Escolar Documentos
Profissional Documentos
Cultura Documentos
JAN
FAGERBERG
T H E O X F O R D H A N D B O O K OF
INNOVATION
Edited by
JAN F A G E R B E R G
DAVID C. MOWERY
AND
R I C H A R D R. NELSON
OXFORD
U N IV E R S IT Y PR ESS
P r e f a c e
In a famous poem, “ The Blind Men and the Elephant,” John Godfrey Saxe (1816-87)
described what may happen when different observers approach the same phenom
enon from rather di fferent starting points. In the poem Saxe lets one of the blind men
approach the elephant’s side. The man finds it to be “ very like a wall.” Another fits
around its leg and concludes that it resembles a tree. And so on. They end up
disputing “ loud and long.” Saxe drew the following moral:
So oft in theologic wars,
The disputants, I ween,
Rail on in utter ignorance
Of what each others mean,
And prate about an Elephant
Not one of them has seen!
The point is, of course, that each “ disputant” has a valid insight, but needs to
combine it with the insights o f others to reach a holistic understanding. If we
substitute “ innovation” for the elephant and the “ social scientists from different
disciplines” for the blind men, we come close to understanding the motives that led
to the creation o f this handbook. Innovation is a multifaceted phenomenon that
cannot be easily squeezed into a particular branch of the social sciences or the
humanities. Consequently, the rapidly increasing literature on innovation is char
acterized by a multitude of perspectives based on— or cutting across— existing
disciplines and specializations. There is a danger, however, that scholars studying
innovation do it from starting points so different that they become unable to— or
not interested in— communicating with each other, preventing the development o f a
more complete understanding of the phenomenon.
The purpose o f this volume is to contribute to a holistic understanding of
innovation. The volume includes twenty-one carefully selected and designed con
tributions, each focusing on a specific aspect o f innovation, as well as an introduc
tory essay that sets the stage for the chapters that follow. The authors are leading
academic experts on their specific topics, and include economists, geographers,
historians, psychologists, and sociologists. Some contributors have engineering
degrees in addition to their social science degree. Each chapter can be read separately,
but most readers will benefit from reading the introductory essay first. Readers
interested in pursuing further study on specific topics will find suggestions for
¥1 PREFACE
additional reading (m arked with asterisks) in the reference list at the end o f each
chapter.
As with all books there is a history behind it. In fact there are several. There is a
long history, related to h ow innovation studies have evolved over the years. M any o f
the contributions presented here, C hapter 1 in particular, give elem ents o f that story.
The shorter history begins in the m id-1990s w ith the big im petus to innovation
research in Europe provided b y the “ Fram ew ork” program m es o f the European
C om m ission. H aving participated actively in this research for som e tim e, several o f
the contributors to this volum e becam e interested in establishing a netw ork that
could support
X JT
discussion and evaluation o f its results. F o r this xpu rp
x
ose Jan Fager-
K-r
berg organized in 1999, with the support o f the N orw egian Research C ouncil, an
international netw ork for innovation studies that m et occasionally to discuss
selected topics within innovation research. The m eetings o f this group led to a
proposal for a book reflecting o u r current know ledge on innovation. O xford
University Press was contacted and w elcom ed the idea. E con om ic su p port from
the European C om m ission and the N orw egian Research C ou n cil m ade it possible
for the contributors to meet twice to exchange ideas and com m ent on each other
contributions, greatly enhancing the quality and consistency o f the volum e.
One o f the central participants in the netw ork that led to this volu m e was Keith
Pavitt, Professor at SPR U (U niversity o f Sussex) and editor o f Research Policy, the
leading journal in the field. W ith a background in both engineering and econom ics,
Keith was one o f the pioneers in cross-disciplin ary research on innovation. C h arac
terized by a “ fact-finding” approach and a lack o f respect for received “ grand
theories” not supported by solid evidence, he influenced generations o f younger
researchers and helped put innovation studies on its current “ issue-driven,” em pir
ically oriented track. Keith enthusiastically supported this b oo k initiative, very
quickly (before anybody else) circulated a full draft o f a chapter and participated
actively in the discussions during the first w orkshop in Lisbon in N o vem ber 2002. He
died unexpectedly shortly afterwards. The editors and contributors dedicate this
book to his m em ory.
J.F., D .M ., R.N .
Oslo, Berkeley; and New York
January 2004
A c k n o w l e d g e m e n t s
W ithout financial su pport from the N orw egian Research C ouncil (projects
131468/510 and 139867/510) and the European C om m ission (the T E A R I project—
H PSE-C T-2002-60052) this b ook w ould not have been realized. We thank Trygve
Lande and Helge Rynning from N orw egian Research C ouncil and N icholaus Kas~
trinos from the European C om m ission for their cooperation. The Centre for
Technology, Innovation, and Culture (T IK ), U niversity o f Oslo gave Jan Fagerberg
a leave o f absence to start w orking with this project, which he spent at ISEG ,
Technical U niversity o f Lisbon. H e w ould also like to thank ISE G and the G ulben-
kian Foundation for helping to m ake this possible. Sim ilarly, D avid M ow ery would
like to thank the D ivision o f Research o f the H arvard Business School for a Bow er
Fellowship during 20 0 3-4 that aided his w ork on this volum e. M anuel G odinh o o f
ISEG helped organize the first w orkshop in Lisbon in N ovem ber 2002, and Bart
Verspagen o f E C IS (U niversity o f Eindhoven) sim ilarly assisted in organizing the
second w orkshop in R oerm ond in June 2003. In addition to the contributors several
people participated in these w orkshops and contributed to the progress o f the work,
we would particularly like to m ention Fulvio Castellacci, Joao Caraca, M aureen
McKelvey, Sandro M endon^a, R ichard Stankiewicz, and M ona W ibe. D uring the
final phases o f preparing the m anuscript for publication, M ike Hobday, C hris
Freeman, Ian M iles, and Susan Lees provided invaluable assistance in editing,
proofreading, and preparing Keith P a v itfs chapter for publication. D u rin g the
final phase Charles M cC an n provided valuable advice to the non-English/A m erican
authors. O var A ndreas Johansson at T IK w as a very efficient and helpful project
assistant. At O xford U niversity Press D avid M usson and M atthew D erbyshire were
inspiring and patient partners.
C o n t e n t s
List o f Tables xi ii
List o f Boxes xv
1. In n o v a tio n : A G u id e to th e L ite ra tu re i
Jan Fa g er b er g
PARTI IN N O V A T IO N IN T H E M A K IN G
In tr o d u c tio n 28
2. T h e In n o v a tiv e F ir m 29
W illiam Lazonick
3. N e tw o rk s o f In n o v a to r s 56
W a l t e r W. P o w e l l a n d S t i n e G r oda l
4. In n o v a tio n P ro c e sse s 86
Keith Pa v i t t
6. M e a s u r in g In n o v a tio n 148
Ke it h Sm it h
X CONTENTS
P A R T II T H E S Y S T E M I C N A T U R E
OF IN N O V A T IO N
In tr o d u c tio n 18 0
9. F in a n c e a n d In n o v a tio n 240
M a r y O ’S u l l i v a n
PART III H O W I N N O V A T I O N D I F F E R S
Introduction 348
1 6. In n o v a tio n in S e rv ic e s 433
Ian M iles
In tr o d u c tio n 486
19. In n o v a tio n a n d C a tc h in g -u p 5 14
J a n F a g e r b e r g a n d M a n u e l M . G od in ho
Bjorn Asheim Professor, Departm ent o f Social and Econom ic G eography and
Centre for Innovation, Research and Com petence in the Learning E conom y
(C IR C LE), U niversity o f Lund, Sweden, and Centre for Technology, Innovation
and Culture (T IK ), U niversity o f Oslo, Norway.
Meric G ertler Professor, D epartm ent o f G eography and M u n k Centre for Inter
national Studies, U niversity o f Toronto, Canada, and Centre for Technology, In n o v
ation and Culture (T IK ), U niversity o f Oslo, N orway.
Ove G ran stran d Professor, Center for Intellectual Prop erty Studies (C IP ), D epart
ment o f Industrial M anagem ent and Econom ics, School o f Technology M anage
ment and Econom ics, C halm ers U niversity o f Technology, Sweden.
Stine G rod al D octoral Candidate in M anagem ent Science and Engineering, Stan
ford University, U SA .
Bhaven N. Sam pat Assistant Professor, School o f Public Policy, G eorgia Institute o f
Technology, USA.
Keith Sm ith Professor, D epartm ent o f Industrial D ynam ics, C h alm ers U niversity
o f Technology, Sweden.
B art Verspagen Professor, Eindhoven Centre for In novation Studies (Ecis), Eind
hoven U niversity o f Technology, the N etherlands, and Centre for Technology,
Innovation and Culture (T IK ), U niversity o f Oslo, N orw ay.
A ntonello Zan fei Professor, Faculty o f Econom ics, U niversity o f U rbino, Italy.
C H A P T E R 1
IN N O V A T I O N
A GUIDE TO THE
* ль* Ж. « Ж Ж . л Ж* Ж*. .Ж. JL Ж J
LITERATURE
JAN FAGERBERG
i.i In t r o d u c t io n 1
As a result, our knowledge about innovation processes, their determ inants and social
and econom ic im pact has been greatly enhanced.
W hen innovation studies started to em erge as a separate field o f research in the
1960s, it did so m ostly outside the existing disciplines and the m ost prestigious
universities. An im portant event in this process w as the form ation in 1965 o f the
Science Policy Research Unit (SPR U ) at the U niversity o f Sussex (see B o x 1.1). The
nam e of the center illustrates the tendency for innovation studies to develop unde
other (at the time m ore acceptable?) terms, such as, for instance, “ science studies” or
“ science policy st udies.” But as we shall see in the follow ing, one o f the m ain lessons
from the research that came to be carried out is that science is only one am ong several
ingredients in successful innovation. As a consequence o f these findings, not only the
focus o f research in this area but also the notions used to characterize it changed.
D uring the late twentieth/early twenty-first century, a n um ber o f new research
centers and departm ents have been founded, focusing on the role o f innovation in
innovation : a g u i d e to t h e l it e r a t u r e 3
SPRU— Science Policy Research Unit— at the University o f Sussex, UK was founded in
1965 with Christopher Freeman as its first director. From the beginning, it had a cross-
disciplinary research staff consisting o f researchers with backgrounds in subjects as
diverse as economics, sociology, psychology, and engineering. SPRU developed its own
cross-disciplinary Master and Ph.D. programs and carried out externally funded
research, much o f which came to focus on the role o f innovation in economic and
social change. It attracted a large number o f young scholars from other countries who
came to train and work here.
The research initiated at SPRU led to a large number o f projects, conferences, and
publications. Research Policy* which came to be the central academic journal in the
field, was established in 1972, with Freeman as the first editor (he was later succeeded by
Keith Pavitt, also from SPRU). Freemans influential book, The Economics o f Industrial
Innovation* was published two years later, in 1974, and has since been revised twice. In
1982, the book, Unemployment and Technical Innovation* written by Freeman, Clark,
and Soete, appeared, introducing a systems approach to the role o f innovation in long
term economic and social change. Freeman later followed this up with an analysis o f
the national innovation system in Japan (Freeman 1987). He was also instrumental in
setting up the large, collaborative IFIAS project which in 1988 resulted in the very
influential book, Technical Change and Economic Theory* edited by Dosi, Freeman,
Nelson, Silverberg, and Soete (both Dosi and Soete were SPRU Ph.D. graduates).
In many ways, SPRU came to serve as a role model for the many centers/institutes
within Europe and Asia that were established, mostly from the m id- 1980s onwards,
combining cross-disciplinary graduate and Ph.D. teaching with extensive externally
funded research. Most o f these, as SPRU itself, were located in relatively newly formed
(so-called “ red-brick” ) universities, which arguably showed a greater receptivity to
new social needs, initiatives, and ideas than the more inert, well-established academic
“ leaders,” or at other types o f institutions such as business or engineering schools.
SPRU graduates were in many cases instrumental in spreading research and teaching
on innovation to their own countries, particularly in Europe.
econom ic and so cial change. M a n y o f these have a cro ss-d iscip lin ary o rien tatio n ,
illustrating the need fo r in n o va tio n to be studied fro m different perspectives. Several
jou rn als an d p ro fessio n al association s have also been fo u n d ed .
