the second face of appropriability: … investment made tends to be a first-order or primary goal....
TRANSCRIPT
THE SECOND FACE OF APPROPRIABILITY:GENERATIVE APPROPRIABILITY AND
ITS DETERMINANTS
GAUTAM AHUJAUniversity of Michigan
CURBA MORRIS LAMPERTFlorida International University
ELENA NOVELLICity University London
We distinguish between two forms of appropriability: primary appropriability—effectiveness in exploiting inventions as problem-solving mechanisms and capturinga share of their profits—and generative appropriability—effectiveness in exploitinginventions as concepts and capturing a share of the future inventions they spawn.Recognizing that generative appropriability has two components—cumulative inven-tion and preclusion of others—we identify its key managerially manipulable deter-minants and discuss the implications of the construct for the literature on the re-source-based view and the literature on organizational learning.
The issue of appropriability—that is, howfirms can appropriate value from their inven-tions—represents a fundamental problem instrategy research (e.g., Teece, 1986). In most ap-propriability research, scholars have focused onthe idea that inventions create value for firms bybeing translated into products or by being li-censed out. However, from the time of Arrow(1962), and possibly earlier, researchers haverecognized that there are usually at least twofacets to any invention. On the one hand, aninvention is a solution to some technoeconomicproblem, a source of enhanced utility or lowercost for some set of beneficiaries; on the otherhand, an invention is a concept, an idea thatadds to our universe of concepts and ideas andthat can itself become a seed for future conceptsand ideas. Thus, Arrow’s (1962) abstraction sug-gests that any invention potentially creates twotypes of value: an intrinsic value that relates tothe problem-solving aspect of the invention anda fecundity, or generative, value that relates toits potential as a springboard for future inven-tions (Hopenhayn & Mitchell, 1999).
It follows, then, that for every invention a firmfaces two distinct appropriability issues: (1) how
it can capture the greatest share of profits fromthe problem-solving invention it has devel-oped—that is, benefit from the utility that itsinvention directly creates for a user—and (2)how it can capture the greatest share of futureinventions that are spawned by its inventionand thereby benefit from the new element it hasadded to the universe of ideas (cf. Flem-ing, 2001).
The first type of appropriability, “primary” ap-propriability (PA), refers to a firm’s effectivenessin exploiting a given invention by translating itinto a product or licensable solution for users(e.g., Teece, 1986)—“primary” because for eachgeneration of inventions, the achievement ofprofits and, thus, the realization of a return onthe investment made tends to be a first-order orprimary goal. The second type, “secondary” or(henceforth) “generative” appropriability (GA),refers to a firm’s effectiveness in capturing thegreatest share of future inventions spawned byits existing inventions. Future inventions couldbe enhanced or improved versions of the origi-nal invention (addressing the same needs as theoriginal invention and, hence, potential substi-tutes for it), or derived inventions that use theideas of the original invention in a related butcomplementary market (e.g., iPad tablets build-ing on iPhone) or even in unrelated marketsAll authors contributed equally to this work.
� Academy of Management Review2013, Vol. 38, No. 2, 248–269.http://dx.doi.org/10.5465/amr.2010.0290
248Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyrightholder’s express written permission. Users may print, download, or email articles for individual use only.
(e.g., skin cleanser Clarisonic building on theprinciple of ultrasound vibration embodied inan electronic toothbrush, Sonicare).
The strategy literature has focused mainly onPA. However, it is likely that in contexts whereinventions occur with some frequency and exist-ing products are supplanted fairly often, firmsmay seek both to enjoy the profits associatedwith a given invention as above (PA) but also toensure that their research efforts bear fruit inthe form of multiple sequential inventions (GA)that sustain their value creation over time. Inthis article we elaborate the construct of GA,highlighting its theoretical and managerial dis-tinctiveness and significance; identify key man-agerially manipulable determinants of GA; andexplore its implications for the literature on theresource-based view (RBV) and the literature onorganizational learning.
DEVELOPING AND CLARIFYING THEGA CONSTRUCT: THE DISTINCTION
BETWEEN PA AND GA
Imagine a firm that in Period 1 releases Inven-tion 1. At that time the firm faces a PA and a GAproblem. First, it needs to use the problem-solving aspect of Invention 1 to create a profit,identifying the best way to monetize it—as acommercialized product or through licensing.Second, recognizing the conceptual componentof Invention 1, the firm may use the principlesembodied in the current invention to originatenew inventions. Success in this second endeavorwill lead the firm to release an Invention 2 inPeriod 2 that builds on the concepts embodied inInvention 1. This begins a cycle where the firmhas to find (1) how to translate Invention 2 into astream of financial returns and (2) how to buildon the idea in Invention 2 to come up with In-vention 3 in Period 3 (see Figure 1), and so on.Thus, the two appropriability questions ariseiteratively for an initial invention as well as forany subsequent inventions that originate fromit.1
Consider Apple’s invention of the iPod. Interms of PA, the iPod was an extremely success-
ful product, generating significant profits for thecompany. However, Apple’s GA performance isalso notable. Apple took the iPod and combinedthe basic concept with a telecommunicationfunction to create the iPhone. It also created aset of additional products, such as the iPodNano, Shuffle, and so on, where it took the basicidea of the original iPod and developed newversions of products in which key attributessuch as capacity, size, and weight were modi-fied to fit the needs of users with different pref-erences. Subsequently, it took the basic conceptof the iPhone (in itself a very successful productin terms of PA) and expanded its surface areaand computing functions to create yet anothernew product, the iPad. PA and GA are thus twodistinct outcomes reflecting two different as-pects of a firm’s invention performance.
To highlight the added value of GA as a con-struct, we note that building successfully onprior inventions is important for firms for multi-ple reasons. First, initial iterations of most in-ventions are relatively rudimentary. Subsequentversions improve significantly and often havefar greater commercial viability than the origi-nal. Second, even when inventions become com-mercially viable, fully exploiting a firm’s invest-ments in research often entails being able tocreate subsequent generations of inventionsthat are increasingly differentiated for specificuses and market segments (e.g., iPod Nano,Shuffle) or that offer better price/performancevalues (McEvily & Chakravarthy, 2002; Roberts,1999). Third, the most valuable application of agiven invention is often not the application forwhich it was originally conceived. Coupling theconceptual kernel of an invention with newproblems to be solved may then be extremelyvaluable (Hargadon & Sutton, 1997). For in-stance, Corning’s Gorilla Glass product, nowused very successfully for smartphone screens,is derived from a tough glass that was originallydesigned for use in car windshields. Simply put,then, firms need to care about GA to maximizethe returns from their investments in knowledgeand to safeguard against the possibility of be-ing rendered obsolete by improved substitutesthat build on their own prior efforts.
Multiple arguments also suggest that distin-guishing between GA and PA is important. First,factors highlighted in the literature as enhanc-ing appropriability, such as complexity, andcausal ambiguity influence the two forms of ap-
1 Although the GA concept could be expanded to nontech-nological domains in that ideas from marketing or humanresource practices could also be built on to create new ideas,to retain theoretical focus and clarity we limit the scope ofour investigation to technological inventions.
2013 249Ahuja, Lampert, and Novelli
FIG
UR
E1
Prim
ary
and
Gen
erat
ive
App
ropr
iabi
lity
Tim
e
t 1
t 2
New
inve
ntio
n 1
Gen
erat
ive
appr
opri
abili
ty is
sue:
How
can
we
crea
te fu
ture
in
vent
ions
that
are
spa
wne
d fr
om th
is in
vent
ion?
Gen
erat
ive
appr
opri
abili
ty is
sue:
How
can
we
crea
te fu
ture
in
vent
ions
that
are
spa
wne
d fr
om th
is in
vent
ion?
Lead
s to
cho
ices
rela
ted
to p
rofit
crea
tion
in t 1
—i.e
.: Le
ads
to c
hoic
es re
late
d to
pro
fitcr
eatio
n in
t 2—
i.e.:
Lead
s to
cho
ices
rela
ted
to p
rofit
crea
tion
in t n
—i.e
.:
• Pa
tent
or n
ot?
• Li
cens
e or
not
? •
Dev
elop
into
a p
rodu
ct
or n
ot?
Inve
ntio
n 2
(bui
lt on
inve
ntio
n 1)
…
Inve
ntio
n n
(b
uilt
on s
ome
prev
ious
in
vent
ion)
t n
• Pa
tent
or n
ot?
• Li
cens
e or
not
? •
Dev
elop
into
a p
rodu
ct
or n
ot?
• Pa
tent
or n
ot?
• Li
cens
e or
not
? •
Dev
elop
into
a p
rodu
ct
or n
ot?
Prim
ary
appr
opri
abili
tyis
sue:
H
ow c
an w
e cr
eate
pro
fits
from
this
inve
ntio
n?
Prim
ary
appr
opri
abili
tyis
sue:
H
ow c
an w
e cr
eate
pro
fits
from
this
inve
ntio
n?
Prim
ary
appr
opri
abili
tyis
sue:
H
ow c
an w
e cr
eate
pro
fits
from
this
inve
ntio
n?
Ima
gin
ea
firm
tha
tin
Per
iod
1re
lea
ses
an
Inve
nti
on1.
Att
ha
ttim
eth
efi
rmfa
ces
ap
rim
ary
an
da
gen
era
tive
ap
pro
pri
ab
ilit
yp
rob
lem
,th
efi
rstb
ein
g“H
owca
nw
ecr
eate
prof
its
from
this
inve
nti
on?”
an
dth
ese
con
db
ein
g“H
owca
nw
ecr
eate
futu
rein
vent
ions
tha
ta
resp
aw
ned
from
this
inve
nti
on?”
Ifth
efi
rmsu
cces
sfu
lly
an
swer
sb
oth
qu
esti
ons,
then
inP
erio
d1
itw
ill
earn
ap
rofi
tfr
omIn
ven
tion
1,b
ut
inP
erio
d2
itw
ill
rele
ase
an
Inve
nti
on2
tha
tb
uil
ds
onth
eco
nce
pts
emb
odie
din
Inve
nti
on1.
