks to asymmetries in
existing customersand the incumbent'ssegment
firms' initial cost positions and their abilities to engage in costincumbents'
specifiesthe magnitudeof the impactthatsustainreducing process innovations. To the extent that
process innovation options are limited, they will have an even
ing innovationsalong the preferencelines of one
harder time holding on to less demanding consumers and lose
segmentwill have on consumerevaluationin the
market share at the low end of the market.
Copyright d3 2002 John Wiley & Sons, Ltd.
Strat. Mgmt. J., 23: 667-688 (2002)
686
R. Adner
other segment. Second, the degree of preference
asymmetryspecifies firms' differentialincentives
to competefor new marketsegments.Finally,critical to a disruptiveoutcome is the price at which
the invaderoffers its product.The attackingfirm
must have the incentive and ability to offer its
technically inferior, yet nonetheless satisfactory
productat a sufficientlylower unit price to consumers than its rival (e.g., 3.5-inch disk drives
displaced5.25-inch drives, but notebookcomputers have not displaced desktops from their core
marketseven thoughthey offer sufficientprocessing power along with benefits such as smaller
size, lowerweight,andlowerpowerconsumption).
While disruptionis enabled by sufficientperformance, it is enacted by price. Thus, to identify
potentialdisruptions,managersshouldplot notjust
e-provided andperformance-demanded
performanc
curves, but also the price trajectoriesof the competing offers.
A demand-based view of the emergence of
competition
In examiningthe developmentof firmcapabilities
(Wernerfelt,1984;Teece,Pisano,andShuen,1997)
and the natureof firm competition(Porter,1980;
Hamel and Prahalad,1994), the strategyliterature
has tendedto exhibita 'supply-side'bias. Focused
on firms' activities and interactions,the literature
has largely overlooked the role of the demand
environmentin whichthese interactionstakeplace.
The demand context, however, affects both the
immediatesuccess of firms' activities as well as
the nature of future activities. A demand-based
view, focused on how firms' offers are evaluated
in the market, complements the resource-based
view that has focused on how firms create these
offers.
The demand landscapeshapes the opportunity
structurethat firms face and affects individual
firms' incentives to innovate. Firms' innovation
activities, in turn,affect consumers'expectations
and, throughthese expectations,the demandenvironmentfaced by rival firms.Of particularinterest
is how, as a technology's performancebegins to
addressthe needs of consumersin multiplemarket segments, the distinctionbetween these segments is blurred,leading to radical changes in
firms' competitivepositions. By focusing on the
interactionbetween firms and their environments,
a demand-basedview of competitionpresents a
Copyright t 2002 John Wiley & Sons, Ltd.
complementaryapproachto examining many of
the dynamicsthat are of interestto the strategy
field and offers a link between the evolution of
firms' individualactivities and the evolution of
.
.
competltlon.
ACKNOWLEDGEMENTS
I thankJavierGimeno,RebeccaHenderson,Daniel
Levinthal, Marvin Lieberman,ChristophLoch,
SubramanianRangan,ChristopheVan Den Bulte,
Sidney Winter, Peter Zemsky, ChristophZott,
Associate Editor Will Mitchell, the anonymous
SMJ reviewers, and seminar participants at
INSEAD,Harvard,and Informsfor theirvaluable
comments.I am also gratefulto Roger Bohn, the
InformationStorage IndustryCenter at UC San
Diego andDisk/TrendReportfor sharingtheirhard
disk drive data.
REFERENCES
AbernathyW Clark K. 1985. Innovation: mapping the
winds of creative destruction. Research Policx 14:
3-22.
Adner R, Levinthal D. 2001. Demand heterogeneity
and technology evolution: implications for product
and process innovation. Management Science 47(5):
611-628.
Adner R Zemsky P. 2001. Disruptive technologies and
the emergence of competition. INSEAD working
paper.
Afuah A. 2000. How much do your co-opetitors
capabilities matterin the face of technological change?
Strategic ManagementJournal 21(3): 387-404.
Afuah A Bahram N. 1995. The hypercube of innovation.