The leaning tow ards cross-d isciplin arity that characterizes m uch scholarly w o rk in
this area reflects the fact that no single discipline deals w ith all aspects o f innovation.
Hence, to get a com preh ensive overview , it is necessary to com bine insights from
several disciplines. Traditionally, fo r instance, econom ics has dealt primarily w ith the
allocation o f resources to in n ovation (in com petition with other ends) and its
econom ic effects, w hile the in n o vation process itself has been m o re or less treated as
a “ black b o x ” W h at h ap pen s w ithin this “box” has been left to scholars from other
disciplines. A lo t o f w h at h appens o b vio u sly has to d o w ith learning, a central topic in
cognitive science. Such, learn in g occu rs in organized settings (e g . groups* teams, firms.
4 IAN F A G E R B E R G
and networks), the working o f which is studied w ithin disciplines such as sociology,
organizational science, management, and business studies. M oreover, as economic
geographers point out, learning processes tend to be linked to specific contexts or
locations. The way innovation is organized and its localization also undergo im port
ant changes through time, as underscored by the w o rk w ithin the field o f economic
history. There is also, as historians o f technology have pointed out, a specific techno
logical dim ension to this; the way innovation is organized, as well as its econom ic and
social effects, depends critically on the specific nature o f the technology in question.
Two decades ago, it was still possible for a h ard -w orkin g student to get a fairly
good overview o f the scholarly w ork on innovation b y devoting a few years o f
intensive study to the subject. N ot any m ore. Today, the literature on innovation is
so large and diverse that even keeping up-to-date w ith one specific field o f research is
very challenging. The purpose o f this volum e is to provide the reader w ith a guide to
this rapidly expanding literature. We do this under the follow in g b road headings:
I Innovation in the M aking
II The System ic N ature o f Innovation
III H ow Innovation Differs
IV Innovation and Perform ance.
Part One focuses on the process through which innovations occu r and the actors
that take part: individuals, firm s, organizations, and netw orks. A s we will discuss in
m ore detail below, innovation is by its very nature a system ic phenom enon, since it
results from continuing interaction between different actors and organizations. Part
Two outlines the systems perspective on innovation studies and discusses the roles o f
institutions, organizations, and actors in this process at the national and regional
level. Part Three explores the diversity in the m anner in w hich such system s work
over time and across different sectors or industries. Finally, Part Fo u r exam ines the
broader social and econom ic consequences o f innovation and the associated policy
issues. The rem ainder o f this chapter sets the stage for the discussion that follows by
giving a broad overview o f some o f the central topics in innovation studies (includ
ing conceptual issues).
1 .2 W h a t is In n o v a t io n ?
Joseph Schumpeter ( 1883- 1950) was one o f the most original social scientists o f A&
twentieth century* He grew up in Vienna around the turn o f the century* where he
studied law and economics. For most o f his life he worked as an academic* but he also
tried his luck as politician* serving briefly as finance minister in the first post-World
War I (socialist) government* and as a banker (without much success). He became
professor at the University o f Bonn in 1925 and later at Harvard University in the USA
( 1932), where he stayed until his death. He published several books and papers in
German early on, among these the Theory o f Economic Development, published in 1911
and in a revised edition in English in 1934*Among his most well-known later works are
Business Cycles in two volumes (from 1939) > Capitalism , Socialism and Democracy
(1943), and the posthumously published History o f Economic Analysis (1954)*
Very early he developed an original approach, focusing on the role o f innovation in
economic and social change. It was not sufficient, Schumpeter argued, to study the
economy through static lenses, focusing on the distribution o f given resources across
different ends. Economic development, in his view, had to be seen as a process of
qualitative change, driven by innovation, taking place in historical time. As examples
o f innovation he mentioned new products, new methods o f production, new sources
of supply, the exploitation of new markets, and new ways to organize business. He
defined innovation as “ new combinations” o f existing resources. This combinatory
activity he labeled “ the entrepreneurial function” (to be fulfilled by “ entrepreneurs” ),
to which he attached much importance. One main reason for the important role played
by entrepreneurs for successful innovation was the prevalence o f inertia, or “ resistance
to new ways” as he phrased it, at all levels of society that entrepreneurs had to fight in
order to succeed in their aims. In his early work, which is sometimes called “ Schump
eter Mark I ” Schumpeter focused mostly on individual entrepreneurs. But in later
works he also emphasized the importance o f innovation in large firms (so-called
“ Schumpeter Mark II” ), and pointed to historically oriented, qualitative research
(case studies) as the way forward for research in this area.
In his analysis of innovation diffusion, Schumpeter emphasized the tendency for
innovations to “cluster” in certain industries and time periods (and the derived effects
on growth) and the possible contribution o f such “clustering” to the formation of
business cycles and long waves” in the world economy (Schumpeter 1939). The latter
suggestion has been a constant source of controversy ever since. No less controversial,
and perhaps even better known, is his Inspired discussion o f the institutional changes
under capitalism (and its possible endogenous transformation into “ socialism” ) in the
book Capitalism, Socialism and Democracy (1943).
Sources. Swedberg 1991; Shionoya 1997; Fagerberg 2003.
organize business, H ow ever, in econom ics, m ost o f the focus has been on the tw o
first o f these. S ch m o o k ler (1966), fo r instance, in his classic work on “ Invention and
E con om ic G row th ,” argu ed that the distinction betw een “ product technology” and
“ p ro d u ctio n tech n o lo g y” w as “ critical” for o u r u n d erstan d in g o f this phenomenon
(ibid. 166). H e defin ed the fo rm er type as know ledge abou t h o w to create or improve
products, and the latter as know ledge ab ou t h ow to produce them . Similarly, the
term s “ p ro d u ct in n o va tio n ” and “ process in n o va tio n ” have been used to character-
ize the occurren ce o f new o r im p ro ved good s an d services, and im provem ents in the
w ays to p ro d u ce these g o o d and services, respectively,5 T h e argu m en t for focusing
particu larly on the d istin ctio n betw een p ro d u ct and process in n o vatio n often rests
on the assu m p tio n that their eco n o m ic and social im pact m a y differ. For instance,
while the in tro d u ctio n o f new p ro d u cts is c o m m o n ly assum ed to have a clear,
positive effect on grow th o f in com e and em plo ym en t, it has been argu ed that process
in n ovation , due to its co st-cu ttin g nature, m ay have a m o re am b ig u o u s effect
(Edquist et al. 2001; P ian ta in this vo lu m e). H ow ever, w h ile clearly d istin guish able
at the level o f the in d ivid u al firm or indu stry, such differences tend to becom e
blurred at the level o f the overall econom y, because the p ro d u ct o f o n e firm (or
industry) m ay end up as b ein g used to p ro d u ce good s o r services in another.6
The focus o n p ro d u ct and process in n o vation s, w h ile u sefu l fo r the analysis o f
som e issues, sh o u ld not lead us ign ore other im p o rtan t aspects o f in n o vatio n . F o r
instance, d u rin g the first h a lf o f the tw entieth century, m an y o f the in n o vatio n s that
m ade it possible fo r the U n ited States to “ forge ah ead ” o f other capitalist econ o m ies
were o f the o rgan izatio n al k in d , in vo lvin g entirely new w ays to organ ize p ro d u ctio n
and d istrib u tion (see B ru la n d an d M o w e ry in this vo lu m e, w hile L a m pro vid es an
overview o f o rgan izatio n al in n o vatio n ). E d q u ist et al. (2001) have suggested d iv id
ing the catego ry o f process in n o va tio n into “ tech n ological process in n o va tio n s” and
“ organizational p ro cess in n o vation s,” the fo rm er related to n ew types o f m ach in ery,
and the latter to n ew w ays to organ ize w o rk . H ow ever, organ izatio n al in n o vatio n s
are not lim ited to n ew w ays to organize the process o f p ro d u ctio n w ith in a given
firm . O rgan ization al in n o va tio n , in the sense used b y Sch u m p eter ,7 also includes
arrangem ents across firm s su ch as the reo rgan izatio n o f entire indu stries. M oreover,
as exem plified b y the case o f the U S A in the first h a lf o f the p revio u s century, m an y o f
the m ost im p o rta n t o rgan izatio n al in n o vatio n s have occu rred in d istrib u tio n , w ith
great consequences fo r a w h o le ran ge o f indu stries (C h an d ler 1990)*
A n oth er ap p ro ach , also based o n S c h u m p e te rs w o rk , has been to classify innov
ations acco rd in g to h o w rad ical th ey are co m p ared to cu rren t technology (Freeman
and Soete 1997). F ro m this perspective, co n tin u o u s im p ro vem en ts o f the type
referred to above are o ften characterized as “ in crem en tal” or “ marginal” innov
ations,8 as o p p o se d to “ rad ical” in n o vatio n s (such as the introduction o f a totally
new type o f m ach in ery) o r “ tech n o logical rev o lu tio n s” (con sistin g o f a cluster o f
in n o vation s that togeth er m a y have a v e ry far-reaching impact)* Schumpeter
focused in p a rtic u la r o n the latter tw o categories, which he believed to be o f greater
8 JAN F A G E R B E R G
im portance. It is a w idely held view, however, that the cum ulative im pact of
increm ental innovations is just as great ( if not greater), and that to ignore these
leads to a biased view o f long run econom ic and social change (Lun d vail et aL 1992).
M oreover, the realization o f the econom ic benefits fro m radical innovations in
m ost cases (including those o f the airplane and the autom obile, discussed earlier)
requires a series o f incremental im provem ents. A rguably, the b u lk o f economic
benefits com e from incremental innovations and im provem ents.
There is also the question o f how to take different contexts into account. I f A for
the first time introduces a particular innovation in one context, w hile B later
introduces the sam e innovation in another, w ou ld we characterize both as innov
ators? This is a m atter o f convention. A w idely used practice, based on Schum peter’s
work, is to reserve the term innovator for A and characterize B as an im itator. But
one might argue that, follow ing Schum peter’s ow n d efin ition, it w ou ld be equally
consistent to call B an Innovator as well, since B is in trodu cin g the innovation for the
first tim e in a new context. This is, for instance, the position taken b y H obday (2000)
in a discussion o f innovation in the so-called “ new ly indu strializing countries” in
A sia.9 One might object, though, that there is a qualitative difference between (a)
com m ercializing som ething for the first tim e and ( b) copyin g it and introducing it in
a different context. The latter arguably includes a larger dose o f im itative behavior
(im itation), or what is som etim es called “ technology transfer.” T h is does not
exclude the possibility that im itation m ay lead to new in n o vation (s). In fact, as
pointed out by Kline and Rosenberg (1986, see B o x 1.3), m any econom ically sig
nificant innovations occur while a product or process is diffusing (see also Hall in
this volum e). Introducing som ething in a new context often im plies considerable
adaptation (and, hence, increm ental innovation) and, as h isto ry has show n, organ
izational changes (or innovations) that m ay significantly increase productivity and
competitiveness (see G odinho and Fagerberg in this v o lu m e ).10
The problems with this model, Kline and Rosenberg point out, are twofold. First,
it generalizes a chain of causation that only holds for a minority o f innovations.
Although some important innovations stem from scientific breakthroughs, this is
not true most o f the time. Firms normally innovate because they believe there is a
commercial need for it, and they commonly start by reviewing and combining existing
knowledge. It is only if this does not work, they argue, that firms consider investing in
research (science). In fact, in many settings, the experience o f users, not science, is
deemed to be the most important source o f innovation (von Hippel 1988; Lundvall
1988). Second, “ the linear model” ignores the many feedbacks and loops that occur
between the different “ stages” o f the process. Shortcomings and failures that occur at
various stages may lead to a reconsideration o f earlier steps, and this may eventually
lead to totally new innovations.