Bu
ton
cea
ga
inth
efi
rmw
ill
ha
veto
thin
ka
bou
t(1
)how
totr
an
sla
teIn
ven
tion
2in
toa
stre
am
offi
na
nci
al
retu
rns
thro
ug
hp
rod
uct
emb
odim
ent
orli
cen
sin
ga
nd
(2)h
owto
bu
ild
onth
eid
eain
Inve
nti
on2
inor
der
toco
me
up
wit
hIn
ven
tion
3in
Per
iod
3.
propriability differently, making it unlikely thatthe same managerial choices will maximizeboth forms of appropriability (see Table 1). Sec-ond, PA implies focusing immediately and di-rectly on obtaining rents, whereas GA impliesinvesting in creating a future opportunity forrents. Thus, the two forms of appropriabilitysuggest very different (and potentially conflict-ing) foci for organizational resource allocation,systems, staffing, and attention. For instance,maximizing PA entails attention to commercial-ization concerns and requires focusing on theneeds of a given set of existing users and opti-mally configuring the invention to those needs(Christensen, 1997; Christensen & Bower, 1996).In contrast, enhancing GA pushes attention to-ward inventiveness and requires thinking abouthow different groups of potential users mightfind the principle underlying the invention use-ful, or thinking about how that principle may beused to respond to other needs. Failing to explic-itly recognize this trade-off and the accompany-ing distinctions may lead to a loss of competi-tive position. For instance, after its acquisitionby News Corporation, Myspace significantly fo-cused on realizing advertising revenue insteadof allocating resources to develop improvedproducts that incorporated features that couldlead to an enhanced experience for the user.This choice was controversial within the com-pany and even blamed for its decline, as com-petitors constantly upgraded and enhancedtheir offerings but Myspace did not (Gil-lette, 2011).
IDENTIFYING THE DETERMINANTS OF GA
The construct of GA has not been used in theliterature before; however, in three streams ofresearch, scholars have examined related con-cepts: (1) the RBV literature on threats from sub-stitution and imitation (e.g., King, 2007; King &Ziethaml, 2001; McEvily & Chakravarthy, 2002),(2) the organizational learning literature on ex-ploitation and reuse of knowledge (e.g., Flem-ing, 2001; Majchrzak, Cooper, & Neece, 2004), and(3) the economics literature on sequential inven-tion (e.g., Hopenhayn, Llobet, & Mitchell, 2006).Table 2 presents an analysis of how GA and theconstructs emerging from previous literature(imitation, substitution, exploitation, knowledgereuse, and sequential invention) are differenti-ated across multiple dimensions.
For any firm, GA is likely to be driven by twocomponents: (1) the cumulative invention com-ponent, defined as firms’ effectiveness in creat-ing new inventions that build on their own ex-isting inventions, and (2) the preclusivecomponent, defined as firms’ effectiveness inpreventing others from building inventions basedon the firms’ inventions. The two components ofGA can but do not necessarily covary. Firms canoutperform on both components, underperformon both components, or perform well on one andpoorly on the other. For instance, Xerox has cre-ated many inventions but has not necessarilybeen successful in building on many of them(Chesbrough, 2002) or in preventing others fromdoing so. As previously mentioned, Apple is anexample of a firm that has managed to be cre-ative in both building on its own inventions (e.g.,the iPhone and iPad based on the iPod) and inprecluding (or at least delaying) other compa-nies from successfully building on its ideas (Bur-rows, 2009). Conversely, Epilady, the originatorof pull epilators, is an example of a companythat was high on the cumulative invention com-ponent of GA but low on the preclusive compo-nent (Geer, 1990).
Researchers in knowledge management andcognitive psychology have noted that breakinga concept into components and studying thecomponents separately reduces the scope of theproblem being analyzed and, thus, simplifiesthe search for identifying underlying relation-ships (Goldenberg, Mazursky, & Solomon, 1999).Using this approach, we distinguish betweenthe cumulative and preclusive component ofGA, identify determinants of the construct, andarticulate propositions relating GA to these de-terminants. A key assumption we build on isthat a firm’s GA outcomes are likely to dependon access to and usage of the inventive knowl-edge of the firm. Enhancing the access and us-age of the firm’s inventive knowledge by inven-tors inside the firm and reducing the access andusage of the firm’s knowledge by inventors out-side the firm are then natural paths to enhanc-ing GA.
We adopt a strategic, firm-based perspective.In uncovering the determinants of GA, we focusour attention on key levers that lie within thecontrol of management and are commonly usedto influence organizational outcomes: organiza-tion structure, organizational systems and pro-cesses, and organization strategy (Grant, 2007).
2013 251Ahuja, Lampert, and Novelli
TABLE 1Primary Appropriability versus Generative Appropriability
Key ComparisonDimensions Primary Appropriability (PA) Generative Appropriability (GA)
Domain of construct Profits/economic rents (e.g., Teece, 1986): PAis defined in terms of the firm’s share ofprofits from its inventions.
Ideas/inventions: The currency of GA isideas; GA is defined in terms of thefirm’s share of the ideas or inventionsthat are spawned by its earlierinventions.
Relationship betweenconstruct and firmperformance
PA influences the relationship between aninvention and firm financialperformance (i.e., revenues and profitsearned by the firm). Only if the firm isable to translate the invention into avalue proposition for a customer (be itanother firm or the final consumer) canit commercialize the invention andcreate financial returns from it.However, PA is not directly related tofirm invention performance (defined asthe rate of inventions developed by thefirm): a firm with a higher PA willappropriate a higher share of revenuesand profits from each invention butwill not necessarily come up with newinventions.
Conversely, GA (defined as the firm’ssuccess in building on its own priorinventions) influences firm inventionperformance by increasing the rate ofinventions generated from the firm’sprior inventions but may not be directlyrelated to firm financial performance inthe short run.
Possible measures Profits from a new product; incrementalprofits made possible by new processes;licensing income from new products orprocesses
The proportion of inventions/productsspawned by the focal firm’s inventionsthat are created by the focal firm; of thetotal citations of the firm’s patents, theproportion that come from the firm’sown patents
Primary effect of patentingan invention
Generally increases PA for that invention,although the strength of the benefitvaries in different industries, for differenttypes of inventions (e.g., process versusproducts), and under different patentregimes (e.g., Cohen, Nelson, & Walsh,2000; Levin, Klevorick, Nelson, & Winter,1987; Teece, 1986)
May decrease GA because the doctrine ofenabling disclosure enshrined in patentlaw makes building on the focal firm’sinventions easier for others; patent lawdemands that patent disclosures shouldenable a person skilled in the art tocreate the claimed invention
Primary effect of keepingthe invention secret
Generally increases PA, although the effectmay vary across industries and contexts(e.g., Cohen et al., 2000; Levin et al., 1987;Teece, 1986); in a relative sense, though,in many contexts secrecy may leave thefocal firm with weaker protection thanpatents—if competitors figure out how theinvention works and the focal firm hasnot patented it, they could imitate it moreeasily
Improves GA—e.g., by diminishing thelikelihood that somebody else willeasily learn about the innards of afirm’s new product or process
Primary effect of weakproperty rights
Reduces PA by reducing the inventor’sability to prevent imitation by enforcingproperty rights: competitors can moreeasily imitate the firm’s extant inventionswith a reduced risk of being sued (e.g.,Teece, 1986)
Reduces GA by reducing theinventor’s ability to preventimitation by enforcing property rights:competitors can more easily build onthe firm’s extant inventions togenerate new inventions with areduced risk of being sued (effect ofweak property rights is stronger on PAthan on GA)
(Continued)
252 AprilAcademy of Management Review
We recognize that GA is likely to be driven bymany additional factors, including context-specific and idiosyncratic factors such as theintrinsic fecundity potential and complexity in-herent in a given invention, but we leave theseto be identified by future work. Moreover, evenwithin our chosen set of managerial levers, thelist of predictors we identify is illustrative ratherthan exhaustive (see Table 3).
THE CUMULATIVE COMPONENT OF GA
GA and Nearly DecomposableOrganization Structures
Enhancing cumulative invention requires thattwo goals be accomplished simultaneously—that is, that invention be enhanced within thefirm and that the inventions generated build onprior ones. Organization structure, in particularthe decentralization versus centralization pat-tern (Nickerson & Zenger, 2002) of research unitswithin the organization, shapes the primarypaths of information flow inside the firm (Sig-gelkow & Levinthal, 2003) and, thus, plays a keyrole in determining the extent to which the or-ganization succeeds in cumulative invention.
Research activity in an organization can beconducted through multiple decentralized sub-groups or units or through a single centralizedone (Argyres & Silverman, 2004). From the per-spective of enhancing invention and cumula-tiveness, these alternatives both imply a trade-off. Decentralized structures foster invention buthinder cumulativeness, whereas centralizedstructures foster cumulativeness but hinder in-vention. Decentralized structures—modular orfully decomposable in Simonian terms (Sanchez& Mahoney, 1996)—permit each subgroup to spe-cialize in a limited domain of knowledge andbecome highly sensitive and responsive toneeds in that domain (Nickerson & Zenger, 2002;Yayavaram & Ahuja, 2008). This depth of knowl-edge of the domain and the sensitivity to cus-tomer needs within that domain enhance theli-kelihood of breaching the knowledge frontier,fostering invention (Katila & Ahuja, 2002). Yetdecentralization hinders cumulativeness in in-vention. Decentralized structures provide in-adequate coordination across subgroups(Nickerson & Zenger, 2002), since individualsubgroups tend to be isolated from the cre-ations and knowledge flow of other subgroups.Consequently, many potential inventions
TABLE 1(Continued)
Key ComparisonDimensions Primary Appropriability (PA) Generative Appropriability (GA)
Primary effect of complexity Generally increases PA by making imitationdifficult (because it increases the difficultyfor the imitator to comprehend how thesystem works; McEvily & Chakravarthy,2002)
Overall may not enhance GA becausecomplexity may make building oninventions more difficult both from theinventor’s and the competitors’perspectives
Primary effect of causalambiguity
Generally increases PA:Interfirm, generally increases PA of inventor
by making imitation difficult, since thepotential imitator does not understandwhat and how to copy (e.g., King, 2007;Lippman & Rumelt, 1982; Reed &DeFillippi, 1990)
Intrafirm, usually will not affect PA sincethe inventor can profit from its inventionseven if it does not understand theinternal linkages that led to the creationof the invention; where understanding thelinkages in the causal process ofinventing is necessary to the invention’scommercial success, it could reduce PA
May or may not increase GA:Interfirm, enhances GA by making
subsequent invention by competitorsdifficult, since the competitor may findit difficult to understand the innards ofthe focal firm’s prior inventions
Intrafirm, may reduce GA—e.g., byimpeding the understanding of “the linkbetween a competency and itsperformance outcomes” and by blockingthe “ability to learn about and adaptthat competency” (King, 2007: 168),which is required to build on priorinventions (e.g., King, 2007; McEvily,Das, & McCabe, 2000; Reed & DeFillippi,1990)
2013 253Ahuja, Lampert, and Novelli
TABL
E2
Con
stru
ctD
isti
ncti
vene
ssfr
omR
elat
edC
onst
ruct
s
Con
stru
ct/
Out
com
eD
efin
itio
nof
Con
stru
ct
IsTh
ere
aSe
cond
-O
rder
“Inv
enti
on”
Req
uire
dfo
rth
eC
onst
ruct
toBe
Mea
ning
ful?