Research Policy 14: 3-22.
Amit R, Schoemaker PJH. 1993. Strategic assets and
organizational rent. Strategic Management Journal
14(1): 33-46.
Barney JB. 1986. Strategic factor markets: expectations,
luck, and business strategy. Management Science
32(10): 1231-1241.
Becker GS. 1962. Investmentin humancapital: effects on
earnings. Journal of Political Economy 70(0ctober):
9-49.
Bower JL Christensen C. 1995. Disruptive technologies:
catching the next wave. Harvard Business Review
73(1): 43-53.
Christensen CM. 1992. Exploring the limits of the technology S-cure. Production and Operations Management 1(4): 334-366.
Christensen CM. 1997. The Innovator's Dilemma.
HarvardBusiness School Press: Boston, MA.
Strat. Mgml. J. 23: 667-688 (2002)
Whenare Technologies Disruptive?
Christensen CM, Rosenbloom R. 1995. Explaining the
attacker's advantage: technological paradigms, organizational dynamics, and the value network. Research
Policy 24(2): 233-257.
Cohen WM, Klepper S. 1996. Firm size and the nature
of innovation within industries: the case of process
and product R&D. Review of Economics and Statistics
78(2): 232-243.
Cooper A, Schendel D. 1976. Strategic responses to
technological threats. Business Horizons 19: 61-69.
David PA. 1969. A contribution to the theory of
distribution.Research memorandumno. 71. Research
Center in Economic Growth, Stanford University.
Day GS. 1990. Market Driven Strategy: Processes for
Creating Value. Free Press: New York.
Dosi G. 1982. Technological paradigmsand technological
trajectories.Research Policy 11: 147- 162.
Foster R. 1986. Innovation, the Attacker's Advantage.
Simon and Schuster: New York.
Gavetti G, Levinthal D. 2000. Looking forward and
looking backward: cognitive and experiential search.
AdministrativeScience Quarterly 45(1): 113- 137.
GranovetterM. 1978. Threshold models of collective behavior. American Journal of Sociology 83:
1420- 1443.
Green PE, Wind Y. 1973. Multiattribute Decisions in
Marketing:A MeasurementApproach. Dryden Press:
Hinsdale, IL.
Griliches Z. 1961. Hedonic Price indexes for automobiles: an econometric analysis of quality change.
In Price Statistics of the Federal Government.
National Bureau of Economic Research: New York;
137-146.1tauthqueryQ4Pagenumbers?
Hamel G, PrahaladCK. 1994. Competingfor the Future.
HarvardBusiness School Press: Boston, MA.
Hannan MT, Freeman J. 1977. The population ecology
of organizations. American Journal of Sociology 82:
929-964.
Henderson R. 1995. Of life cycles real and imaginary:
the unexpectedly long old age of optical lithography.
Research Policy 24: 631 -643.
Henderson RM, Clark KB. 1990. Architectural innovation: the reconfigurationof existing product technologies and the failure of established firms.Administrative
Science Quarterly 35(1): 9-30.
Kamien MI, Schwartz NL. 1982. Market Structure and
Innovation. Cambridge University Press: New York.
Kim WC, Mauborgne RA. 1997. Value innovation: the
strategic logic of high growth. Harvard Business
Review 75(1): 102- 113.
Klepper S. 1996. Entry exit, growth and innovation over
the product life cycle. American Economic Review
86(3): 562-583.
Lancaster KJ. 1979. Variety, Equity, and Efficiency.
Columbia University Press: New York.
Lancaster KJ. 1991. Modern Consumer Theory. Edward
Elgar: Brookfield, VT.
Leonard-BartonD. 1992. Core capabilities and core
rigidities: a paradox in managing new product development. Strategic ManagementJournal, Summer Special Issue 13: 111-125.
Copyright C) 2002 John Wiley & Sons, Ltd.
687
Levinthal DA.1998. The slow pace of rapidtechnological
change: gradualism and punctuation in technological
change. Industrial and Corporate Change 7(2):
217-249.