1 .3 In n o v a t io n in t h e M a k in g
inertia, with the latter seen as (partly) endogenous. T h is may, to som e extent, have
been an adequate interpretation o f events in Europe aro u n d the turn o f the nine
teenth century. But during the first decades o f the tw entieth century, it becam e clear
to observers that innovations increasingly involve team w ork and take place within
larger organizations (see Bruland and M o w ery (Ch. 13)? L am (C h. 5)> and Lazonick
(Ch. 2) in this volum e). In later work, Schum peter acknow ledged this and empha
sized the need for systematic study o f "coop erative” entrepreneurship in big firms
(so-called "Schum peter M ark II” ). However, he did not analyze the phenom enon in
m uch detail (although he strongly advised others to ).11
System atic theoretical and em pirical w ork on in n ovation -projects in firm s (and
the m anagem ent o f such projects) was slow to evolve, but d u rin g the last decades a
quite substantial literature has emerged (see chapters b y Pavitt and Lam in this
volum e). In general, research in this area coincides w ith Schum peter's em phasis on
uncertainty (Nelson and W inter 1982; N onaka and Takeuchi 1995; Van de Yen et a l
1999). In particular, for potentially rew arding innovations, it is argued, one may
sim ply not know what are the m ost relevant sources o r the best options to pursue
(still less how great the chance is o f success).12 It has also been em phasized that
innovative firm s need to consider the potential problem s that "p ath dependency”
m ay create (Arthur 1994). For instance, if a firm selects a specific innovation path
very early, it m ay (if it is lucky) enjoy “ first m over” advantages. But it also risks being
“ locked in” to this specific path through various self-reinforcing effects. I f in the end
it turns out that there actually existed a superior path, w hich som e other firm
equipped with m ore patience (or luck) happened to find, the early m over may be
in big trouble because then, it is argued, it m ay sim p ly be too costly or too late to
switch paths. It has been suggested, therefore, that in the early phase o f an inno vation
project, before sufficient knowledge o f the alternatives is generated, the best strategy
m ay sim ply be to avoid being “ stuck” to a particular path, and rem ain open to
different (and com peting) ideas/solutions. At the level o f the firm , this requires a
pluralistic leadership” that allows for a variety o f com peting perspectives ( Van de
Ven et al. 1999), in contrast to the hom ogenous, u n itary leader style that, in the
m anagem ent literature, is som etim es considered as the m ost advan tageou s.13
Openness to new ideas and solutions? is considered essential for innovation
projects, especially in the early phases. The principal reason for this has to do with a
fundam ental characteristic o f innovation: that every new in n ovation consists o f a
new com bination o f existing ideas, capabilities, skills, resources, etc. It follows
logically from this that the greater the variety o f these factors w ithin a given system,
the greater the scope for them to be com bined in different ways, producing new
innovations which will be both m ore com plex and m ore sophisticated. This evolu
tionary logic has been used to explain why, in ancient tim es, the inhabitants of the
large Eurasian landmass came to be m ore innovative, and technologically sophisti
cated, than small, isolated populations elsewhere around the globe (D iam on d 1998).
Applied m echanically on a population o f firm s, this logic m ight perhaps be taken to
innovation : a g u i d e to t h e l i t e r a t u r e П
1.4 T h e S y s t e m ic N a t u r e of I n n o v a t io n
As is evident from the preceding discussion, a central fin d in g in the literature is that,
in m ost cases, innovation activities in firm s depend h eavily on external sources. One
recent study sums it up well: “ Popular folklore notw ithstandin g, the innovation
journey is a collective achievem ent that requires key roles fro m n u m erou s entrepre
neurs in both the public and private sectors” (Van de Yen et al. 1999:149)* In that
particular study, the term “ social system for innovation d evelopm en t” was used to
characterize this “ collective achievem ent ” H ow ever, this is ju st one am ong several
exam ples from the last decades o f h ow system concepts are applied to the analysis of
the relationship between innovation activities in firm s and the w ider fram ework in
which these activities are em bedded (see Edquist, C h. 7 in this volu m e).
One m ain approach has been to delineate system s on the basis o f technological,
industrial, or sectoral characteristics (Freem an et al. 1982; H ughes 1983; Carlsson and
Stankiewicz 1991; M alerba, Ch. 14 in this volum e) but, to a varyin g degree, to include
other relevant factors such as, for instance, institutions (laws, regulations, rules,
habits, etc.), the political process, the public research infrastructure (universities,
research institutes, support from public sources, etc.), financial institutions, skills
(labor force), and so on. To explore the technological dynam ics o f innovation, its
various phases, and how this influences and is influenced b y the w ider social,
institutional, and econom ic fram eworks has been the m ain focus o f this type of
analysis. Another im portant approach in the innovation-system s literature has
focused on the spatial level, and used national or regional borders to distinguish
between different systems. For exam ple, Lund vail (1992) and N elson et al. (1993)
have used the term national system o f innovation” to characterize the systemic
interdependencies within a given cou ntry (see Edquist in this volum e), while
Braczyk et al. (1997) sim ilarly have offered the notion o f “ regional innovation
systems (see Asheim and Gertler, Ch. 11 in this volu m e). Since the spatial systems
are delineated on the basis o f political and adm inistrative borders, such factors
innovation : a g u i d e to t h e l i t e r a t u r e 13
n atu rally tend to play an im p o rtan t role in analyses based on this approach, which
has proven to be influ ential am o n g po licy m akers in this area, especially in Europ e
(see Lu n dvall and B o rras, C h. 22 in this volu m e). (Part II o f this v o lu m e analyzes
som e o f the constituent elem ents o f such system s in m o re d e t a il15)
W hat are the im p licatio n s o f ap p lyin g a system perspective to the study o f
in n ovation ? System s are— as netw orks— a set o f activities (or actors) that are
interlinked, and this leads n atu rally to a focus on the w o rk in g o f the linkages o f
the system .16 Is the poten tial fo r co m m u n icatio n and interaction th rou gh existing
linkages su fficien tly exploited? A re there po ten tial linkages w ith in the system that
m ight p ro fitab ly be established? Such questions ap p ly o f course to netw orks as well
as system s. H ow ever, in the n o rm al usage o f the term , a system w ill typ ically have
m o re “ stru ctu re” than a netw ork, and be o f a m ore en d u rin g character. T h e structu re
o f a system w ill facilitate certain patterns o f in teraction and ou tcom es (an d constrain
others), and in this sense there is a parallel to the role o f “ in ertia” in firm s. A d yn am ic
system also has feedbacks, w h ich m ay serve to reinforce— o r w eaken— the existin g
structure/ fu n ctio n in g o f the system , lead in g to “ lo ck in “ (a stable co n fig u ratio n ), o r
a change in o rien tatio n , o r— eventually— the d issolu tion o f the system . H ence,
system s m ay— ju st as firm s— be locked into a specific path o f d evelopm en t that
su p po rts certain types o f activities and constrains others. T h is m a y be seen as an
advantage, as It pu shes the p articip atin g firm s an d other actors in the system in a
direction that is d eem ed to be beneficial. B u t it m ay also be a d isadvantage, i f the
co n figu ratio n o f the system leads firm s to ign ore p o ten tially fru itfu l avenues o f
exp loratio n . T h e ch aracter o f such processes w ill be affected b y the extent to w h ich
the system exchanges im pu lses w ith its en viron m en t. T h e m o re open a system is fo r
im pulses fro m o u tsid e, the less the chance o f being “ locked o u t” fro m p ro m isin g
new path s o f d evelo p m en t that em erge ou tsid e the system . It is, therefore, im p o rtan t
for “ system m an ag ers” — such as p o licy m akers— to keep an eye o n the openness o f
the system , to avo id the p o ssib ility o f in n o vatio n activities b eco m in g u n d u ly c o n
strained b y self-rein fo rcin g path -d ep en den cy.
A n o th er im p o rta n t feature o f system s that h as com e into focu s is the stron g
com plem en tarities that co m m o n ly exist betw een the co m p o n en ts o f a system . If, in a
d yn am ic system , on e critical, co m p le m e n tary co m p o n en t is lackin g, o r foils to
progress o r develop, this m a y b lo ck o r slo w d o w n the grow th o f the entire system .
T h is is, as p o in ted o u t earlier, one o f the m ain reasons w h y there is o ften a v e ry
co n siderab le tim e lag betw een in ven tion an d in n o vatio n . E c o n o m ic h isto rian s h ave
co m m o n ly used concep ts such as “ reverse salients” an d “ b ottlen ecks” to ch aracter
ize such p h en o m en a (H u gh es 1983; R o sen b erg 1982). H ow ever, such co n strain ts
need n o t be o f a p u re ly tech n ical ch aracter (such as, fo r instance, the failu re to invent
a decent b attery, w h ic h has severely con strain ed the d iffu sio n o f electric cars fo r
m o re th an ce n tu ry ), b u t m a y h ave to d o w ith lack o f p ro p e r in firn tractu re, fin an ce,
skills, etc. Some o f the m o st im p o rta n t in n o va tio n s o f this century, su ch as d e e tric it j
an d au to m o b ile s (M o w e ry a n d R o sen b erg 19 9 8 ), w ere d e j» id e n t o n v e ry ш т ш т
14 JAN FAGERBERG
1.5 How In n o v a t io n D if f e r s
One o f the striking facts about innovation is its variab ility over tim e and space* It
seems, as Schum peter (see Box 1.2) pointed out, to “ cluster,” not only in certain
sectors but also in certain areas and tim e periods. O ver tim e the centers o f innov
ation have shifted from one sector, region, and co u n try to another. Fo r instance, for a
long period the worldw ide center o f innovation was in the U K , and the productivity
and incom e o f its population increased relative to its n eighb orin g countries, so that
by the m id-nineteenth century its productivity (and incom e) level was 50 per cent
higher than elsewhere; at about the beginning o f the twentieth cen tu ry the center of
innovation, at least for the m odern chem ical and electrical technologies o f the day,
shifted to Germ any; and now, for a long tim e, the w orld w ide center o f innovation
has been in the U SA, which during m ost o f the twentieth centu ry enjoyed the highest
productivity and living standards in the world. As explained b y Bruland and
M ow ery in this volum e, the rise o f the U S to w orld technological leadership was
associated with the growth o f new industries, based on the exploitation of
econom ies o f scale and scope (Chandler 1962, 1990) and m ass production and
distribution.
H ow is this dynam ic to be explained? Schum peter, extending an earlier line of
argum ent dating back to Karl M a rx ,17 held technological com petition (competition
through innovation) to be the driving force o f econom ic developm ent. I f one firm in
a given industry or sector successfully introduces an im p o rtan t innovation, the
argum ent goes, it will be am ply rewarded b y a higher rate o f profit. This functions
innovation : a g u i d e to t h e l i t e r a t u r e 15
as a signal to other firm s (the im itators), w hich, if entry conditions allow, will
“ sw arm ” the in d u stry o r sector w ith the h ope o f sh aring the benefits (with the result
that the initial in n o vato r's first m over advantages m ay be quickly eroded). This
“ sw arm in g ” o f im itators im plies that the grow th o f the sector or industry in which
the in n o vatio n occu rs w ill be quite high for a while. S o o n er o r later, however, the
effects on grow th (created b y an in n o vatio n ) w ill be depleted and growth w ill slow
dow n.
To this essentially M a rx ia n sto ry Sch u m p eter added an im p o rtan t modification.
Im itators, he argued, are m u ch m o re likely to succeed in their aim s if they improve
on the origin al in n o vation , i.e., becom e in n o vato rs them selves. T h is is all the more
natural, he co n tin u ed , because one (im p o rtan t) in n o va tio n tends to facilitate
(induce) other in n o vatio n s in the sam e or related fields. In this way, in n o v a tio n -
d iffu sio n becom es a creative process— in w h ich one im p o rtan t in n o vatio n sets the
stage for a w h o le series o f subsequent in n o vatio n s— and n o t the passive, adaptive
process often assu m ed in m u ch d iffu sio n research (see H all in this v o lu m e). T h e
system ic interdepend en cies betw een the initial and indu ced in n o vatio n s also im ply
that In n ovation s (and grow th) “ tend to concentrate in certain sectors and their
su rro u n d in g s” o r “ clu sters” (Sch u m p eter 1939: 10 0 - 1) . Schum peter, as is well
kn o w n , lo o ked at this d yn am ic as a possible exp lan ato ry facto r b eh in d business
cycles o f vario u s lengths (Freem an and Lou^a 2001).