Sour
ceof
Seco
nd-
Ord
erIn
vent
ion
orPr
oduc
t
Scop
eof
App
lica
tion
ofSe
cond
-Ord
erIn
vent
ion
Sour
ceof
Kno
wle
dge
Buil
ton
inSe
cond
-O
rder
Inve
ntio
nor
App
lica
tion
(inca
seit
isno
tan
inve
ntio
n)
Pers
pect
ive
Ado
pted
byLi
tera
ture
/Nat
ure
ofSo
luti
onto
Add
ress
the
Prob
lem
Illu
stra
tive
Art
icle
sH
ighl
ight
ing
the
Con
stru
ct
Gen
erat
ive
appr
opri
abil
ity
(GA
)
GA
refe
rsto
afi
rm’s
effe
ctiv
enes
sin
cap
turi
ng
ash
are
ofth
efu
ture
inve
nti
ons
spa
wn
edb
yit
sex
isti
ng
inve
nti
ons.
Yes
—T
oob
tain
GA
the
orig
ina
lin
ven
tin
gfi
rmh
as
tocr
eate
an
ewin
ven
tion
ba
sed
onit
sea
rlie
rin
ven
tion
s.
Foc
al
firm
orot
her
firm
s—If
the
foca
lfi
rmis
the
sou
rce
ofth
ese
con
d-o
rder
inve
nti
ona
nd
oth
erfi
rms
ha
ven
otb
een
ab
leto
dev
elop
seco
nd
-or
der
inve
nti
ons
bu
ild
ing
onth
efo
cal
firm
’sin
ven
tion
s,th
efo
cal
firm
ha
sh
igh
GA
.If
oth
erfi
rms
ha
vesu
cces
sfu
lly
bu
ilt
onth
efo
cal
firm
’sin
ven
tion
s,th
efo
cal
firm
ha
slo
wG
A.
Sa
me
ma
rket
ord
iffe
ren
tm
ark
ets—
Th
ese
con
d-o
rder
inve
nti
onco
uld
be
asu
bst
itu
tefo
rth
efo
cal
pro
du
ct(e
.g.,
an
imp
rove
dve
rsio
nof
it),
targ
etin
gth
esa
me
ma
rket
,or
itco
uld
be
aco
mp
lem
enta
ryor
un
rela
ted
pro
du
ctta
rget
ing
dif
fere
nt
ma
rket
s.
Foc
al
firm
—F
orG
A,
the
seco
nd
-ord
erin
ven
tion
bu
ild
son
the
con
cep
tsem
bod
ied
inth
efo
cal
firm
’sp
rece
din
gin
ven
tion
s.
Foc
al
firm
’sp
ersp
ecti
ve—
Th
efo
cus
ison
firm
-le
vel
solu
tion
sto
the
pro
ble
mof
GA
.S
ug
ges
ted
solu
tion
sto
incr
ease
GA
are
the
stra
teg
ies
outl
ined
inth
ea
rtic
le,a
llof
wh
ich
are
wit
hin
the
con
trol
offi
rms.
Th
isa
rtic
le
Imit
atio
nR
efer
sto
con
scio
usl
ya
ssoc
iati
ng
elem
ents
com
mon
toth
eob
serv
ab
lesu
cces
sof
ag
iven
firm
an
dre
pli
cati
ng
them
(Alc
hia
n,1
950)
No—
Imit
ati
onca
noc
cur
wh
enth
ese
con
dfi
rmsi
mp
lyre
pli
cate
sth
eor
igin
al
inve
nti
on.
Th
ere
isn
ore
qu
irem
ent
tha
ta
nim
ita
tor
bu
ild
an
ewin
ven
tion
ba
sed
onth
eor
igin
al
inve
nti
onto
talk
ab
out
imit
ati
on.
Sec
ond
-ord
erin
ven
tion
isn
otre
qu
ired
.Im
ita
tion
isw
hen
aco
mp
etit
orre
pli
cate
sa
nex
isti
ng
pro
du
ctor
serv
ice.
No
nov
else
con
d-o
rder
inve
nti
onis
crea
ted
.
Sa
me
ma
rket
—In
the
con
text
ofim
ita
tion
,th
eco
py
ista
rget
eda
tth
esa
me
ma
rket
as
the
orig
ina
lin
ven
tion
.
Foc
al
firm
—T
ob
eca
lled
imit
ati
on,
the
cop
yn
eed
sto
bu
ild
onth
efo
cal
firm
’skn
owle
dg
e.
Foc
al
firm
’sp
ersp
ecti
ve—
Ca
usa
la
mb
igu
ity,
com
ple
xity
,sca
le,
etc.
are
key
mec
ha
nis
ms
tha
tp
rote
ctth
efi
rmfr
omim
ita
tion
.
Die
rick
x&
Coo
l(1
990)
;Kin
g&
Zeit
ha
ml
(200
1);
Lip
pm
an
&R
um
elt
(198
2);R
eed
&D
eFil
lip
pi
(199
0);
Riv
kin
(200
1);
Tee
ce(1
986)
Subs
titu
tion
Ref
ers
toth
ed
evel
opm
ent
ofa
nin
ven
tion
fun
ctio
na
lly
equ
iva
len
tto
the
foca
lin
ven
tion
Yes
—S
ub
stit
uti
onoc
curs
wh
ena
nin
ven
tion
fun
ctio
na
lly
recr
eate
sth
ep
rod
uct
orse
rvic
eb
enef
its
ofa
nex
isti
ng
inve
nti
on.
Th
ea
uth
orof
the
seco
nd
-ord
erin
ven
tion
isth
esu
bst
itu
tor.
Foc
al
firm
sor
oth
erfi
rms—
Th
efo
cal
firm
can
itse
lfco
me
up
wit
ha
sub
stit
uti
ng
inve
nti
on,o
rsu
cha
nin
ven
tion
can
be
crea
ted
by
an
oth
erfi
rm.
Sa
me
ma
rket
—In
the
con
text
ofsu
bst
itu
tion
,th
ese
con
din
ven
tion
ista
rget
eda
tth
esa
me
ma
rket
as
the
orig
ina
lin
ven
tion
.
Foc
al
firm
orot
her
firm
s—T
he
sub
stit
uti
ng
inve
nti
onco
uld
be
ba
sed
onth
ekn
owle
dg
eof
the
foca
lfi
rmth
at
wa
sth
ecr
eato
rof
the
firs
tin
ven
tion
orco
uld
be
ba
sed
onth
ekn
owle
dg
eof
oth
erfi
rms.
Foc
al
firm
’sp
ersp
ecti
veM
cEvi
ly&
Ch
akr
ava
rth
y(2
002)
;McE
vily
,D
as,
&M
cCa
be
(200
0)
Expl
oita
tion
Exp
loit
ati
onco
nce
rns
the
refi
nem
ent
an
du
seof
afi
rm’s
exis
tin
gca
pa
bil
itie
sa
nd
know
led
ge
ba
ses,
wh
ile
exp
lora
tion
isre
late
dto
the
crea
tion
ofn
ewon
es(L
evin
tha
l&
Ma
rch
,198
1;M
arc
h,1
991;
Win
ter,
1971
).
No—
Exp
loit
ati
onm
ay
not
invo
lve
inve
nti
ona
ta
ll.
For
inst
an
ce,
exp
loit
ati
onm
ay
enta
ilex
pa
nd
ing
sale
sof
an
exis
tin
gp
rod
uct
from
one
geo
gra
ph
yto
an
oth
er.
Foc
al
firm
Sa
me
ma
rket
(e.g
.,B
enn
er&
Tu
shm
an
,200
3;M
arc
h,1
991)
,in
gen
era
l—E
xpa
nd
ing
toa
dif
fere
nt
ma
rket
ma
yb
eco
nsi
der
edex
plo
rati
on.
Foc
al
firm
—E
xplo
ita
tion
imp
lies
usi
ng
afi
rm’s
exis
tin
gca
pa
bil
itie
s.
Foc
al
firm
’sp
ersp
ecti
veB
enn
er&
Tu
shm
an
(200
3);M
arc
h(1
991)
Kno
wle
dge
reus
eK
now
led
ge
reu
sere
fers
tow
hen
know
led
ge
acq
uir
edin
one
situ
ati
onis
ap
pli
edto
an
oth
ersi
tua
tion
(Ma
jch
rza
k,C
oop
er,&
Nee
ce,
2004
).
No—
Kn
owle
dg
eco
uld
be
reu
sed
ina
dif
fere
nt
con
text
wit
hou
t,n
eces
sari
ly,t
he
crea
tion
ofa
new
inve
nti
on.F
orin
sta
nce
,afi
rmco
uld
use
its
rou
tin
esd
evel
oped
from
run
nin
ga
nof
fice
inth
eU
nit
edS
tate
sto
run
an
offi
cein
Eu
rop
e.
Foc
al
firm
orot
her
sS
am
eor
dif
fere
nt
ma
rket
No
dis
tin
ctio
nm
ad
eb
etw
een
the
know
led
ge
ofth
efo
cal
firm
an
dth
at
ofot
her
s
Foc
al
firm
’sp
ersp
ecti
veF
lem
ing
(200
1);
Ma
jch
rza
ket
al.
(200
4);M
urr
ay
&O
’Ma
hon
y(2
007)
.
(Con
tin
ued
)
TABL
E2
(Con
tinu
ed)
Con
stru
ct/
Out
com
eD
efin
itio
nof
Con
stru
ct
IsTh
ere
aSe
cond
-O
rder
“Inv
enti
on”
Req
uire
dfo
rth
eC
onst
ruct
toBe
Mea
ning
ful?
Sour
ceof
Seco
nd-
Ord
erIn
vent
ion
orPr
oduc
t
Scop
eof
App
lica
tion
ofSe
cond
-Ord
erIn
vent
ion
Sour
ceof
Kno
wle
dge
Buil
ton
inSe
cond
-O
rder
Inve
ntio
nor
App
lica
tion
(inca
seit
isno
tan
inve
ntio
n)
Pers
pect
ive
Ado
pted
byLi
tera
ture
/Nat
ure
ofSo
luti
onto
Add
ress
the
Prob
lem
Illu
stra
tive
Art
icle
sH
ighl
ight
ing
the
Con
stru
ct
Sequ
enti
alin
vent
ion
Inve
nti
onor
der
iva
tive
imp
rove
men
tsb
ase
don
an
init
ial
inve
nti
on(G
reen
&S
cotc
hm
er,1
995;
Sco
tch
mer
&G
reen
,199
0)
Yes
—W
ith
out
asu
ccee
din
gin
ven
tion
tha
tb
uil
ds
onth
ep
rior
inve
nti
ons,
the
con
stru
ctis
not
mea
nin
gfu
l.
Foc
al
firm
orot
her
firm
s(G
reen
&S
cotc
hm
er,1
995)
—E
ith
erca
nb
ea
sou
rce
ofin
ven
tion
sth
at
bu
ild
onp
rior
inve
nti
ons.
Sa
me
ord
iffe
ren
tm
ark
ets
(Gre
en&
Sco
tch
mer
,199
5)—
Th
eli
tera
ture
ofte
nfo
cuse
son
the
sam
em
ark
et.