Lynn GS, Morone JG, Paulson AS. 1996. Marketing and
discontinuous innovation: the probe and learn process.
CaliforniaManagement
Review38(3): 8 - 37.
MacMillan I, McGrath R. 2000. Technology strategy in
lumpy market landscapes. In Whartonon Managing
Emerging Technologies, Day G, Schoemaker PJH
(eds). John Wiley: New York; 150- 171.
Malerba F. 1985. Demand structure and technological
change: the case of the European semiconductor
industry. ResearchPolicy 14(5): 283 -297.
Malerba F, Nelson R, Orsenigo L, Winter S. 1999.
History friendly models of industryevolution: the case
of the computer industry. Industrialand Corporate
Change8: 3-40.
McFadden D. 1986. The choice theory approach
to marketing research. MarketingScience 5(Fall):
275-298.
Meyer RJ, Kahn BH. 1991. Probabilistic models of
consumer choice. In Handbook
of Consumer
Behavior,
Robertson TS, KassarjianHH (eds). Prentice-Hall:
Englewood Cliffs, NJ; 85- 123.
Meyer R, Johnson EJ. 1995. Empirical generalizations in
the modeling of consumer choice. MarketingScience
14(3): Part 27: G180-G189.
Moore GA. 1991. Crossingthe Chasm.Harper Business
Press: New York.
Nelson R, Winter S. 1982. An EvolutionaryTheory
of Economic Change. Harvard University Press:
Cambridge, MA.
Pfeffer J, Salancik GR. 1978. The ExternalControlof
Organizations:
A ResourceDependencePerspective.
Harper& Row: New York.
Porter ME. 1980. Competitive
Strategy. Free Press: New
York.
Saloner G. 1991. Modeling, game theory, and strategic
management. StrategicManagement
Journal,Winter
Special Issue 12: 119- 136.
Sastry MA. 1997. Problems and paradoxes in a model
of punctuated organizational change. Administrative
ScienceQuarterly42(2): 237-275.
Smith CG, Cooper AC. 1988. Established companies
diversifying into young industries: a comparison of
firms with different levels of performance. Strategic
Management
Journal9(2): 111- 121.
Smith DK, Alexander RC. 1988. Fumblingthe Future:
HowXeroxInvented,thenIgnored,the FirstPersonal
Computer.W. Morrow: New York.
Sutton J. 1998. Technology
andMarketStructure:Theory
and History.MIT Press: Cambridge, MA.
Teece DJ, Pisano G, Shuen A. 1997. Dynamic capabilities and strategic management. StrategicManagement
Journal18(7): 509-533.
Tirole J. 1988. The Theoryof IndustrialOrganization.
MIT Press: Cambridge, MA.
TrajtenbergM. 1990. Economic Analysis of Product
Innovation:The Case of CT Scanners. Harvard
University Press: Cambridge, MA.
Strat. Mgmt. J., 23: 667-688 (2002)
688
R. Adner
Tripsas M. 2001. Understandingthe timing of technological transitions:the role of user preference trajectories.
HarvardBusiness School Working Paper #02-028.
Tushman ML, Anderson P. 1986. Technological discontinuities and organizationalenvironments.AdministrativeScienceQuarterly31(3): 439 -466.
Tushman M, Romanelli E. 1985. Organizational evolution: a metamorphosis model of convergence and
Behavior,
reorientation.In Researchin Organizational
Vol. 7, Staw B, Cummings L (eds). JAI Press: Greenwich, CN; 171-222.
Copyright (¢ 2002 John Wiley & Sons, Ltd.
Utterback J, Abernathy W. 1975. A dynamic model
of process and product innovation. Omega 3(6):
639-356.
Analysis.W.W.Norton:
VarianH. 1978. Microeconomic
New York.
von Hippel E. 1988.TheSourcesof Innovation.Oxford
University Press: New York.
WernerfeltB. 1984. A resource-based view of the firm.
Journal5(2): 171-180.
StrategicManagement
Strat. Mgmt. J., 23: 667-688 (2002)