This sim ple schem e has been rem arkably successful in inspiring applications in
different areas. Fo r instance, there is a large am ou n t o f research that has adapted the
M an c-Sch u m p eter m odel o f technological com petition to the study o f industrial
grow th, international trade, and com petitiveness,18 although som etim es, it m ust be
said, w ith ou t acknow ledging the source for these ideas. A n early and very influential
contribution w as the so-called “ product-life-cycle th eo ry” suggested b y V ernon
(1966), in w hich indu strial grow th follow in g an im po rtan t product in n o vation w as
seen as com posed o f stages, characterized b y changing conditions o f and location o f
p ro d u ctio n .19 Basically w hat w as assum ed w as that the ab ility to d o produ ct in n o v
ation m attered m ost at the early stage, in w hich there w ere m an y different and
com peting versions o f the produ ct o n the m arket. H ow ever, w ith tim e, the p ro d u ct
was assum ed to standardize, and this was assum ed to be accom panied by a greater
em phasis o n process innovation, scale econom ics, and cost-com petition. It was
argued that these changes in co m p etitive co n d itio n s m igh t initiate tran sfer o f the
tech n o lo gy fro m the in n o va to r c o u n try (h igh in com e) to co u n tries w ith large
m arkets an d /o r lo w costs. Su ch transfers m igh t also be associated w ith in tern atio n al
capital flo w s in the fo rm o f so-called fo reign d irect investm ents (F D Is), an d the
th eo ry has therefore also b eco m e k n o w n as a fra m e w o rk fo r exp lain in g such flo w s
(see N aru la an d Z a n fe i in this vo lu m e).
The “ product-life-cycle theory,” attractive as it was in its simplicity, was not
always corroborated by subsequent research. While it got some o f the general,
conjectures (b o rro w ed fro m Schumpeter) right, the rigorous scheme it aided.
16 JAN F A GE RB E R G
Ch. 14 by M alerb a; C h . 15 by V onT unzelm ann and Acha; Ch, 16 by M iles). Inspired, to
a large extent, b y the sem inal w o rk by N elson and W inter (see Box 1.5), research in
this area has exp lored the m an n er in w hich in du stries and sectors differ in terms o f
their internal d yn am ics (or “ tech nological regim es” : see M alerb a and O rsenigo
1997), fo cu sin g, in particular, on the differences across sectors in knowledge bases,
actors, n etw orks, and in stitu tion s (so called “ sectoral system s” : see M alerb a, Ch. 14
in this volu m e). A n im p o rtan t result fro m this research is that, since the factors that
influence in n o vatio n differ across indu stries, p o licy m akers have to take such
differences into accou n t w h en designing policies. T h e sam e p o licy (and policy
instrum ents) w ill not w o rk equ ally well everyw here.
The book An Evolutionary Theory o f Economic Change (1982) by Richard Nelson and
Sidney Winter is one o f the most important contributions to the study o f innovation
and long run economic and social change. Nelson and Winter share the Schumpeterian
focus on “capitalism as an engine o f change” However, building on earlier work by
Herbert Simon and others (so-called “procedural” or “ bounded” rationality), Nelson
and Winter introduce a more elaborate theoretical perspective on how firms behave. In
Nelson and Winter’s models, firms’ actions are guided by routines, which are repro
duced through practice, as parts o f the firms’ “ organizational memory.” Routines
typically differ across firms. For instance, some firms may be more inclined towards
innovation, while others may prefer the less demanding (but also less rewarding)
imitative route. I f a routine leads to an unsatisfactory outcome, a firm may use its
resources to search for a new one, which— if it satisfies the criteria set by the firm— will
eventually be adopted (so-called “ satisficing” behavior).
Hence, instead o f following the common practice in much economic theorizing o f
extrapolating the characteristics o f a “ representative agent” to an entire population
(so-called “ typological thinking” ), Nelson and Winter take into account the social, and
economic consequences o f interaction within populations o f heterogeneous actors (so-
called “ population thinking” ). They also emphasize the role o f chance (the stochastic
element) in determining the outcome o f the interaction. In the book, these outcomes
are explored through simulations, which allow the authors to study the consequences
off varying the value o f key parameters (to reflect different assumptions on techno-
logical progress, firm behavior, 'etc.). ■ They, 'distinguish ■ between a n ' тштттШш:
regime” Ы which. Ae-teAnologi^d/.iromfier Is assumed t o - o f f -
1.6 In n o v a t io n a n d E c o n o m ic
Per fo r m an ce
The M arx-Schum peter model was not intended as a m odel o f industrial dynamics; its
prim ary purpose was to explain long run econom ic change, what Schum peter called
“ development ” The core o f the argument was (1) that technological com petition is the
m ajor form o f com petition under capitalism (and firm s not responding to these
demands fail), and (2) that innovations, e.g. 'n e w com binations"5 o f existing know
ledge and resources, open up possibilities for new business opportunities and future
innovations, and in this way set the stage for continuing change. This perspective, while
convincing, had little influence on the econom ics discipline at the tim e o f its publica
tion, perhaps because it did not lend itself easily to form al, m athem atical modeling of
the type that had become popular in that field. M ore recently, however, economists
(Rom er 1990), drawing on new tools for m athem atical m odeling o f econom ic phe
nomena, have attempted to introduce som e o f the above ideas into form al growth
models (so-called4new growth theory55 or “ endogenous grow th theory55).21
In developing this perspective, Schum peter (1939) w as, as noted, particularly
concerned with the tendency o f innovations to “ cluster55 in certain contexts, and
the resulting structural changes in production, organization, dem and, etc. Although
these ideas were not well received b y the econom ic co m m u n ity at the tim e, the big
slum p in econom ic activity w orldw ide during the 1970s led to renew ed attention,
and several contributions emerged view ing long run econom ic and social change
from this perspective. Both M ensch (1979) and Perez (19 8 3,19 8 5 ), to take just two
exam ples, argued that m ajor technological changes, such as, fo r instance, the ICT
revolution today, or electricity a century ago, require extensive organizational and
institutional change to run their course. Such change, how ever, is difficult because of
the continuing influence o f existing organizational and institutional patterns. They
saw this inertia as a m ajor grow th-im peding factor in periods o f rapid technological
change, possibly explaining some o f the variation o f grow th over tim e (e.g. booms
and slumps) in capitalist econom ies. W hile the latter p ro p o sitio n rem ains contro
versial, the relationship between technological, organizational, and institutional
change continues to be an im portant research issue (Freem an and Lou^a 2001),
with im portant im plications both for the analysis of the d iffu sion o f new technolo
gies (see Hall in this volum e) and the policy discourse (see Lu ndvall and Borras in
this volum e).
A lthough neither M arx nor Schum peter applied their dynam ic perspective to the
analysis o f cross-national differences in growth perform ance, fro m the early 1960s
onwards several contributions emerged that explore the potential o f this perspective
for explaining differences in cross-country grow th. In what cam e to be a very
influential contribution, Posner (1961) explained the difference in econom ic growth
innovation : a g u i d e to t h e l i t e r a t u r e 19
1.7 W h a t do w e K n o w a b o u t In n o v a t io n ?
A n d w h a t d o w e N e e d t o
L e a r n m o r e a b o u t ?
Arguably, we have a good understanding of the role played by innovation in long run
economic and social change, and many of its consequences:
20 JAN F A G E RB E R G
N otes
1. I wish to thank my fellow editors and contributors for helpful comments and suggestions.
Thanks also to Ovar Andreas Johansson for assistance in the research, Sandro Mendon^a
for his many creative inputs (which I unfortunately have not have been able to follow to
the extent that he deserves), and Louise Earl for good advice. The responsibility for
remaining errors and omissions is mine.
2. A consistent use o f the terms invention and innovation might be to reserve these for the
first time occurrence o f the idea/concept and commercialization, respectively. In practice
it may not always be so simple. For instance, people may very well conceive the same idea
independently o f one another. Historically, there are many examples o f this; writing,
for instance, was dearly invented several times (and in different cultural settings)
throughout history (Diamond 1998). Arguably, this phenomenon may have been
reduced in importance over time, as communication around the globe has progressed.
3. In the sociological literature on diffusion (i.e, spread o f innovations), it is common to
characterize any adopter o f a new technology, product, or service an innovator. This then
leads to a distinction between different types o f innovators, depending on how quick they
are in adopting the innovation, and a discussion o f which factors might possibly explain
such differences (Rogers 1995). While this use o f the terminology may be a useful one in
the chosen context, it dearly differs from the one adopted elsewhere. It might be preferable
to use terms such as “ imitator” or “ adopter” for such cases.
4. Similarly for automobiles: while the idea o f a power-driven vehicle had been around for a
long time, and several early attempts to commercialize cars driven by steam, electricity,
and other sources had been made, it was the incorporation o f an internal combustion
engine driven by low-cost, easily available petrol that made the product a real hit in the
market (Mowery and Rosenberg 1998).
5. A somewhat similar distinction has been suggested by Henderson and Clark (1990). They
distinguish between the components (or modules) o f a product or service and the way these
components are combined, e.g. the product “ d e s ip ” or “ architecture” A change only in
the former is dubbed “ modular innovation” change only in the latter “architectural
innovation” They a r p e that these two types o f innovation rely on different types o f
knowledge (and, hence, create different challenges for the firm).
22 IAN F A G E R B E R G
6. In fact, many economists go so far as to argue that the savings in costs, following a process
innovation in a single firm or industry, by necessity will generate additional income and
demand in the economy at large, which will “ compensate” for any initial negative effects
of a process innovation on overall employment. For a rebuttal, see Edquist 2001 and
Pianta, Ch. 21 in this volume.
7. Schumpeter 1934: 66.
8. In the sociological literature on innovation the term "reinvention” is often used to
characterize improvements that occur to a product or service, while it is spreading in a
population of adopters (Rogers 1995)*
9. In the Community Innovation Survey (CIS) firms are asked to qualify novelty with
respect to the context (new to the firm, industry or the world at large). See Smith in this
volume for more information about these surveys.
10. Kim and Nelson (2000a) suggest the term "active imitation” for producers who, by
imitating already existing products, modify and improve them.
11. For Instance, in one o f his last papers, he pointed out: "To let the murder out and start my
final thesis, what is really required is a large collection o f industrial and locational
monographs all drawn up according to the same plan and giving proper attention on
the one hand to the incessant historical change in production and consumption func
tions and on the other hand to the quality and behaviour o f leading personnel”
(Schumpeter 1949/1989:328).
12. Even in cases where the project ultimately is successful in aims, entrepreneurs face the
challenge of convincing the leadership o f the firm to launch it commercially (which may
be much more costly than developing it). This may fail if the leadership o f the firm has
doubts about its commercial viability. It may be very difficult for management to foresee
the economic potential of a project, even if it is "technically” successful Remember, for
instance, IBM director Thomas Watsons dictum in 1948 that “ there is a world market for
about five computers” (Tidd et al. 1997: 60)!
13. "A unified homogenous leadership structure is effective for routine trial-and-error
learning by making convergent, incremental improvements in relatively stable and
unambiguous situations. However, this kind o f learning is a conservative process that
maintains and converges organizational routines and relationships towards the existing
strategic vision. . . although such learning is viewed as wisdom in stable environments, it
produces inflexibility and competence traps in changing worlds” (Van de Yen et al.
1999:117).
14. It would also imply that large countries should be expected to be more innovative than
smaller ones, consistent with, for instance, the prediction o f so-called "new growth”
theory (Romer 1990)- See Verspagen in this volume.
15. See, in particular, Ch, 10 by Granstrand (intellectual property rights), Ch. 8 by Mowery
and Sampat (universities and public research infrastructure), and Ch. 9 by O'Sullivan
(finance).
16. This is essentially what was suggested by Porter (1990).
17. See Fagerberg 2002, 2003 for a discussion of this “ Marx-Schumpeter” model.
18. See Fagerberg (1996), Wakelin (1997), and Cantwell, Ch. 20 in this volume for overviews
o f some of this literature.
19. For a more recent analysis in this spirit, with a lot o f empirical case-studies, see Utterback
(1994).
in n o v a t io n : a g u id e to th e lit e r a t u r e 23
20** Available econometric evidence suggests that innovation, measured in various ways (see
Smith in this volume), matters in many industries, not only those which could be
classified as being in the early stage o f the product-cycle (Soete 1987; Fagerberg 1995),
21, For an overview, see Aghion and Howitt (1998). See also the discussion in Fagerberg
(2002, 2003), and Ch. 18 by Verspagen in this volume.