Foc
al
firm
Soc
ieta
lor
pu
bli
cp
olic
yp
ersp
ecti
ve—
Sol
uti
ons
pre
sen
ted
toth
ep
rob
lem
ofse
qu
enti
al
inve
nti
onin
volv
ea
lter
ing
the
des
ign
ofth
ep
ate
nt
syst
emin
term
sof
bre
ad
tha
nd
len
gth
ofp
ate
nts
—va
ria
ble
sth
at
are
bey
ond
the
con
trol
ofin
div
idu
al
firm
s(G
reen
&S
cotc
hm
er,1
995)
.
Gre
en&
Sco
tch
mer
(199
5);H
open
ha
yn,
Llob
et,&
Mit
chel
l(2
006)
;Sco
tch
mer
&G
reen
(199
0)
that lie at the intersection of the knowledgebases of different subgroups and, hence, buildon prior inventions are never created.
In contrast, centralized structures—nonde-composable in Simonian terms—foster buildingcumulatively on a firm’s past inventions butmay reduce inventiveness in the first place. In acentralized research structure, decision makingand authority are concentrated at the centrallevel (Argyres & Silverman, 2004), and decisionsare not optimized for each subgroup (Siggelkow& Levinthal, 2003). Rather, subgroups coalescearound specific compromises and paths andthen prefer not to depart from them to minimizeorganizational and political struggles, facilitat-ing cumulation (Siggelkow & Levinthal, 2003).However, this stickiness around specific com-promises reduces exploration, and that, together
with the bureaucracy associated with central-ization, leads to lower inventiveness (Nickerson& Zenger, 2002; Yayavaram & Ahuja, 2008).
One solution to the trade-off outlined above isprovided by the creation of a small number oflinkages between the subgroups of a decentral-ized research structure, in effect converting adecomposable structure into a nearly decom-posable one (Sanchez & Mahoney, 1996; Simon,1962). Such linkages could consist of internalconsulting (Hargadon & Sutton, 1997), cross-group liaison teams (Fang, Lee, & Schilling,2010), or shared senior management roles acrosssubgroups (Tushman & O’Reilly, 2004). In anearly decomposable structure, structural sepa-ration between the subgroups fosters the infor-mational diversity and local application thatgive decentralization its power to develop in-
TABLE 3Organizational Determinants of Generative Appropriability
Components of GA Proposition
Cumulative component:Firms’ effectiveness in creatingnew inventions that build on theirown existing inventions
Proposition 1: The use of nearly decomposable organization structuresfor research leads to higher GA for firms.
Proposition 2: The use of personalization-intensive approaches toknowledge management leads to higher GA for firms.
Proposition 3: The use of incentive systems fostering stretch goals oninvention, across-group collaboration, and joint problem solvingleads to higher GA for firms.
Proposition 4a: The use of time-paced creative processes leads tohigher GA for firms.
Proposition 4b: The use of semistructured creative processes leads tohigher GA for firms.
Proposition 5: Moderate levels of organizational resource availabilityfor invention lead to higher GA for firms than very high or verylow levels.
Preclusive component:Firms’ effectiveness in preventingor precluding others from buildinginventions on the firms’ inventions
Proposition 6a: Moderate dispersion of firms’ research activitiesacross geographic locations leads to higher GA for firms than veryhigh or very low dispersion.
Proposition 6b: The use of moderately sized research teams leads tohigher GA for firms than the use of very large or very smallresearch teams.
Proposition 7: The use of retention-oriented HR practices leads tohigher GA for firms.
Proposition 8: Focusing inventive activity within a narrow domain orwithin closely related domains leads to higher GA for firms.
Proposition 9: Owning a broader global infrastructure of supportingassets leads to higher GA for firms.
Proposition 10a: Developing inventive knowledge through R&Dalliances (as opposed to through in-house efforts) leads to lower GAfor firms. This effect will strengthen as the number of R&D alliancesincreases.
Proposition 10b: Forming link alliances leads to higher GA for firmsthan forming scale alliances.
2013 257Ahuja, Lampert, and Novelli
ventions and exploit the opportunities in eachlocal ecology. However, the few yet criticalcross-subgroup linkages enable the transfer ofknowledge from one subgroup to another,thereby fostering broad diffusion, assimilation,and reuse of the organization’s inventive knowl-edge (Fang et al., 2010). These arguments sug-gest that nearly decomposable structures fostercumulative invention and, thus, enhance GA.
Proposition 1: The use of nearly de-composable organization structures forresearch leads to higher GA for firms.
GA and Knowledge Management Systems
The organization’s knowledge managementsystems provide another path to bridge time andspace and to transmit knowledge from the orga-nization’s past inventions to future potentialusers of that knowledge (Hollingshead, 2001;Lewis, Lange, & Gillis, 2005; Wegner, 1986).Knowledge management systems can facilitatecumulative invention by making knowledge ofthe firm’s prior inventions more widely and eas-ily accessible within the organization. Twobroad formal approaches to knowledge man-agement have been identified: codification andpersonalization (Hansen, Nohria, & Tierney,1999). Although these approaches have beenconceived in the context of organizationalknowledge in general, they are also meaningfulin the context of inventive knowledge and,hence, are useful for understanding GA.
We suggest that personalization-intensiveknowledge systems will be more useful for thekind of information exchange required by cumu-lative invention than codification-intensiveones. Successfully using the knowledge fromprior inventions to create new inventions oftenrequires information-rich contextual detail (Har-gadon & Sutton, 1997) or “metacontextual” infor-mation from the original invention (Majchrzak etal., 2004). Much information of this type tends tobe tacit, socially embedded, and hard to codify(Hansen et al., 1999; Kogut & Zander, 1992). Cod-ification and personalization approaches varyin their effectiveness in providing access to suchtacit knowledge. We draw on Hansen and col-leagues’ (1999) rich description of these ap-proaches to identify their key characteristics be-low and explore the implications of thesecharacteristics for cumulative invention.
In codification, knowledge is classified usinga people-to-documents approach. Information isextracted from the person who originally cre-ated it, converted into a knowledge object that isindependent of the individual who created it,and stored in an electronic database, makingthe information available to all subsequent us-ers in document form (Hansen et al., 1999). Con-versely, personalization follows a person-to-person logic (Hansen et al., 1999). Although thisapproach also entails embedding basic ele-ments of knowledge in electronic media, it em-phasizes mechanisms for linking people to peo-ple rather than people to documents. Thisfacilitates connections with the actual peoplewho worked on the original project, enabling thetransfer of noncodifiable, tacit knowledgethrough direct person-to-person contacts andenhancing cumulative invention.
In addition, each of these two approaches re-quires a set of complementary components(Hansen et al., 1999). In the codification ap-proach, firms invest in skilled abstraction, cod-ification staff, and sophisticated software formaking connections between previously devel-oped knowledge objects and currently facedproblems. In the personalization approach,firms instead invest in a collegial and coopera-tive culture, enabling capabilities that fosteractivation of the person-person network connec-tion (e.g., videoconferencing), and organiza-tional processes, such as collective brain-storming sessions that are used to convertindividual memories into organizational memo-ries (Hansen et al., 1999; Sutton & Hargadon,1996). Such culture facilitates exchange of priorknowledge and is therefore conducive to cumu-lative invention.
Although most organizations usually use bothcodification and personalization systems (Han-sen et al., 1999), since the complementary invest-ments required for the systems are different,most organizations tend to use one more inten-sively than the other. The prior arguments sug-gest that personalization-intensive knowledgesystems facilitate cumulative invention andthereby enhance GA.
Proposition 2: The use of personaliza-tion-intensive approaches to knowl-edge management leads to higher GAfor firms.
258 AprilAcademy of Management Review
GA and Incentive Systems for Inventors
Three common mechanisms have been iden-tified for controlling and incentivizing researchpersonnel: (1) output control, which directs re-sults and outcomes; (2) input control, which reg-ulates what resources are used in the process;and (3) behavior control, which modulates themeans leading up to the results (Cardinal, 2001).In the context of cumulative invention, all threematter. Specifically, precommitted stretch goals,collaboration across subgroups, and jointproblem-solving behaviors are likely to fosterhigher levels of GA.
Under output control, reward systems incen-tivize the achievement of certain outcomes.Some organizations precommit to certain highinvention output levels through publicly statedpolicies. For instance, 3M has a rule that 30 per-cent of sales must come from products launchedin the last five years (Eisenhardt & Brown, 1998).This public pronouncement creates a strong mo-tivation to invent; failure to successfully launchnew products would imply both lowered finan-cial rewards and a public loss of face. Further,the “stretch” nature of the goal implies that or-ganizational members are likely to feel somepressure in delivering this outcome. Given thelimited availability of time, the stretch goal islikely to encourage further evaluation of avail-able knowledge that can be reused, as opposedto exploration of new knowledge (Majchrzak etal., 2004).
However, reusing knowledge for invention re-quires integration of familiar knowledge with atleast some unfamiliar knowledge or perspective(Hargadon & Sutton, 1997; Majchrzak et al., 2004).This element of novelty can emerge from facili-tating connections across different subgroups orunits of the organization that have developedlocally differentiated knowledge as they servetheir own local contexts (Hansen et al., 1999).Input control systems that encourage collabora-tion between peers in different subgroups canprovide the necessary novelty in knowledge in-puts and cross-fertilization of the organization’sexisting knowledge to create new inventionsbased on past ones.
In addition, firms can foster behaviors thatfavor such internal cross-fertilization, which isconducive to cumulative invention. Incentiviz-ing joint problem-solving behaviors implies thatseeking help from others and providing help to
others become not just legitimate but norma-tively desirable (Hargadon & Sutton, 1997). Helpseeking and providing become perceived jointlyas markers of the organization’s solution-seeking orientation, in which solving the prob-lem is more important than being seen as aself-sufficient, extremely intelligent individual(Hargadon, 1998). These arguments suggest thatincentive systems fostering stretch goals,across-group collaboration, and joint problem-solving facilitate cumulative invention and,hence, enhance GA.
Proposition 3: The use of incentive sys-tems fostering stretch goals on inven-tion, across-group collaboration, andjoint problem solving leads to higherGA for firms.
GA and Creative Processes
Organizations enhance or diminish the abilityof their members to deliver on outcomes throughtheir design of organizational processes androutines (Brown & Eisenhardt, 1997; Eisenhardt &Brown, 1998; Sutton & Hargadon, 1996). We iden-tify two types of organizational creative process-es—time-paced and semistructured creativeprocesses (Brown & Eisenhardt, 1997)—as facili-tating higher GA.
Creative processes in organizations can besubjected to different types of temporal con-straints. For instance, Brown and Eisenhardt(1997) and Eisenhardt and Brown (1998) de-scribed two distinct temporal approaches tolaunching new products: event paced and timepaced. Event-paced launches are driven by ex-ternal occurrences, such as competitive actionsor market developments. In contrast, time-pacedcreative processes are driven by the calendar,with the organization following a rhythmic,time-paced approach (e.g., a new productlaunch every eighteen months).