22. For a discussion o f the role o f different types o f knowledge in economics, including the
organizational dimension, see Cowan et a l (2000) and Ancori et a l (2000).
R efer en c es
K rugm an , P. (i979)> UA Model o f Innovation, Technology Transfer and the World Distribu
tion o f Incom e” Journal of Political Economy 87: 253-66.
Lu n d vall , B. A. (1988), "‘Innovation as an Interactive Process: From User-Producer Inter
action to the National System o f Innovation” in Dosi et al. 1988: 349—69.
------ (ed.) (1992), National Systems of Innovation: Towards a Theory of Innovation a n d
Interactive Learning, London: Pinter.
M a ler b a , R, and O rsenigo , L. (1997), “ Technological Regimes and Sectoral Patterns o f
Innovative Activities” Industrial and Corporate Change6: 83-117.
------ N elson , R. R., O rsenigo , L., and W in ter , S. G. (1999), “ ‘History-friendly' Models o f
Industry Evolution: The Computer Industry,” Industrial Dynamics and Corporate Change
8 :1-36.
M ensch , G. (1979), Stalemate in Technology, Cambridge, Mass.: Ballinger Publishing C om
pany.
M etc a lfe , J. S. (1998), Evolutionary Economics and Creative Destruction, London: Routledge.
*M ow ery , D., and Ro senberg , N. (1998), Paths of Innovation, Technological Change in 20th-
Century America, Cambridge: Cambridge University Press.
N elson , R. R. (ed.) (1993), National Systems of Innovation: A Comparative Study; Oxford:
Oxford University Press.
------ and W in ter , S. G. (1982), An Evolutionary Theory o f Economic Change, Cambridge,
Mass.: Harvard University Press.
*N o naka , I., and T ak eu ch i , H. (1995), The Knowledge Creating Company, Oxford: Oxford
University Press.
*P a v it t , K. (1984), “ Patterns o f Technical Change: Towards a Taxonomy and a Theory,”
Research Policy 13: 343-74.
Perez , C. (1983), “ Structural Change and the Assimilation o f New Technologies in the
Economic and Social System,” Futures 15: 357-75.
— (1985), “ Micro-electronics, Long Waves and World Structural Change,” World Develop
ment 13: 441-63.
Porter , M. E. (1990), “ The Competitive Advantage ofNations,” H arvard Business Review 68:
73- 93.
Posner , M. V. (1961), “ International Trade and Technical Change,” Oxford Economic Papers
13: 323-41.
^Rogers , E. (1995), Diffusion o f Innovations, 4th edn., New York: The Free Press.
Romer , P. M. (1990), “ Endogenous Technological Change,” Journal o f Political Economy 98:
S71-S102.
Rosenberg , N. (1976), Perspectives on Technology, New York: Cambridge University Press.
------ (1982), Inside the Black Box: Technology and Economics, New York: Cambridge Univer
sity Press.
S chmookler , J. (1966), Invention and Economic Growth, Cambridge, Mass.: Harvard Uni
versity Press.
Sch u m peter , J. (1934), The Theory o f Economic Development, Cambridge, Mass.: Harvard
University Press.
------ (1939)» Business Cycles: A Theoretical, Historical, and Statistical Analysis o f the Capitalist
Process, 2 vols., New York: McGraw-Hill.
* ------ (1943), Capitalism, Socialism and Democracy, New York: Harper.
-— — (1949), “ Economic Theory and Entrepreneurial History* Change m i the Ernmprmmm
63-84, repr. in J. Schumpeter (1989), Essays on Entrepreneurs, Innovations» Business
26 JAN FAGERBERG
and the Evolution of Capitalism, ed. Richard V. Clemence, New Brunswick, NJ: Transaction
Publishers, 253-61.
Schumpeter , R. (1954), History o f Economic Analysis, New York: Allen & Unwin.
Shionoya, Y. (1997), Schumpeter and the Idea o f Social Science, Cambridge: Cambridge
University Press.
S oete , L. (1987), “ The Impact o f Technological Innovation on International Trade Patterns;
The Evidence Reconsidered” Research Policy 16 :101-30 .
Sw edberg , R, (1991), Joseph Schumpeter: His Life and Work, Cambridge: Polity Press.
T idd, J., B ess an t , J., and Раѵітт , K. (1997), Managing Innovation: Integrating Technological
Market and Organizational Change, Chichester: John Wiley 8c Sons.
T ushman , M. L., and A nderson , P. (1986). “ Technological Discontinuities and Organiza
tional Environments ” Administrative Science Quarterly 31(3): 439--65.
U tterba ck , J. M. (1994), Mastering the Dynamics of Innovation, Boston: Harvard Business
School Press.
V an de V en , A., P olley , D. E., G arud , R., and V en k a ta r a m a n , S. (1999), The Innovation
Journey; New York: Oxford University Press.
V ernon , R. (1966), “ International Investment and International Trade in the Product Cycle,”
Quarterly Journal of Economics So: 190-207.
V on H ip p e l , E. (1988), The Sources of Innovation, New York: Oxford University Press.
W a k elin , K. (1997)> Trade and Innovation: Theory and Evidence, Cheltenham: Edward Elgar.
W a lk er , W. B. (1979), Industrial Innovation and International Trading Performance, Green
wich: JAI Press.
I NNOVATI ON IN
THE M A K I N G
In t r o d u c t io n t o P a r t I
THE INNOVATIVE
FIRM
WILLIAM LAZONICK
2.1 In t r o d u c t io n
Over the past few decades, the greatest challenges to the U S m anagerial corporation
have com e from Japan. Section 2.5 identifies the social co n d itio n s o f innovative
enterprise that have characterized the Japanese m odel, w hile Section 2.6 outlines the
distinctive characteristics o f the U S N ew E con om y firm that has gained competitive
advantage in a num ber o f critical product m arkets in the in fo rm atio n and commu
nication technology (IC T) industries. Section 2.7 draw s som e general conclusions
from this essay's com parative—historical perspective con cern in g strategy, finance,
and organization in the innovative firm , and the m eth o d o lo g y for studying these
phenom ena.
2.2 S o c i a l C o n d i t i o n s o f
In n o v a t iv e E n t e r p r is e
Firm s strategize when they choose the product m arkets in w h ich they want to
com pete and the technologies w ith which they h ope to be com petitive. Firm s finance
when they m ake investments to transform technologies and access m arkets that can
only be expected to generate revenues som etim e in the future. Firm s organize when
they com bine resources in the attem pt to tran sform them into saleable products,
To strategize, finance, and organize is not necessarily to innovate. B y definition,
innovation requires learning about how to tran sform technologies and access
markets in ways that generate higher quality, low er cost products. Learning is a
social activity that renders the innovation process uncertain, cum ulative, and
collective (O 'Sullivan 2000b). The innovation process is uncertain because, by
definition, what needs to be learned about tran sform in g technologies and accessing
markets can only becom e know n through the process itself. By investing in learning,
an innovative strategy confronts the uncertain character o f the innovation process.
The innovation process is cum ulative when learning cannot be done all at once; what
is learned today provides a foundation for what can be learned tom orrow . Invest
ments in cum ulative learning, therefore, require sustained, com m itted finance. The
innovation process is collective when learning cannot be done alone; learning
requires the collaboration o f different people with different capabilities. Investments
in collective learning, therefore, require the integration o f the w ork o f these people
into an organization.
W hat is the theory o f the firm that can com prehend h o w strategizing, financing,
and organizing can support the Innovation process? O ver the past century, the
theoretical efforts o f econom ists have focused m ain ly on the optim izing firm
rather than the innovating firm . The optim izing firm takes as given technological
THE I NN OV AT I VE FIRM 3]
capabilities and m arket prices (for inputs as well as outputs), and seeks to m axim ize
profits on the basis o f these technological and m arket constraints. In sharp con trast,
in the attem pt to generate higher quality, low er cost products than had previously
been available, and thus differentiate itself from com petitors in its industry, the
innovating firm seeks to tran sform the technological and m arket conditions that the
optim izing firm takes as “ given” constraints. Hence, rather than constrained o p ti
m ization, the innovating firm engages in what I call “ historical tra n sfo rm a tio n ” a
m ode o f resource allocation that requires a theoretical perspective on the processes
o f industrial and organizational change (Lazonick 2002a).
The distinction between the innovating and optim izing firm is im plicit in the
w ork o f A lfred M arshall, w hose Principles o f Economics, published in eight editions
between 1890 and 1920, placed the th eo ry o f the firm at the center o f econom ic
analysis. A lth ou gh M arshall's follow ers used his argum ents to construct the theory
o f the optim izing firm that rem ains entrenched in econom ics textbooks, M arshall
(1961:3x5) h im self displayed considerable insight into the dynam ics o f the in n o vat
ing firm , as revealed in the follow in g passage:
An able man, assisted by some strokes o f good fortune, gets a firm footing in the trade, he
works hard and lives sparely, his own capital grows fast, and the credit that enables him to
borrow more capital grows still faster; he collects around him subordinates of more than
ordinary zeal and ability; as his business increases they rise with him, they trust him and he
trusts them, each o f them devotes himself with energy to just that work for which he is
specially fitted, so that no high ability is wasted on easy work, and no difficult work is
entrusted to unskillful hands. Corresponding to this steadily increasing economy o f skill, the
growth o f his firm brings with it similar economies o f specialized machines and plants o f all
kinds; every improved process is quickly adopted and made the basis o f further improve
ments; success brings credit and credit brings success; success and credit help to retain old
customers and to bring new ones; the increase o f his trade gives him great advantages in
buying; his goods advertise one another and thus diminish his difficulty in finding a vent for
them. The increase o f the scale o f his business increases rapidly the advantages which he has
over his competitors, and lowers the price at which he can afford to sell.
W hat then constrains the grow th o f such a firm ? In Industry and Trade>published
in 1919, A lfred M arsh all acknow ledged that over the previou s decades the large-scale
enterprise had becom e d o m in an t in advanced nations such as the U nited States and
Germ any. H e invoked, how ever, the aphorism , “ shirtsleeves to shirtsleeves In three
generations” (M arsh all 1961: 621) to explain the lim it to the grow th o f the firm that
w ould prevent a sm all n u m b er o f large firm s fro m d om inatin g an industry. A n
ow ner-entrepreneur o f exceptional ab ility w o u ld fou nd and build a successful firm .
In the second generation, control w o u ld pass to descendants w h o could not be
expected to have the capabilities o r drive o f the founder, and as a result the firm
would grow m o re slowly o r even stagnate. The third generation w o u ld lose touch
w ith the innovative legacy o f the first generation, and the firm w ou ld w ither aw ay in
the face o f n ew en trepreneurial com petition.
32 WILLIAM LAZONICK
W riting in the first decades of the twentieth century, Jo sep h Schum peter (1934)
also focused on the innovative entrepreneur w ho, b y creating “ n ew combinations”
o f productive resources, could disrupt the “ circu lar flo w o f econom ic life as
conditioned by given circum stances ” In effect, Schu m peter was arguing that,
through entrepreneurship, which he called the fu n dam en tal phenom enon of
econom ic developm ent,” innovating firm s could challenge o p tim izin g firms, and
thereby drive the developm ent o f the econom y. In i9it> w hen he first published
The Theory of Economic Development (in G erm an ), Schum peter, like Marshall,
viewed the innovative firm as the result o f the entrepreneurial w o rk o f an extra
ord in ary individual. Over the subsequent decades, how ever, as Schum peter ob
served the actual developm ent o f the leading econom ies, he cam e to see the large
corporation as the innovating firm , engaged in what he called a process o f creative
destruction” ; the creation o f new m odes o f p rodu ctive tran sfo rm ation destroyed
existing m odes that had themselves been the result o f in n ovative enterprise in
the past.