Time-paced creative processes encourage cu-mulative invention and GA in multiple ways.First, they generate a constant time pressure foreach invention (Brown & Eisenhardt, 1997). Fre-quent invention under these pressures suggeststhat the salience-in-memory of the organiza-tion’s prior inventions will be higher: knowledgeused in the recent past is likely to be recalledmore easily. Further, the time pressure creates“a sense of urgency” and energy (Eisenhardt &
2013 259Ahuja, Lampert, and Novelli
Brown, 1998) as each individual becomes con-scious that his or her failure will hold up others.This enhances individual motivation to find asolution and, thus, leads to invention (Shere-mata, 2000). Second, the use of an internalrhythm decouples the organization from the lesscontrollable changes in the external environ-ment while enhancing its sensitivity to techni-cal changes within the organization (Brown &Eisenhardt, 1997). This reduced importance ofexternal stimuli further encourages inventors tobecome cognizant of invention possibilities pre-sented by the organization’s own portfolio ofpast inventions. Third, the rhythm also makespossible a form of entrainment (Brown & Eisen-hardt, 1997) as the various subgroups within theorganization synchronize their activities to thesame internal clock and even build on each oth-er’s inventions. For instance, Intel’s introductionof a series of time-paced chips throughout the1990s was probably enabled by the fact that ineach case the firm was building on its own pre-vious architecture and could rely on the manu-facturing process for the ever-denser chips to besimultaneously coordinated.
Proposition 4a: The use of time-pacedcreative processes leads to higher GAfor firms.
Creative processes in organizations also varyin their degree of formal structure. Highly struc-tured creative processes that involve tight timeaccountability and costing, resource-gated pro-cedures, and “disciplined problem solving” im-prove the efficiency of invention but may reduceits creativity (Brown & Eisenhardt, 1995; Clark &Fujimoto, 1991). In contrast, less structured cre-ative processes enhance motivation and creativ-ity (Amabile, 1987; Brown & Eisenhardt, 1997),which are beneficial to invention, but may leadto very scattershot invention outcomes ratherthan cumulative invention. Between these two isthe possibility of what we call “semistructuredcreative processes” (see Brown & Eisenhardt,1997, for the broader notion of semistructure),which we argue are likely to support high GA.
Semistructured creative processes foster cu-mulative invention by combining some structurewith some freedom. They provide direction, butwithout explicit fiat, through “unobstrusiveguidance” practices. For instance, policies pro-viding a certain amount of time for inventors topursue their own projects (e.g., 3M’s 15 percent
“personal” time) while restricting the bulk oftheir time to corporate projects, provide bothcorporate direction and individual autonomy.Such “personal” time fractions, while being sub-stantive and thus motivating to inventors whovalue “freedom,” may not be so high as to en-able inventors to go on complete flights of fancy.Given that 85 percent of the scientist’s time isoccupied with corporate projects, it is likely thatsuch “personal” time will be spent in the neigh-borhood of these existing corporate projects, forat least three reasons: (1) the human tendency oflocal search, (2) the constraint of scientists’ ex-pertise and attention being limited to specificareas, and (3) the social context that is likely todirect attention to commonly shared company-wide concerns. Similarly, providing a commonsubsidized cafeteria where all scientists eat(and thereby talk) or bringing together scientistsfrom different units into company-wide inven-tion fairs is a subtle way to encourage workalong certain trajectories of knowledge that uti-lize the organization’s earlier inventions. Theabove arguments suggest that using semistruc-tured creative processes fosters cumulative in-vention and, thus, enhances GA for firms.
Proposition 4b: The use of semistruc-tured creative processes leads tohigher GA for firms.
GA and Resources Available for Invention
The amount of resources available to pursuethe invention process is likely to influence thedirection of inventive effort and, eventually, thefirm’s degree of GA. We suggest that moderatelevels of resource availability should lead tohigher GA, relative to very low or very highlevels of resource availability.
Pursuing inventions is expensive and risky.For exploratory activity to be supported, suffi-cient resources need to be made available to theorganization (Lounamaa & March, 1987; Nohria& Gulati, 1996). As resources for invention areincreased from a very low level, inventive explo-ration is encouraged and inventiveness is en-hanced. However, when the amount of resourcesmade available for invention is very high, theincentives to examine and mine the firm’s exist-ing portfolio of inventions for new possibilitiesmay be limited—exploration is fun and chal-lenging, the money is available, and discipline
260 AprilAcademy of Management Review
is lacking (Nohria & Gulati, 1996). In contrast,moderate levels of resource availability permitsome exploration but also encourage reuse ofexisting inventive knowledge as firms mustmake the most of what is available and adaptexisting knowledge to address new problems(Majchrzak et al., 2004). These arguments sug-gest that cumulative invention and, therefore,GA will be curvilinearly related to the resourcesavailable for invention.
Proposition 5: Moderate levels of orga-nizational resource availability for in-vention lead to higher GA for firmsthan very high or very low levels.
THE PRECLUSIVE COMPONENT OF GA
GA and the Division of Inventive Labor
Hiring away scientists who embody a firm’sinventive knowledge is a potent weapon in en-abling a competitor to build on the firm’s inven-tions. This is especially true in the context oftacit and hard to codify knowledge (Kogut &Zander, 1992). Task design can be used to miti-gate such hazards (Liebeskind, 1996). The prin-ciples of division of labor and task design canbe applied at two different levels within theorganization to limit the degree to which infor-mation can leave the firm: first, to the divisionand distribution of tasks across geographic lo-cations, and, second, to the design of individ-ual jobs.
A firm can strategically limit the degree towhich information can leave the firm by imple-menting a specific geographic distribution ofresearch tasks across the organization (Zhao,2006). Inventions create value through the syn-thesis of inventive knowledge elements withother complementary inventive knowledge ele-ments or assets of the firm (Zhao, 2006). Individ-ual knowledge elements are less valuable in theabsence of their complementary components(Anand & Galetovic, 2004; Zhao, 2006). By effec-tively separating research projects into distinctand mutually complementary components, witheach key component being based in a differentgeographic location, firms can make the act ofrecombination difficult for competitors. Success-fully exploiting these knowledge elements todevelop new inventions would require otherfirms to be able to tap into knowledge frommultiple locations simultaneously, a task that is
significantly more difficult than tapping intoone location. Thus, firms can use the dual mech-anisms of specialization (and differentiation) atthe location level and integration at the organi-zational level to simultaneously enhance theirinventive output while reducing the danger ofknowledge expropriation (Demsetz, 1991;Grant, 1996).
Firms can also limit the degree to which in-formation can leave by implementing a distri-bution of research tasks at the individual levelthrough job design and employee conduct rules.Using larger project teams implies that individ-ual scientists have a proportionately smallerrole in the project. Such a division of labor en-sures that the knowledge available with eachindividual member for each project is smaller.As a result, to reconstruct a given level of knowl-edge about a project at the original firm, thecompetitor must hire away more people (Ahuja,2002). This is likely to be difficult. Further, withthis arrangement the marginal product of eachindividual team member to a given project de-clines. Hence, to extract the maximal value fromhis or her own inventive knowledge, each scien-tist in the focal firm needs more complementaryunits of knowledge (from other scientists), thusmaking his or her departure decision more dif-ficult. Finally, to the extent that some scientistsdo move, this task organization limits the dis-ruption of the focal project since the lost scien-tists represent only a small part of the project’sscientific expertise.
However, while the managerial choices ofgeographically dispersing research activitiesand enlarging the size of the resource team pro-vide a fundamental impetus toward preclusion,these choices can also have a simultaneous im-plication for cumulative invention. Splitting re-search tasks across geographies raises integra-tion and coordination challenges for the focalfirm. Indeed, if tasks are extremely fragmented,these coordination challenges could interferewith the firm’s own ability to invent, reducingGA. Similarly, as team size becomes very large,it is possible that the costs of integrating theknowledge to develop new inventions will in-crease and the coordination complexities willreduce inventiveness. Furthermore, with verylarge teams, the close ties that are necessary forthe transfer of rich information between mem-bers may become more difficult to build andmaintain, limiting the development of a social
2013 261Ahuja, Lampert, and Novelli
community. Such a social community is impor-tant since it can provide a superior work envi-ronment to foster the productivity of the teamand serve as a retention device making it moredifficult for individual inventors to leave (Paru-churi, Nerkar, & Hambrick, 2006).2 These nega-tive implications of the division of labor acrosslocations or teams could limit the firm’s owninventiveness and hurt its GA.
This suggests the existence of an inverted-Urelationship between firms’ division of inventivelabor and their GA:
Proposition 6a: Moderate dispersion offirms’ research activities across geo-graphic locations leads to higher GAfor firms than very high or very lowdispersion.
Proposition 6b: The use of moderatelysized research teams leads to higherGA for firms than the use of very largeor very small research teams.
GA and Retention-Oriented HR Practices
In addition to being hired away by competi-tors, key inventors might leave the firm volun-tarily to start their own companies. In this case anatural knowledge input for the inventive activ-ity of the new-born enterprise would be the in-ventor’s recent inventive knowledge, possiblyoriginated within the context of the employer’sinventive activity (Klepper & Sleeper, 2005). Thissuggests that a path to precluding other firmsfrom building on the inventive knowledge of thefocal firm is to reduce the turnover of researchemployees through retention-oriented HR prac-tices. Moreover, retention also contributes to en-hancing cumulative invention by ensuring thatemployees who developed earlier inventionsstay within the firm and contribute their knowl-edge to later inventions. Common retention-oriented practices include higher relative pay,internal promotion opportunities, provision ofemployment security, and restrictions on em-ployee activity, such as noncompete agree-ments.
Research on retention has emphasized thatlong-term incentives and inducements reduce a
firm’s turnover rate (Batt & Colvin, 2011; Doe-ringer & Piore, 1971). Such long-term incentivesand inducements can help to enhance a firm’sability to preclude others from building on itsinventions. For instance, higher relative pay in-creases the inventor’s opportunity cost of leav-ing the firm. Similarly, noncompete agreementsthat prevent employees from working in areasthat are close to the areas they worked in fortheir employer (Franco & Mitchell, 2008; Marx,Strumsky, & Fleming, 2009) also reduce the at-tractiveness of departure for employees, sincethey reduce employees’ value to other employ-ers and also significantly diminish employees’ability to work in their primary domains of ex-pertise (Marx et al., 2009).