In Capitalism , Socialism, and Democracy, first published in 1942, Schumpeter
( 19 5 0 :118 ,13 2 ) argued that “ technological 'progress’ tends, through systemization
and rationalization o f research and m anagem ent, to becom e m ore effective and
sure-footed” as it is undertaken as “ the business o f team s o f trained specialists who
turn out what is required and make it w ork in predictable w ays ” In a series o f major
works, Alfred Chandler (1962, 1977, 1990) docum ented the rise o f the managerial
corporation in the United States from the last decades o f the nineteenth century, the
evolution o f its m ultidivisional structure from the 1920s, and the emergence of
m anagerial enterprise in Britain and Germ any. In The Theory o f the Growth of the
Firm , first published in 1959, Edith Penrose (1995) conceptualized the modem
corporate enterprise as an organization that adm inisters a collection o f human
and physical resources. People contribute labor services to the firm , not merely as
individuals, but as m em bers o f teams who engage in learning about how to make
best use o f the firm ’s productive resources— including their ow n.
At any point in time, this learning endow s the firm w ith experience that gives it
productive opportunities unavailable to other firm s, even in the sam e industry, that
have not accum ulated the same experience. The accu m u lation o f innovative experi
ence enables the firm to overcom e the “ m anagerial lim it” that in the theory o f the
optim izing firm causes the onset o f increasing costs and constrains the growth of
the firm (Penrose 19 95- chs. 5,7, and 8). The innovating firm can transfer and reshape
its existing productive resources to take advantage o f new m arket opportunities.
Each m ove into a new product m arket enables the firm to utilize unused productive
services accum ulated through the process o f organizational learning. These unused
productive services can provide a foundation for the grow th o f the firm , through
both in-house com plem entary investments in new product developm ent and the
acquisition o f other firms that have already developed co m p lem en tary productive
services.
THE INNOVATIVE FIRM S3
*E lbaum , B., and L azo nick , W. (eds.) (1986), The Decline of the British Economy>Oxford:
Oxford University Press.
Fa r n ie > D. (1990), uThe Textile Machinery-Making Industry and the World Market,
1870—1960,” Business History 32(4): 150-70.
G a l b r a it h , J. K. (1967), The New Industrial State, Boston: Houghton Mifflin.
G im e in , M., D a s h , E., M unoz , L., and S u n g , J. (2002), “ You Bought They Sold/' Fortune
146(4): 64-72.
G om pers , P., L erner , J., and Sch a r fstein , D. (2003), “ Entrepreneurial Spawning: Public
Corporations and the Genesis o f New Ventures, 1986-1999,” N BER Working Paper 9816,
July.
G ordon , A. (1985), The Evolution of Labor Relations in Japan: Heavy Industry; 1853-1955,
Cambridge, Mass.: Harvard University Press.
* H a n n a h , L. (1983), The Rise of the Corporate Economy: The British Experience, 2nd edn.,
London: Methuen.
H o u n sh ell , D. (1984), From the American System to Mass Production, 1800-1932, Baltimore:
Johns Hopkins University Press.
H u b e r m a n , M. (1996), Escape from the Market: Negotiating Work in Lancashire, Cambridge:
Cambridge University Press.
J a ik u m a r , R. (1989), “ Japanese Flexible Manufacturing Systems: Impact on the United
States,” Japan and the World Economy 1(2): 113-43.
K ogut , B., and Z a n d e r , U. (1996), “ What Firms Do? Coordination, Identity, and Learning,”
Organization Science 7: 502-18.
K o rtu m , S., and L e r n e r , J. (2000), “Assessing the Contribution o f Venture Capital to
Innovation,” Rand Journal of Economics 31(4): 674-92.
La zo n ic k , W. (1986), “ Strategy, Structure, and Management Development in the United
States and Britain,” in K. Kobayashi and Н. Morikawa (eds.), Development of Managerial
Enterprise, Tokyo: University o f Tokyo Press, 101-46.
------ (1990), Competitive Advantage on the Shop Floor, Cambridge, Mass.: Harvard University
Press.
------ (1999), “ The Japanese Economy and Corporate Reform: What Path to Sustainable
Prosperity?” Industrial and Corporate Change8(4): 607-33.
------ (2001), “ Organizational Learning and International Competition: The Skill-Base
Hypothesis,” in W. Lazonick and M. O'Sullivan (eds.), Corporate Governance and Sustain
able Prosperity, Basingstoke: Palgrave, 37- 17*
------ (2002a), “ Innovative Enterprise and Historical Transformation,” Enterprise & Society
3(i): 35- 54-
------ (2002b), “ Innovative Enterprise, The Theory of,” in M. Warner, (ed.), International
Encyclopedia of Business and Management, 2nd edn., Stamford, Conn.: Thomson Learning,
3055-76.
-----(2002c), “The US Industrial Corporation and The Theory of the Growth ofthe Firm,” in
Christos Pitelis (ed.), The Growth of the Firm: The Legacy o f Edith Penrose, Oxford: Oxford
University Press, 249-77.
— —(2003), “The Social Foundations of Innovative Enterprise,” Franco Momigliano Lec
ture, Institute per la Cultura e la Storia dTmpresa, Terni, Italy, July 10.
— — (2004), “ Corporate Restructuring ” in S. Ackroyd, R* Batt, R Thompson, and R Tolbert
(eds.), The Oxford Handbook o f Work and Organization, Oxford: Oxford University Pies
(forthcoming).
34 WILLIAM LAZONICK
along the cum ulative paths that are the fou n d ation s o f its distinctive competitive
success. The perspective em phasizes the role o f h u m an agency in determining
whether and how the enterprise accum ulates in n ovative capability, and thus adds
an explicitly social dim ension to w ork on dynam ic capabilities. Specifically,
strategic control determines h ow strategic decision m akers choose to build on
“ asset position s"; financial com m itm ent determ ines w hether the enterprise will
have the resources available to it to persist along an “ evo lu tio n ary path" to the
point where its accum ulation o f innovative capability can generate financial returns;
and organizational integration determ ines the structure o f incentives that charac
terize “ organizational processes" that can tran sform in d ivid u al actions and individ
ual capabilities (including those o f strategic m anagers) into collective learning.
O f central im portance to the accum ulation and tran sfo rm atio n o f capabilities in
knowledge-intensive industries is the skill base in w hich the firm invests in pursuing
its innovative strategy. W ithin the firm , the d ivisio n o f lab or consists o f different
functional specialties and hierarchical responsibilities. A t any point in time a
firm's functional and hierarchical division o f lab or defines its skill base. In the
effort to generate collective and cum ulative learning, those w h o exercise strategic
control can choose how to structure the skill base, in clu d in g h o w em ployees move
around and up the functional and hierarchical d ivisio n o f lab or over the course of
their careers. At the same tim e, however, the organization o f the skill base will be
constrained by both the particular learning requirem ents o f the industrial activities
in which the firm has chosen to com pete and the alternative em ploym ent oppor
tunities o f the personnel for w hom the firm m ust com pete.
In cross-national com parative perspective, the skill base that enterprises employ
to transform technologies and access m arkets can v a ry m arkedly even in the same
industrial activity during the same historical era, w ith different innovative out
comes. Precisely because innovative enterprise depends on social conditions, the
developm ent and utilization o f skill bases that occu r in one institutional environ
ment m ay not, at a point in time at least, be possible in another institutional
environm ent. M oreover, even within the sam e in d u stry and sam e nation, dynamic
capabilities that yielded innovative outcom es in one historical era m ay becom e static
capabilities that inhibit innovative responses in a subsequent historical era.
I h e innovative firm requires that those w ho exercise strategic control be able
to recognize the com petitive strengths and weaknesses o f their firm 's existing
skill base and, hence, the changes in that skill base that w ill be necessary for an
innovative response to com petitive challenges. These strategic decision makers must
also be able to mobilize com m itted finance to sustain investm ent in the skill base
until it can generate higher quality, lower cost produ cts than were previously
available. As the follow ing com parative-h istorical syntheses illustrate, given stra
tegic control and financial com m itm ent, the essence o f the innovati ve firm is the
organizational integration of a skill base that can engage in collective and cumulative
learning.
T H E I N N O V A T I V E FIRM.' 33
From the 1980s m an y business sch ool academ ics, w o rk in g in the strategy area,
cited Penrose's 1959 b o o k as an intellectual fo u n d atio n for a “ reso u rce-b ased ” view
o f the firm . R esou rce-b ased th eo ry focused on the characteristics o f valuab le re
sources that one firm possessed and that co m p etitor firm s fo u n d it d ifficult to
im itate. R esou rce-b ased theory, how ever, pro vid ed no perspective on w h y and
h o w som e firm s rather than others accu m u lated valuable and in im itab le resources,
o r indeed w h at m ad e these resources valu able and in im itab le (see Lazonick 2002a).
In depen d en tly o f the resou rce-based perspective, how ever, R ich ard N elson and
Sid n ey W in ter (1982) fashion ed a th eo ry o f the persistence o f the large in du strial
co rp o ratio n based o n o rgan izatio n al capabilities, characterized b y tacit know ledge
and em bed ded in o rgan izatio n al routines, thus ad d in g a cu m u lative d im en sio n to
the th e o ry o f the firm . D raw in g on a h igh ly eclectic set o f sources fro m a n u m b er o f
disciplines, B ru ce K o gu t and U d o Z a n d er (1996: 502) argu ed that “ [f firm s are
o rganizations that represent social know ledge o f co o rd in atio n and learning,” thus
em ph asizing the collective d im en sio n in the th eo ry o f the firm .
In “ W h y D o F irm s D iffer, and H o w D oes It M atter?” N elson (19 9 1:7 2 ) argu ed that
“ it is o rgan izatio n al differences, especially differences in abilities to generate and
gain fro m in n o vatio n , rather than differences in co m m an d over p articu lar tech n o lo
gies, that are the sou rce o f durable, n ot easily im itable, differences a m o n g firm s.
Particular tech n ologies are m u ch easier to understand, and im itate, than b ro ad er
firm d yn am ic capabilities.” D a v id Teece, G a ry Pisan o, an d A m y Shuen (1997: 516)
defined “ d yn a m ic capabilities as the fir m s ab ility to integrate, b u ild , and reconfigure
internal and extern al com petences to address rap id ly ch an gin g environ m en ts.” T h e y
also argu ed that the firm 's strategy entails ch oo sin g am o n g and co m m ittin g to lo n g
term path s o r trajectories o f com petence d evelopm en t (Teece at al. 1997: 524).
W hereas the firm 's asset p o sitio n s determ ine its com petitive advantage at an y
p o in t in tim e an d its e vo lu tio n a ry path constrains the types o f in d u strial activities
in w h ich a firm can be co m petitive, its o rgan izatio n al processes tran sfo rm the
capabilities o f the firm o ver tim e.
W h ile Teece et al. (19 9 7 :5 19 ) stressed the im p o rtan ce o f learn in g processes that are
“ in trin sically so cial an d collective,” their d yn am ic capabilities perspective lacks
social content. T h e fra m e w o rk does n o t ask w h at typ es o f peo p le are able an d w ilin g
to m ake the strategic investm ents that can result in in n o vatio n , h o w these strategic
d ecision m akers m o b ilize the necessary fin an cial resources, and h o w they create
incentives fo r th o se people w ith in the firm 's h ierarch ical and fu n ctio n a l d iv isio n o f
labor to co o p erate in the im p lem en tatio n o f the in n o vative strategy. These qu estio n s
ab ou t the roles o f strategizing, fin an cin g, an d o rgan izin g in the in n o va tin g firm are
have called the “ so cial co n d itio n s o f
at the center o f w h at M a r y O 'S u liv a n and I
in n o vative e n terp rise” perspective (L azo n ick and O 'S u lliv a n 2000; O 'S u lliv a n
2000b; L a z o n ick 2002b).