Long-term incentives and perceptions of jobsecurity also facilitate cumulative invention. Forinstance, delayed-vesting option compensationthat provides employees with a right to buyshares of stock at set historical prices and holdthem for a certain number of years inhibits de-partures and, hence, loss of knowledge to otherfirms. It also fosters cumulative invention bymotivating inventors to continue inventing,since subsequent inventions represent both anexercise of their creative capabilities and ameans of enhancing the value of their options.Further, such devices and long-term job securityin general discourage inventors from exten-sively engaging in job searches (Direnzo &Greenhaus, 2011), facilitating the developmentof a longer-term research agenda within the or-ganization, again fostering cumulative inven-tion. The above arguments suggest that the useof retention-oriented HR practices fosters bothpreclusion and cumulative invention and, thus,enhances GA.
Proposition 7: The use of retention-oriented HR practices leads to higherGA for firms.
GA and Invention Domain
Whenever a firm commits resources to the de-velopment of a new invention, it faces a choicebetween building inventions in inventiveknowledge domains that it is already active inor domains that are conceptually similar to thedomains it is active in and building inventionsin new or unrelated knowledge domains. Ourargument draws on the literature on local
2 We are grateful to an anonymous reviewer for this valu-able suggestion.
262 AprilAcademy of Management Review
search and absorptive capacity to predict thatas a firm develops inventions in the same orrelated domains, its GA is enhanced as its abil-ity to build on its inventions increases and theincentives and capability of competitors to buildon its inventions decrease.
Understanding and knowledge are often cu-mulative (Cohen & Levinthal, 1990; Helfat, 1997;Nelson & Winter, 1982). The extent of a firm’sprior experience in an inventive knowledge do-main is, thus, a good predictor of the firm’ssubsequent invention success in that domain(Stuart & Podolny, 1996). Focusing inventive ac-tivities within a narrow domain or withinclosely related domains therefore fosters cumu-lative invention. Further, it is also likely to favorpreclusion of other firms. Entry into a knowledgedomain where the focal firm has been inventingextensively is likely to be unattractive to com-petitors, because as a firm progressively ex-pands its dominance in a specific domain, itsgreater experience in the domain increases itsrelative competence (Cohen & Levinthal, 1989,1990; Lane, Koka, & Pathak, 2006) vis-à-vis otherfirms. Recognizing this relative disadvantage,and given the finite carrying capacity of anydomain, when faced with a firm who dominatesa given knowledge domain, potential entrantswill choose to forgo entry into that domain andto seek other domains where their disadvantageis not as pronounced. The above arguments sug-gest that focusing inventive activities within anarrow domain or within closely related do-mains fosters both cumulative invention andpreclusion and, thus, enhances GA.
Proposition 8: Focusing inventive ac-tivity within a narrow domain orwithin closely related domains leadsto higher GA for firms.
GA and Supporting Assets
Firms can also influence the incentives ofcompetitors to build on their inventions throughownership of the supporting assets necessary tocommercialize the stream of inventions. Manu-facturing facilities, marketing and sales forces,and a global infrastructure to commercialize aninvention are all critical for profiting from inven-tions in general (Teece, 1986). We emphasizethat such assets also retain their value in thecontext of GA. An established global infrastruc-
ture can create a speed advantage for the firmthat owns the infrastructure by enabling thetranslation of a given invention into a marketedproduct or service across the globe in an effi-cient and effective fashion. This speed advan-tage is useful for converting a given invention toprofits and also acts as a deterrent to other com-petitors who want to enter that invention nichewith derivative inventions.
Consider two firms, A and B. Assume that Ahas an infrastructure of supporting assets intwenty countries, whereas B has such an infra-structure in only two countries. Now considerfirm C, which is considering building on an in-ventive opportunity opened up by the inventionsof either firm A or B. Other things being equal,firm C would prefer to build on the opportunitycreated by B’s inventions rather than on the op-portunity created by A’s inventions. The concernfor C would be that if it launched a new deriv-ative invention based on A’s inventions, A couldreplicate C’s efforts and launch a similar prod-uct relatively quickly in the twenty markets inwhich A is currently active. Yet, if C picks B’sideas to build on, it knows that B’s speed advan-tage exists in only two markets. Therefore, otherthings being equal, C would face less competi-tion if it picked B’s ideas over A’s to build on.Based on the above arguments, we suggest thata broader global infrastructure fosters preclu-sion of other firms and, thus, enhances GA.
Proposition 9: Owning a broaderglobal infrastructure of supporting as-sets leads to higher GA for firms.
GA and R&D Alliances
Firms often generate inventive knowledgethrough R&D alliances with other firms. We ar-gue here that this choice of mode for sourcinginventive knowledge has implications for firms’GA. Specifically, developing inventive knowl-edge internally enhances GA more than devel-oping it through joint research projects withother firms. First, in-house development limitsoutsiders’ access to the firm’s knowledge. In con-trast, joint knowledge development efforts withother firms imply that the focal firm risks expos-ing its knowledge core (Thompson, 1967) to itspartners, who may then use that knowledge tobuild on its inventions (Garcia-Canal, Valdes-Llaneza, & Sanchez-Lorda, 2008; Hamel, Doz, &
2013 263Ahuja, Lampert, and Novelli
Prahalad, 1989). Second, in-house developmentallows firms to make idiosyncratic inventivechoices relating to the design and creation ofthe product or process such that their choicesare maximally and uniquely complementaryonly with their own bundle of activities andprior inventions (Anand & Galetovic, 2004). This,in turn, limits the opportunity for other firms touse this knowledge in an effective fashion.
Although, in principle, having many alliancescould also provide an opportunity for the focalfirm to benefit from more knowledge spilloveropportunities, it is likely that this effect may notplay a significant role in increasing its GA. Withmany alliances being enacted simultaneously,given finite absorptive capacity, the firm’s abil-ity to assimilate and absorb knowledge flowsmay be diminished (Burt, 1992). Further, control-ling knowledge flows in alliance settings re-quires tight compartmentalization of individualalliances and tight definition of alliance scope(Khanna, Gulati, & Nohria, 1998; Oxley & Wada,2010). With many alliances, such tight controlmay be difficult. Indeed, in such a contextknowledge may leak from many different pointsof a firm in a sprinkler head fashion (Owen-Smith & Powell, 2004), facilitating the inventiveefforts of others. The above arguments suggestthat developing inventive knowledge throughR&D alliances reduces the preclusion of otherfirms from building on the focal firm’s inven-tions and, thus, reduces GA, an outcome furtherworsened by having many alliances.
Proposition 10a: Developing inventiveknowledge through R&D alliances (asopposed to through in-house efforts)leads to lower GA for firms. This effectwill strengthen as the number of R&Dalliances increases.
When firms do use alliances to develop inven-tive knowledge, different types of alliances havediffering effects on GA; specifically, scale alli-ances entail higher risks to GA than link alli-ances. A scale alliance involves two partnersthat both bring the same capability into the al-liance, with the core objective of facilitatingeconomies of scale (Dussauge, Garrette, &Mitchell, 2004). Thus, Intel and AMD’s forming arelationship to develop a new microprocessorwould be an illustration of a scale R&D alliance,since both firms bring chip development skillsto the alliance. Conversely, in a link alliance the
partners bring different skills to the alliance(Dussauge et al., 2004), as in the case of researchcollaboration between a chip design companyand a chip manufacturing foundry.
Two sets of arguments suggest that the leak-age of inventive knowledge to outsiders is likelyto be worse for scale alliances than for linkalliances. First, given the similarity in theircompetence backgrounds, partners in scale R&Dalliances are likely to have high levels of ab-sorptive capacity for the knowledge of theirpartners (Cohen & Levinthal, 1989; Lane & Lubat-kin, 1998; Park & Russo, 1996). Commonalities ofinterests and competencies between firms inscale alliances (Ahuja, 2000; Dussauge et al.,2004; Koka & Prescott, 2002; Lane & Lubatkin,1998) facilitate information diffusion, and as thenumber of such ties increases, policing them fordiffusion of information may become progres-sively more difficult. In R&D link alliances, in-stead, partners bring different kinds of knowl-edge to the alliance and may not have as strongan absorptive capacity for their partners’ knowl-edge as they did with scale alliances.
Second, unlike scale alliances, which imply ahorizontal relationship between partners, linkalliances reflect vertical ties. The overlap be-tween the interests of partners in link alliancesis lower than that in horizontal ties. Many of theaspects of a firm’s inventive knowledge coreand the trajectories of inventions that are possi-ble from it may be of limited interest to thepartners. For example, a chip foundry firm in analliance with a chip design firm may have lim-ited interest in understanding how higher-orderfeatures can be designed into chips—its interestwould be more in creating new ways of manu-facturing the chips than in new product featuresper se. These arguments suggest that forminglink alliances fosters the preclusion of otherfirms from building on the focal firm’s inven-tions, relative to scale alliances, and, thus, en-hances GA.
Proposition 10b: Forming link alli-ances leads to higher GA for firmsthan forming scale alliances.
DISCUSSION AND CONCLUSIONS
In this article we drew a distinction betweentwo notions of appropriability and identifiedsome of the determinants of the less studied one.
264 AprilAcademy of Management Review
We developed the concept of GA as a potentiallyrelevant outcome in assessing the performanceof inventive firms. We identified two paths toinfluencing this outcome— cumulative inven-tion and preclusion of others—and developedpropositions to explain how firms could use ba-sic levers of organizational structure, systems,and strategy to enhance the firm’s GA. In thissection we suggest directions for further re-search based on the identification of caveats tobe considered in the context of GA, and we dis-cuss the managerial and theoretical implica-tions of GA.
Implications for Future Research on GA
We have indicated several reasons why firmsmight want to enhance GA, and we also haveidentified strategies firms may use to do so.However, an issue worth considering is whetherit makes sense for firms to always try to enhanceGA. It is indeed likely that, as with other perfor-mance outcomes for firms, enhancing GA maybe more or less desirable depending on under-lying conditions. We identify several sets of cir-cumstances under which firms would have toevaluate trade-offs before deciding whether ornot to enhance GA. For instance, the value ofenhancing GA may depend on expectationsabout the nature of the spawned inventions.Spawned inventions could be substitutes (theyaddress the same buyer need and render theoriginal invention less competitive), comple-ments to the existing invention (they boost thesales of the original invention), or inventionsunrelated to the existing invention in product-market terms. Prima facie, it would appear thatfirms may have different attitudes toward theimportance of increasing GA depending onwhether the expected subsequent inventionsare substitutes versus complements.
Other trade-offs may emerge because of con-flicts between GA and other desirable outcomesfor a firm. We noted earlier that the strategies toenhance appropriability may be different acrossPA and GA, thus leading to a conflict betweenthe two. Similarly, different environments maycause the relative importance of PA and GA tovary. For instance, in an environment of ferment,where new competitive products are continu-ously being launched (Abernathy & Utterback,1978), being able to develop newer, enhancedproducts may be more important than trying to
figure out how to marginally improve the mon-etization of a given invention. However, in otherenvironments the relative payoff between PAand GA may be reversed. Developing a “contin-gent hierarchy of appropriabilities” and identi-fying the conditions under which a particularform of appropriability is favored over the otherwould indeed be an exciting and useful re-search direction.3
Trade-offs are also associated with the twocomponents of GA: cumulative invention andpreclusion of others. For instance, cumulativeinvention is likely to foster relatively speedyinvention since building on existing inventiveknowledge enables the firm to respond fasterthan entering a knowledge space de novo. Thiscould be advantageous in the short run and indynamic environments where speed is impor-tant. However, such a strategy may also be asource of rigidity in the long run. A firm focusedon building on its existing inventive knowledgemay not adequately experiment with new inven-tive trajectories and may end up not alignedwith the needs of consumers. Similarly, preclud-ing others from building on the firm’s inventionsprovides the focal firm dominance in a knowl-edge domain. However, since great solutionscan emerge from the cross-fertilization of differ-ent and unrelated inventive knowledge bases,knowledge spillovers from one firm to anothermay open new paths to invention and possiblyeven new inventive fields. These potential ben-efits are at risk when the firm pursues preclu-sion very strongly.
Evaluating GA performance gives firms achance to draw managerial attention to a set ofcounterfactuals that are not commonly focusedon in performance reviews. Much performancemeasurement in firms focuses on evaluations ofmanagers and organizations through their real-ized outcomes on certain milestones, such asprofits. However, a case could be made that op-timizing performance may entail examining notjust how well a firm has performed but also howmuch better it could have performed. GA pro-vides an outcome that draws attention to thisaspect of performance. An organization trackingits GA may recognize not only how successful ithas been in commercializing a product but also
3 We are grateful to an anonymous reviewer for suggest-ing the development of a hierarchy of appropriabilities.
2013 265Ahuja, Lampert, and Novelli
how much more could have been possible byrecognizing and exploiting follow-on inventions.This discussion suggests that identifying poten-tial measures of GA would help to push the GAresearch agenda as well as its usage in prac-tice. In industries where patents are meaning-ful, patent data can be used to measure GA.Citations in a patent are markers of its intellec-tual lineage, since they indicate that a particu-lar invention was built upon a prior invention. Asimple archival measure of GA is provided bythe ratio of self-citations to a firm’s patents tothe total citations received by the firm’s patents.In other contexts one could imagine using prod-uct characteristics and features data to tracelinkages between products within andacross firms.
Implications for Theory
Priem and Butler (2001) have noted that spe-cifically identifying “how” resources create sus-tained value and incorporating a temporal com-ponent that relates a firm’s history to its currentoutcomes are two key issues that need to beaddressed for further development of the RBV.This article addresses these issues. The con-struct of GA draws attention to the possibilitythat sustained value creation occurs not justthrough the “defense” plays of protection fromimitation and substitution, already well dis-cussed in the literature, but also through a thirdpath, the “offense” play of creating new inven-tions that lead into new product markets. Fur-ther, by identifying specific strategies throughwhich firms may leverage the power of theirpast inventions into future inventions, this arti-cle also illuminates a temporal link between anorganization’s past successes and its currentoutcomes and shows how resources, instead ofsimply being eroded over time (Priem & Butler,2001), may also lead to new rents. For instance,Apple’s performance over the last decade has insome part been based on protection from imita-tion and substitution in existing product mar-kets. Yet perhaps equally important is its perfor-mance in creating new inventions building onits own prior inventions, which has helped Ap-ple to target completely new product markets.
This article also identifies implications for theliterature on organizational learning. Much ofthe knowledge literature has focused on theidentification of factors that enhance interorga-
nizational learning and facilitate knowledgetransfer across organizations (Grandori & Kogut,2002). GA draws attention to two less studiedaspects of organizational learning. First, GAemphasizes how a firm can enhance the utiliza-tion of its own existing knowledge over time forthe purposes of invention. A distinguishingcharacteristic of knowledge is that, unlike phys-ical inputs, it is not consumed in its usage in theproduction process. Rather, usage may makeprior knowledge even more useful by indicatingdifferent ways that it can be recombined (cf. ourdiscussion of the conceptual component of aninvention). Yet converting this potential for re-use—which is one of knowledge’s most promis-ing and productive features—from being latentto becoming realized requires a better under-standing of how organizations might systemat-ically improve their utilization of their existingknowledge. Although knowledge recombinationis recognized as one of the core functions of afirm (Grant, 1996), there has been limited re-search on the strategies that enable firms torecombine their knowledge into new syntheses,a notable underemphasis in the context of anincreasingly knowledge-driven economy.
In this article we focus directly on the knowl-edge reuse problem and, in the process, high-light an important puzzle: knowledge is valu-able because it can be reused, but it is difficultto store or retrieve in a manner that makes itavailable at the locus where its reuse would bethe most beneficial. The propositions we havepresented identify how elements of organiza-tional structure, systems, and strategy can beused to address this puzzle in the context ofknowledge reuse for invention. Identifying otherstrategies that can be used to address this puz-zle and targeting knowledge reuse contexts be-yond invention are natural directions for furtherresearch.
Second, the learning literature has commonlyfocused on how organizations can increase theirlearning from other organizations. In contrast,this article draws attention to the strategies thatfirms may use to prevent others from learningfrom them. Although prima facie that appearssocially undesirable, a nuanced considerationdraws our attention to the incentive problemunderlying GA—in a for-profit world, activitiesthat do not yield returns to the actor are eventu-ally not undertaken. In that sense, the strategiesreflected here, while appearing undesirable,
266 AprilAcademy of Management Review
may still be better than the alternatives in pro-moting research effort. Theoretically, this articletherefore represents a call to integrate more di-rectly the consideration of incentives and com-petition into the literature on learning (Grandori& Kogut, 2002). For instance, research modelingknowledge spillovers would need to explicitlyrecognize that firms could strategically be at-tempting to minimize such spillovers. This mayimply a need for a more nuanced theoretical andmethodological approach in studying how orga-nizations learn from each other.
Understanding how the two components ofGA (and the managerial choices underlyingthem) might interrelate with each other wouldbe helpful—another task for future research.
REFERENCES
Abernathy, W. J., & Utterback, J. M. 1978. Patterns of indus-trial innovation. Technology Review, 80(7): 40–47.
Ahuja, G. 2000. Collaboration networks, structural holes, andinnovation: A longitudinal study. Administrative Sci-ence Quarterly, 45: 425–455.
Ahuja, G. 2002. When Atlas shrugged: Pre-emption, complex-ity and division of labor in a theory of appropriability.Paper presented at the annual meeting of the Academyof Management, Seattle.
Alchian, A. A. 1950. Uncertainty, evolution, and economictheory. Journal of Political Economy, 58: 211–221.
Amabile, T. M. 1987. The motivation to be creative. In S.Isaksen (Ed.), Frontiers of creativity research: Beyond thebasics: 223–254. Buffalo, NY: Bearly Limited.
Anand, B., & Galetovic, A. 2004. How market smarts canprotect property rights. Harvard Business Review, 82(12):72–81.
Argyres, N. S., & Silverman, B. S. 2004. R&D, organizationstructure, and the development of corporate technolog-ical knowledge. Strategic Management Journal, 25(Spe-cial Issue): 929–958.
Arrow, K. J. 1962. Economic welfare and the allocation ofresources for invention. In The rate and direction ofinventive activity: Economic and social factors: 609–626.Cambridge, MA: National Bureau of Economic Research.
Batt, R., & Colvin, A. J. S. 2011. An employment systemsapproach to turnover: Human resources practices, quits,dismissals, and performance. Academy of ManagementJournal, 54: 695–717.
Benner, M. J., & Tushman, M. L. 2003. Exploitation, explora-tion, and process management: The productivity di-lemma revisited. Academy of Management Review, 28:238–256.
Brown, J., & Eisenhardt, K. 1995. Product development: Pastresearch, present findings, and future directions. Acad-emy of Management Review, 20: 342–378.
Brown, J., & Eisenhardt, K. 1997. The art of continuouschange: Linking complexity theory and time-paced evo-lution in relentlessly shifting organizations. Administra-tive Science Quarterly, 42: 1–34.
Burrows, P. 2009. Lessons from the Apple-Psystar battle.Bloomberg Business Week, December 2: http://www.businessweek.com/technology/content/dec2009/tc2009121_009697.htm.
Burt, R. S. 1992. Structural holes. Cambridge, MA: HarvardUniversity Press.
Cardinal, L. B. 2001. Technological innovation in the phar-maceutical industry: The use of organizational control inmanaging research and development. Organization Sci-ence, 12: 19–36.
Chesbrough, H. 2002. Graceful exits and missed opportuni-ties: Xerox’s management of its technology spin-off or-ganizations. Business History Review, 76: 803–837.
Christensen, C. M. 1997. The innovator’s dilemma. Boston:Harvard Business School Press.
Christensen, C. M., & Bower, J. L. 1996. Customer power,strategic investment and the failure of leading firms.Strategic Management Journal, 17: 197–218.
Clark, K. B., & Fujimoto, T. 1991. Product development perfor-mance. Boston: Harvard Business School Press.
Cohen, W. M., & Levinthal, D. A. 1989. Innovation and learn-ing: The two faces of R&D. Economic Journal, 99: 569–596.
Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: Anew perspective on learning and innovation. Adminis-trative Science Quarterly, 35: 128–152.
Cohen, W. M., Nelson, R. R., & Walsh, J. P. 2000. Protectingtheir intellectual assets: Appropriability conditions andwhy U.S. manufacturing firms patent (or not). NBERWorking Paper No. 7552, National Bureau of EconomicResearch, Cambridge, MA.
Demsetz, H. 1991. The theory of the firm revisited. In O. E.Williamson & S. Winter (Eds.), The nature of the firm:159–178. New York: Oxford University Press.
Dierickx, I., & Cool, K. 1989. Asset stock accumulation andsustainability of competitive advantage. ManagementScience, 35: 1504–1511.
Direnzo, M. S., & Greenhaus, J. H. 2011. Job search and vol-untary turnover in a boundaryless world: A control the-ory perspective. Academy of Management Review, 36:567–589.
Doeringer, P., & Piore, M. 1971. Internal labor markets andmanpower analysis. Lexington, MA: Heath.
Dussauge, P., Garrette, B., & Mitchell, W. 2004. Asymmetricperformance: The market share impact of scale and linkalliances in the global auto industry. Strategic Manage-ment Journal, 25: 701–711.
Eisenhardt, K. M., & Brown, S. L. 1998. Time pacing: Compet-ing in markets that won’t stand still. Harvard BusinessReview, 76(2): 59–69.
Fang, C., Lee, J., & Schilling, M. A. 2010. Balancing explora-tion and exploitation through structural design: The iso-
2013 267Ahuja, Lampert, and Novelli
lation of subgroups and organizational learning. Orga-nization Science, 21: 625–642.
Fleming, L. 2001. Recombinant uncertainty in technologicalsearch. Management Science, 47: 117–132.