T h is p ersp ective asks h o w an d u n d er w h at co n d itio n s the exercise of strategic
co n tro l ensures- that the en terprise seeks to g ro w u sin g the e o le e tiv e p ro cesses m i
36 W I L L I A M L A Z O N I C K __________ _____________________________________ _
2.3 T h e B r it is h In d u s t r ia l D is t r ic t
2.4 T h e US M a n a g e r ia l C o r p o r a t io n
Marshall located the limits to the growth o f the firm in the problem o f succeeding the
original owner-entrepreneur. In The Theory ofEconomic Development, Schumpeter
(1934: 156) concurred using the sam e aphorism as M arsh all, literally clothed in
different garb and specifically identified as a U S p h en om en on : “A n American
adage expresses it: three generations from overalls to overalls ” Critical to this
perspective were two assum ptions: first, that the entrepreneur w as the essence of
the innovative firm , and second, that the integration o f ow n ersh ip and control was a
necessary condition for entrepreneurship. N otw ith stan d in g his ow n important
study o f com parative trends in industrial organization pu blish ed in Industry and
Trade, M arshall (1919) declined to recognize, as ultim ately Schu m peter did, that the
problem o f innovative succession could be resolved b y the separation o f ownership
and co n tro l
Taking place during the same decades in w hich M arsh all w rote his influential
books, the separation o f share ow nership from strategic control w as the essence of
what Chandler (am ong others) w ould call “ the m anagerial revolu tion ” in American
business. D uring this period G erm any and Jap an also experienced managerial
revolutions (Chandler 1990; Chandler et al. 1997; M o rik aw a and Kobayashi 1986;
M orikaw a 1997). M any British firm s, especially in the science-based chemical and
electrical industries also m ade investm ents in m anagerial organ ization , but in such a
constrained m anner that it can hardly be said that a m anagerial revolution occurred
in Britain during the first h alf o f the twentieth century (H an n ah 1983; Lazonick 1986;
Chandler 1990; Owen 2000).
In the United States, the m anagerial revolution began in the 1890s in industries
such as steel, oil refining, m eatpacking, tobacco, agricultural equipm ent, telecom
m unications, and electric power that ow ner-entrepreneurs had built up over the
previous decades. Wall Street (and especially the firm o f J. P. M o rgan ) organized the
m erger o f the leading com panies, and in the process did w hat w o u ld later become
known as initial public offerings” (IPO s) in order to allow the owner-entrepreneurs
to cash in on their ownership stakes. M an y o f them then retired from active
m anagem ent o f the company. Taking their places in strategic decision-making
positions were salaried m anagers, m ost o f w h o m had them selves been recruited
years or even decades earlier to help build the innovative firm s that they now
controlled. Hence, M arshall’s “ entrepreneurial” lim it to the grow th o f the firm was
overcom e. By the turn o f the century, the separation o f ow nership and control in
m any o f the m ost successful industrial corporations served as a pow erful induce
ment for bright young, and typically W hite, A n g lo -Saxo n , Protestant, men to
consider careers as corporate executives (Lazonick 1986; O ’ Su llivan 2000a: ch. 3).
Also from the beginning o f the twentieth century, а fo u r-year undergraduate
college degree became im portant for entry into m anagerial careers, and in 1908
THE INNOVATIVE FIRM 3J
were u n im p o rtan t to econ om ic p erform an ce, and indeed that they could all be
characterized b y d ep ictin g a “ representative firm ” that optim ized subject to given
tech nological and m arket constraints. W ithin the M arsh allian perspective, even
in n o vation at the district level did not require strategic direction , since the industrial
arts w ere “ in the a ir ” In deed, M arsh all (1919: 6 0 0 -1) described the o rgan ization o f
the Lan cash ire cotton textile industry, w ith its h igh degrees o f h orizon tal co m p e ti
tion and vertical specialization, as “ perhaps the present instance o f concentrated
o rgan isatio n m a in ly a u to m a tic ” Yet ju st as M arsh all w as w ritin g these w ords, the
cotton textile indu stry, w h ich had accoun ted fo r o n e-q u arter o f British exp orts on
the eve o f W orld W ar I, entered into a lo n g -ru n decline fro m w hich it never
recovered, and the other m a jo r B ritish in d u strial districts suffered a sim ilar fate
(E lb au m an d Lazo n ick 1986).
From the late 1970s, how ever, the n o tio n o f the “ M arsh allian in d u strial d istrict” as
a d river o f in n o vative enterprise saw an academ ic resurgence, based on the rap id
grow th d u rin g the 1960s and 1970s o f m an y h igh ly specialized and localized districts
in w hat b ecam e k n o w n as “ the T h ird Italy” (B ru sco 1982; Sabel 1982; Becattini 1990).
O n the basis o f this experience, a n u m b er o f U S academ ics, h ead ed b y C h arles Sabel,
M ich ael P iore, an d Jo n ath an Z eitlin , posited a n ew m o d el o f “ flexible sp ecialization ”
as an alternative to m ass p ro d u ctio n on the U S co rp orate m o d el (P iore and
Sabel 1984; Sabel an d Z eitlin 1985). T h e in d u strial activities o f the districts o f the
T h ird Italy fo cu sed on, am o n g oth er things, textiles, footw ear, and light m ach in ery,
just as the B ritish districts had done. Large n u m bers o f vertically specialized p ro p ri
etary firm s in w h ich craft lab o r w as a p rim e source o f co m p etitive advantage
p o p u lated each in d u strial activity, and m an y entrepreneurs h ad p re vio u sly been
craft w orkers.
There w ere, how ever, tw o im p o rtan t differences betw een the B ritish in d u strial
districts that M arsh all h ad observed in the late nineteenth cen tu ry and those that
experien ced rap id grow th in the T h ird Italy m ore recently. T h e first difference w as
the extent to w h ich in Italy collective in stitu tion s su p p o rted the in n o vative activities
o f sm all firm s. Seb astian o B ru sco (1992) has em phasized the im p o rtan ce o f the “ red ”
local go vern m en ts in E m ilia -R o m a g n a in p ro m o tin g p o licies to su p p o rt the activ
ities o f sm all enterprises, an d in p articu lar in facilitatin g coop eratives that p ro vid e d
these firm s w ith “ real services” related to business ad m in istratio n , m arketin g, and
train in g. W h ile co n su m er co op eratives sp ru n g u p in the B ritish in d u strial d istricts
o f the late nineteenth century, p ro d u cer co op eratives w ere rare. T h e secon d d iffe r
ence, w h ich b ecam e m o re evident in the 1990s, w as the extent to w h ich , in some
districts an d in so m e in du stries, “ lead in g” firm s co u ld em erge, d raw in g o n the
resources o f the in d u strial d istricts w h ile, th ro u g h th eir o w n in tern al growth*
tra n sfo rm in g the in n o vative ca p a b ility o f the d istricts (see, for exam p le, lelussi
1999). In co n trast, w h en in the first h a lf o f the tw entieth ce n tu ry competitive
challenges co n fro n ted the B ritish in d u strial districts, d o m in a n t firms failed to
em erge t o lead a restru ctu rin g process.
40 W IL L IA M LA Z O N JCK _________ ______ ____________________ _
intervention o f a skilled worker to m ake the parts fit together. A s D avid Homshell
(1984) has shown, it took a century o f investm ent in p ro d u ctive capabilities by many
com panies in m any sectors o f U S indu stry before, d u rin g the b o o m o f the 1920s,
mass production, so defined, becam e a reality. T h e p ro d u ctivity o f the mass-
production enterprise, nevertheless, still relied u p on the stable employment of
“ sem i-skilled” production workers w ho tended h igh -th rou gh p u t, and very expen
sive, m achinery (Lazonick 1990: chs. 7-8 ).
D uring the Great Depression o f the 1930s, such stable em ploym en t disappeared,
leading sem i-skilled workers at the m ajor m ass p ro d u cers to turn to industrial
unionism (B rod y 1980: ch. 3). The m ajor achievem ent o f m ass-p ro d u ction unionism
in the United States was long-term em ploym ent secu rity fo r so-called “hourly*
workers, with seniority as the governing principle fo r internal p ro m o tio n to higher
pay grades and continued em ploym ent during co m p an y layoffs. In return, these
unionized employees accepted unilateral m anagerial con trol over the organization
o f w ork and technological change. D uring the post-W o rid W ar II decades, produc
tion workers enjoyed em ploym ent security and rising w ages but they were not in
general integrated with m anagerial personnel into the co m p an y's organizational
learning processes.
The result was that going into the second h a lf o f the tw entieth century US
industrial corporations had pow erful m anagerial organ ization s for developing
new technology. These corporations also had devised arrangem ents with their
unionized labor forces to ensure the high level o f utilization o f these technologies.
In em ploying thousands and in som e cases tens o f thou sand s o f production workers
who were not integrated into the com pany's organizational learning processes,
however, this US m odel o f the innovative firm had a fu ndam ental weakness that,
in the 1970s and 1980s, w ould be exposed in international com petition . The Japanese
in particular w ould dem onstrate the innovative capab ility that could be created by
not only building highly integrated m anagerial organizations, as the Am ericans had
done, but also, as a com plem ent, developing the skills o f sh o p -flo o r workers and
integrating their efforts into the firm's collective learning processes.
Even the m ost insightful o f the theories o f the U S m anagerial corporation could
not, without elaboration, account for the Japanese challenge (Lazon ick 2002c). Both
Penrose (1995) and Chandler (1962 and 1977) focused exclusively on the managerial
organization, as did the influential perspective o f Jo h n Kenneth Galbraith (1967)
with its notion o f the technostructure" as the essence o f the m o d ern firm . Penrose
did not see that, once confronted by the Japanese challenge, the U S managerial
corporation would have to develop the capabilities o f the sh o p -flo o r w orker to make
use o f unused m anagerial resources. Chandler focused on speed or throughput as a
basis for achieving econom ies o f scale and scope, but ignored the role o f the shop-
floor worker in the process o f transform ing high fixed costs into low unit costs, and
hence did not perceive an im portant lim itation o f the U S m anagerial model
(Lazonick 1990).
THE INNOVATIVE FIRM 39
1997), Taking control o f strategic decision m aking w ere "th ird -ra n k executives,”
prim arily engineers plucked from the ranks o f m iddle m anagem ent to take leader
ship positions o f com panies whose challenge was to find n o n -m ilita ry markets for
their com panies' accum ulated capabilities.
W ith the reopening o f the stock m arket in 1949? these yo u n g and ambitious
executives feared that the new public shareholders m igh t jo in forces to demand
their traditional rights as owners. To defend them selves against these outside
interests, the com m unity o f corporate executives engaged in the practice o f cross
shareholding. Com m ercial banks and industrial com panies to o k equities off the
market by holding each other's shares. T h ou gh not contractual, cross-shareholding
was sustained by the willingness o f the entire Japanese business co m m u n ity to accept
that one com pany would not sell its shareholdings o f anoth er com pany.2 By 1975,
according to its broadest, and m ost relevant, defin ition as stock in the hands o f such
stable shareholders, cross-shareholding represented 60 per cent o f outstanding
stocks listed on the Tokyo Stock Exchange. It peaked at 67.4 p er cent in 1988, but
by 2000 had declined to 57.1 per cent, m ainly because the beleaguered banking sector
had been forced to reduce their shareholdings.
D uring the "era o f high-speed grow th" from the early 1950s to the early 1970s,
m ost o f the financial com m itm ent o f Japanese com panies cam e from bank loans,
with the com panies' d eb t-eq u ity ratios often at 6 : 1 or 7 : 1 . Each m ajor industrial
com pany had a "m ain ban k" whose job it was to convince other banks to join it in
m aking loans to the com pany and to take the lead in restru ctu ring its client company
should it fall into financial distress. Som e econom ists (e.g. A o k i and Patrick 1994)
have accorded the m ain banks a m ajor role in m o n ito rin g the behavior o f Japans
corporate managers. In funding the grow th o f Japanese com panies, however, the
Japanese banks were relatively passive agents o f govern m en t developm ent policy,
with "overloans" being m ade by the Bank o f Jap an to its m em b er banks for provid
ing highly leveraged finance to grow ing industrial com panies. Japanese banks, that
is, played a critical role in providing financial com m itm ent, but no significant role in
the exercise o f strategic control.