Franco, A. M., & Mitchell, M. F. 2008. Covenants not to com-pete, labor mobility, and industry dynamics. Journal ofEconomics & Management Strategy, 17: 581–606.
Garcia-Canal, E., Valdes-Llaneza, A., & Sanchez-Lorda, P.2008. Technological flows and the choice of joint ven-tures in technology alliances. Research Policy, 37: 97–114.
Geer, C. T. 1990. Epi-logue? Forbes, August 6: 114.
Gillette, F. 2011. The rise and inglorious fall of Myspace.Bloomberg Business Week, June 22: 1.
Goldenberg, J., Mazursky, D., & Solomon, S. 1999. Towardidentifying the inventive templates of new products: Achanneled ideation approach. Journal of Marketing Re-search, 36: 200–210.
Grandori, A., & Kogut, B. 2002. Dialogue on organization andknowledge. Organization Science, 13: 224–231.
Grant, R. M. 1996. Toward a knowledge-based theory of thefirm. Strategic Management Journal, 17(Winter SpecialIssue): 109–122.
Grant, R. M. 2007. Contemporary strategy analysis (6th ed.).Oxford: Blackwell.
Green, J. R., & Scotchmer, S. 1995. On the division of profit insequential innovation. RAND Journal of Economics, 26:20–33.
Hamel, G., Doz, Y. L., & Prahalad, C. K. 1989. Collaborate withyour competitors—and win. Harvard Business Review,67(1): 133–139.
Hansen, M. T., Nohria, N., & Tierney, T. 1999. What’s yourstrategy for managing knowledge? Harvard BusinessReview, 77(2): 106–116.
Hargadon, A. B. 1998. Firms as knowledge brokers: Lessonsin pursuing continuous innovation. California Manage-ment Review, 40(3): 209–227.
Hargadon, A. B., & Sutton, R. I. 1997. Technology brokeringand innovation in a product development firm. Admin-istrative Science Quarterly, 42: 716–749.
Helfat, C. E. 1997. Know-how and asset complementarity anddynamic capability accumulation: The case of R&D.Strategic Management Journal, 18: 339–360.
Hollingshead, A. B. 2001. Cognitive interdependence andconvergent expectations in transactive memory. Journalof Personality and Social Psychology, 81: 1080–1089.
Hopenhayn, H., Llobet, G., & Mitchell, M. F. 2006. Rewardingsequential innovators: Prizes, patents, and buyouts.Journal of Political Economy, 114: 1041–1068.
Hopenhayn, H., & Mitchell, M. F. 1999. Innovation fertility andpatent design. NBER Working Paper No. 7070, NationalBureau of Economic Research, Cambridge, MA.
Katila, R., & Ahuja, G. 2002. Something old, something new:A longitudinal study of search behavior and new prod-uct introduction. Academy of Management Journal, 45:1183–1194.
Khanna, T., Gulati, R., & Nohria, N. 1998. The dynamics oflearning alliances: Competition, cooperation, and rela-tive scope. Strategic Management Journal, 19: 193–210.
King, A. W. 2007. Disentangling interfirm and intrafirmcausal ambiguity: A conceptual model of causal ambi-guity and sustainable competitive advantage. Academyof Management Review, 32: 156–178.
King, A. W., & Zeithaml, C. P. 2001. Competencies and firmperformance: Examining the causal ambiguity paradox.Strategic Management Journal, 22: 75–99.
Klepper, S., & Sleeper, S. 2005. Entry by spinoffs. Manage-ment Science, 51: 1291–1306.
Kogut, B., & Zander, U. 1992. Knowledge of the firm, combi-native capabilities, and the replication of technology.Organization Science, 3: 383–397.
Koka, B. R., & Prescott, J. B. 2002. Strategic alliances as socialcapital: A multidimensional view. Strategic Manage-ment Journal, 23: 795–816.
Lane, P. J., Koka, B. R., & Pathak, S. 2006. The reification ofabsorptive capacity: A critical review and rejuvenationof the construct. Academy of Management Review, 31:833–863.
Lane, P. J., & Lubatkin, M. H. 1998. Relative absorptive capac-ity and interorganizational learning. Strategic Manage-ment Journal, 19: 461–477.
Levin, R. C., Klevorick, A., Nelson, R., & Winter, S. 1987.Appropriating the returns from industrial research anddevelopment. Brookings Papers on Economic Activity, 3:783–831.
Levinthal, D. A., & March, J. G. 1981. A model of adaptiveorganizational search. Journal of Economic Behaviorand Organization, 2: 307–333.
Lewis, K., Lange, D., & Gillis, L. 2005. Transactive memorysystems, learning, and learning transfer. OrganizationScience, 16: 581–598.
Liebeskind, J. P. 1996. Knowledge, strategy and the theory ofthe firm. Strategic Management Journal, 17(Winter Spe-cial Issue): 93–1072.
Lippman, S. A., & Rumelt, R. P. 1982. Uncertain imitability: Ananalysis of interfirm differences in efficiency undercompetition. Bell Journal of Economics, 13: 418–438.
Lounamaa, P., & March, J. 1987. Adaptive coordination of alearning team. Management Science, 33: 107–123.
Majchrzak, A., Cooper, L. P., & Neece, O. P. 2004. Knowledgereuse for innovation. Management Science, 50: 174–188.
March, J. G. 1991. Exploration and exploitation in organiza-tional learning. Organization Science, 2: 71–87.
Marx, M., Strumsky, D., & Fleming, L. 2009. Mobility, skills,and the Michigan non-compete experiment. Manage-ment Science, 55: 875–889.
McEvily, S. K., & Chakravarthy, B. 2002. The persistence ofknowledge-based advantage: An empirical test forproduct performance and technological knowledge.Strategic Management Journal, 23: 285–305.
268 AprilAcademy of Management Review
McEvily, S. K., Das, S., & McCabe, K. 2000. Avoiding compe-tence substitution through knowledge sharing. Acad-emy of Management Review, 25: 294–312.
Murray, F., & O’Mahony, S. 2007. Exploring the foundations ofcumulative innovation: Implications for organizationscience. Organization Science, 18: 1006–1021.
Nelson, R., & Winter, S. 1982. An evolutionary theory of eco-nomic change. Cambridge, MA: Harvard UniversityPress.
Nickerson, J. A., & Zenger, T. R. 2002. Being efficiently fickle:A dynamic theory of organizational choice. Organiza-tion Science, 13: 547–566.
Nohria, N., & Gulati, R. 1996. Is slack good or bad for inno-vation? Academy of Management Journal, 39: 1245–1264.
Owen-Smith, J., & Powell, W. W. 2004. Knowledge networksas channels and conduits: The effects of spillovers in theBoston biotechnology community. Organization Science,15: 5–21.
Oxley, J., & Wada, T. 2010. Alliance structure and the scope ofknowledge transfer: Evidence from U.S.-Japan agree-ments. Management Science, 55: 635–649.
Park, S., & Russo, M. 1996. When competition eclipses coop-eration: An event history analysis of joint venture fail-ure. Management Science, 42: 875–890.
Paruchuri, S., Nerkar, A., & Hambrick, D. C. 2006. Acquisitionintegration and productivity losses in the technical core:Disruption of inventors in acquired companies. Organi-zation Science, 17: 545–562.
Priem, R. L., & Butler, J. E. 2001. Is the resource-based “view”a useful perspective for strategic management re-search? Academy of Management Review, 26: 22–40.
Reed, R., & DeFillippi, R. J. 1990. Causal ambiguity, barriersto imitation, and sustainable competitive advantage.Academy of Management Review, 15: 88–102.
Rivkin, J. 2001. Reproducing knowledge: Replication withoutimitation at moderate complexity. Organization Sci-ence, 12: 274–293.
Roberts, P. W. 1999. Product innovation, product-market com-petition and persistent profitability in the U.S. pharma-ceutical industry. Strategic Management Journal, 20:655–670.
Sanchez, R., & Mahoney, J. T. 1996. Modularity, flexibility,and knowledge management in product and organiza-tion design. Strategic Management Journal, 17(WinterSpecial Issue): 63–76.
Scotchmer, S., & Green, J. 1990. Novelty and disclosure inpatent law. RAND Journal of Economics, 21: 131–146.
Sheremata, W. A. 2000. Centrifugal and centripetal forces inradical new product development under time pressure.Academy of Management Review, 25: 389–408.
Siggelkow, N., & Levinthal, D. A. 2003. Temporarily divide toconquer: Centralized, decentralized, and reintegratedorganizational approaches to exploration and adapta-tion. Organization Science, 14: 650–669.
Simon, H. 1962. A note on mathematical models for learning.Psychometrika, 27: 417–418.
Stuart, T. E., & Podolny, J. M. 1996. Local search and theevolution of technological capabilities. Strategic Man-agement Journal, 17(Summer Special Issue): 21–38.
Sutton, R. I., & Hargadon, A. 1996. Brainstorming groups incontext: Effectiveness in a product design firm. Admin-istrative Science Quarterly, 41: 685–718.
Teece, D. J. 1986. Profiting from technological innovation:Implications for integration, collaboration, licensingand public policy. Research Policy, 15: 285–305.
Thompson, J. D. 1967. Organizations in action: Social sciencebases of administrative theory. New York: McGraw-Hill.
Tushman, M. L., & O’Reilly, C. A., III. 2004. The ambidextrousorganization. Harvard Business Review, 82(4): 74–81.
Wegner, D. M. 1986. Transactive memory: A contemporaryanalysis of the group mind. In B. Mullen & G. R. Goeth-als (Eds.), Theories of group behavior: 185–208. New York:Springer-Verlag.
Winter, S. 1971. Satisficing, selection, and the innovativeremnant. Quarterly Journal of Economics, 85: 237–261.
Yayavaram, S., & Ahuja, G. 2008. Decomposability in knowl-edge structures and its impact on the usefulness ofinventions and knowledge-base malleability. Adminis-trative Science Quarterly, 53: 333–362.
Zhao, M. Y. 2006. Conducting R&D in countries with weakintellectual property rights protection. Management Sci-ence, 52: 1185–1199.
Gautam Ahuja ([email protected]) is the Harvey C. Fruehauf Professor of BusinessAdminstration at the University of Michigan. He received his Ph.D. from the Universityof Michigan. His research interests include the strategic management of knowledge,technology, and innovation.
Curba Morris Lampert ([email protected]) is an assistant professor in the Department ofManagement and International Business in the College of Business at Florida Inter-national University. She received her Ph.D. from the University of Texas at Austin. Herresearch interests include technology strategy and innovation, diversification anddivestments, and managing entrepreneurship in large corporations.
Elena Novelli ([email protected]) is a lecturer in management at Cass Busi-ness School, City University London. She received her Ph.D. from Bocconi University.Her research interests include the strategic management of knowledge, technology,and innovation.
2013 269Ahuja, Lampert, and Novelli