Integrated organizations o f m anagers and w orkers, not financial interests, moni
tored the behavior o f the top executives o f Japanese co rp oratio n s (Lazonick 1999)’
The m ain m ode o f achieving this organizational integration was the lifetime em
ploym ent system, which extended from top executives to m ale (but not female)
shop-floor workers. The origins o f the lifetim e em ploym ent system can be found in
the widespread em ploym ent in industry ot university graduates as salaried technical
and adm inistrative personnel during the early twentieth cen tu ry (Yonekawa 1984)*
Som e com panies extended the prom ise ot lifetim e em ploym ent to shop-floor
workers as well when dire econom ic conditions and dem ocratization initiatives of
the late 1940s had given rise to a m ilitant labor m ovem ent. The goal o f the new
industrial unions was to im plem ent "p ro d u ctio n co n tro l": the takeover o f idle
factories so that workers could put them into operation and earn a living (Gordon
THE INNOVATIVE FIRM 41
2.5 T h e J a p a n e se C h a l l e n g e
Within the new structure of cooperative industrial relations that emerged out the
conflicts of the depression years, US industrial corporations were able to take
advantage of the post-World War II boom to re-establish themselves as the world’s
pre-eminent producers of consumer durables such as automobiles and electrical
appliances and related capital goods such as steel and machine tools* With the help of
U S go vern m en t research su p p o rt and contracts, U S com panies also becam e the
leaders in the computer and semiconductor industries.
In the 1970s and 1980s, however, Japanese companies challenged the US industrial
corporations in the very mass-production industries— steel, memory chips,
machine tools, electrical machinery, consumer electronics, and automobiles— in
which even as late as the 1960s US corporations seemed to have attained an insur
mountable competitive advantage. During the 1950s and 1960s many Japanese
companies had developed innovative manufacturing capabilities, often on the
basis o f tech nologies b o rro w ed fro m abroad to p ro d u ce m ain ly for the h om e
market. As Japanese exports to the United States increased rapidly in the last half
of the 1970s, many observers attributed the challenge to the lower wages and longer
working hours that prevailed in Japan. By the early 1980s, however, with real wages in
Japan continuing to rise, it became clear that Japanese advantage was based on
superior capabilities for generating higher quality, lower cost products*
The three social institutions that, in combination, formed the foundation for
Japan’s remarkable success were cross-shareholding, the main b an k system, and
lifetime employment. Cross-shareholding provided the managers of Japanese
industrial corporations with the strategic control to allocate resources to invest
ments that could generate higher quality, lower cost products. The main bank system
provided these companies with levels of fin an cial commitment that permitted them
to sustain the innovation processes until they could generate returns, first on home
and then on foreign product markets. Given this financial support for strategic
industries, lifetime employment enabled the companies involved to put in place a
new model of hierarchical and functional integration that enabled them to
mobilize broader and deeper skills bases for collective and cumulative learning
(Lazonick 2001). Let us look briefly at how these institutions became embedded in
the functioning of the Japanese industrial enterprise in the post-World War II
decades.
In 1948 the Supreme Commander for the Allied Powers (SCAP)—the occupation
authority in Japan—began the dissolution of the zaibatsu, the giant holding com
panies that had dominated the Japanese economy from the Meiji era of the late
nineteenth century to World War II. The dissolution process not only dispossessed
the families that owned the zaibatsu but also removed from office the top im rnge-
meut layers of the zaibatsu holding companies and major affiliated firms (Могікаші
44 WILLIAM LAZONICK
developm ent cycles. M uch o f the discussion o f fu n ctio n al integration focused un its
role in “ new product developm ent" in international co m p arative perspective, with,
as Clark and Fujim oto (1991) showed for the au tom o b ile industry, the U S manager
ial corporation perform ing quite poorly.
Given that the innovative pow er o f the U S in d u strial co rp o ratio n resided in its
integrated m anagerial organization, w hy should it have suffered from functional
segm entation in com petition w ith the Japanese? O ne reason w as that, given the
hierarchical segm entation o f sh op -floor activities fro m organizational learning
processes in US com panies, U S engineers were not forced to com m unicate across
their disciplines to solve “ real-w orld" m anu factu ring problem s. A n oth er had to do
with the increasing interfirm m obility o f U S engineers fro m the 1960s— mobility
that, as we shall see, was related to the rise o f the "N e w E c o n o m y " high-tech firm.
The prospects for interfirm m o b ility gave scientists an d engineers an interest in
developing their reputations am ong their peers w ith in their particular area of
specialization, even if it detracted from integrating their specialist knowledge across
functional areas within the particular firm for w hich they w ere w orking. By contrast,
in the Japanese firm both the hierarchical integration o f m anagers and workers and
low levels o f interfirm m obility o f engineering person nel fostered functional
integration.
The evolution o f the sem iconductor in d u stry pro vid es a v iv id exam ple o f the
com petitive power, but also the lim its, o f Japanese o rgan ization al integration. From
the late 1970s the Japanese m ounted a form idable com petitive challenge to US
producers in dynam ic random access m em o ry (D R A M ) chips, forcing most US
com panies, including Intel, to w ithdraw from the m arket after 1985. Already a
powerhouse in sem iconductors before the Japanese challenge, Intel reemerged
even stronger in the 1990s as the leader in m icroprocessors, a product in which it
was the pioneer in the early 1970s and for w hich d u rin g the 1980s it secured the
franchise for the IB M PC and the subsequent IB M clones (B u rgelm an 1994).
O rganizational integration was critical to the Japanese challenge in DRAM s. As
Daniel O kim oto and Yoshio Nishi (1994) have show n, the m ost critical interactions
in product and process developm ent in Japanese sem icon d u ctor com panies were
between personnel in divisional R8cD labs and facto ry engineering labs, with
engineering capability being concentrated in the fa cto ry labs. T h ey argue that in
Japan hands-on m anufacturing exp erien ce. . . is alm ost a requirem ent for upward
career and post-career m obility [whereas] [i]n the U nited States, by contrast,
m anufacturing engineers carry the stigm a o f being second-class citizens" (Okimoto
and Nishi 1994:195).
Value added in m icroprocessors is in the design that determ ines the use of the
product, an activity for which US skill bases in sem icon d u ctors were m ore suited.
Value added in m em ory chips is in process engineering that reduces defects and
increases chip yields, an activity for which Japanese skill bases in semiconductors
were m ore suited. By the 1980s Japanese com panies such as Fujitsu, Hitachi, and
THE INNOVATIVE FIRM 43
1985). Lead in g com pan ies such as Toyota, Toshiba, and Hitachi fired militant
w orkers and created enterprise u nions o f w h ite-collar (technical and adm inistra
tive) and b lu e-collar em ployees. Forem en and su p erviso rs were members o f the
enterprise un io n s, as w ere all u niversity-edu cated personnel, fo r at least the first ten
years o f em p lo ym en t b efore they m ade the official tran sition into “ m anagem ent.”
The m ost im p o rtan t achievem ent o f enterprise u n io n ism w as the in stitu tio n al
ization o f lifetim e em p lo ym en t, a system that, w hile not con tractu ally guaranteed,
gave w h ite-co llar and b lu e-collar w ork ers em p lo ym en t security, at first to the
retirem ent age o f 55, then fro m the 1980s to the age o f 60, and cu rren tly (in
tran sition ) to the age o f 65 (Sako and Sato 1997). T h is em p lo ym en t secu rity both
w on the co m m itm en t o f the w o rk er to the co m p an y and gave the co m p an y the
incentive to d evelop the p ro d u ctive capabilities o f the w orker. The system d id not
differ in p rin cip le fro m the organ izatio n al integration o f technical and a d m in is
trative em ployees that w as at the heart o f the U S m an agerial revolu tion , except in one
extrem ely im p o rta n t respect. In the U nited States there w as a sharp segm entation
betw een salaried m an agers and sh o p -flo o r w orkers, w hereas the Japan ese c o m
panies o f the p o st-W orld W ar II decades integrated sh o p -flo o r w ork ers into a
co m p an y-w id e process o f organ ization al learning.
T h ro u g h their engagem ent in processes o f cost red u ction , Japanese sh o p -flo o r
w orkers w ere co n tin u o u sly involved in a m o re general process o f im p ro vem en t of
products and processes that, b y the 1970s, enabled Japanese com pan ies to em erge as
w orld leaders in fa c to ry a u to m atio n (Ja ik u m a r 1989). B y the early 1990s the stock o f
robots in Jap an ese factories w as o ver seven tim es that o f the U n ited States. Also o f
great im p o rtan ce w as the ab ility o f Japan ese m anu factu rers to elim inate w aste in
p ro d u ctio n ; b y the late 1970s, fo r exam ple, Japan 's com petitive advantage in televi
sion sets w as n ot in lab o r costs o r even scale econ om ies but in a savin gs o f m aterials
costs (O w en 2000: 278; Fagerberg and G o d in h o in this vo lu m e). T h is p ro d u ctive
tran sfo rm atio n b ecam e p articu la rly im p o rtan t in in tern atio n al co m p etitio n in the
1980s as Jap an ese w ages ap p ro ach ed the levels o f those in N o rth A m erica and
W estern E u ro p e and, especially fro m 1985, as the valu e o f the Jap an ese yen d ra m a t
ically strengthened. D u rin g the 1980s an d 1990s, in flu en ced b y n o t only Japan s
exp ort p e rfo rm an ce b u t also the im p act o f Jap an ese direct investm ent in N o rth
A m erica an d W estern E u ro p e, m a n y W estern com pan ies sou gh t, w ith varying
degrees o f success, to im p lem en t Japan ese h igh -q u ality, lo w -co st m ass-p ro d u ctio n
m ethods.
D u rin g the 1980s m o st W estern analyses o f the sources o f Japanese competitive
advantage fo cu sed o n the h ierarch ical in tegration o f the sh o p -flo o r w o rk er into the
o rgan izatio n al learn in g process. B y the early 1990s, how ever, as Japanese companies
captured h ig h er valu e -a d d e d segm ents o f the p ro d u cts m arkets in which they
co m peted , the em p h asis shifted to the role o f “ cro ss-fu n ctio n al management,”
“ c o m p a n y -w id e q u a lity co n tro l ” o r “ co n cu rren t en gin eerin g” in generating not
o n ly lo w er cost b u t also h igh er q u a lity p ro d u cts within highly accelerated product
4б W I L L I A M L A Z O N I C K _________ ________ _____ __________________________
2.6 T h e N ew E c o n o m y M o d e l
During the 1970s and 1980s while Japanese enterprises were challenging established
U S m anagerial corporations in m any industries in which they had been dom inant,
there was a resurgence o f the U S inform ation and com m unications technology
(IC T ) industries, providing the foundation for what by the last h alf o f the 1990s
became known as the “ N ew Econom y.” Historically, underlying the emergence o f the
N ew Econom y were m assive post-W orld W ar II investments by the U S governm ent,
in collaboration with research universities and industrial corporations, in develop
ing com puter and com m unications technologies.
B y the end o f the 1950s, this com bined business-governm ent investment effort
had resulted in not only the first generation o f com puters, w ith IBM as the leading
firm , b u t also the capability o f im bedding integrated electronic circuits on a silicon
48 WILLIAM LAZONICK
com pany taking on any long-term debt. N evertheless, w ith the burstin g o f the New
Econom y bubble from m id-2000, C isco spent b illion s o f dollars repurchasing its
ow n stock to support its sagging stock price (C arpen ter et al. 2003). Even during the
boom , when stock prices were rising, the extent to w h ich N ew E co n o m y companies
issued stock to make acquisitions and com pensate em ployees m eant that some of
them spent billions o f dollars on stock repurchases; d u rin g 19 9 7-20 0 0 , for example,
InteEs stock repurchases totalled $18.8 billion and M ic ro so ft’s $13.4 billion. By way of
com parison, over these years Intel s total expenditures on R8cD were $14.2 billion,
while M icrosoft’s were $11.2 billion.
As in the cases o f Intel, M icrosoft, and Cisco, b y the end o f the twentieth century a
num ber o f N ew Econom y com panies had grow n to be fo rm id ab le grow ing concerns
(Lazonick 2004). In 2002 the top 500 U S-based com panies b y sales included twenty
IC T firms founded no earlier than 1965 that had been neith er sp u n -o ff from nor
merged with an Old Econom y firm . These tw enty com pan ies h ad revenues ranging
from $35.4 billion for Dell C om puter to $3.0 b illio n fo r C o m p u ter Associates
International, with an average o f $10 .4 b illion. T h eir h eadcounts ranged from
78,700 for Intel to 8,100 for Q ualcom m , w ith an average o f 30,084, up from an
average for the same twenty com panies o f 6,347 in *993* N in e o f these twenty
com panies (and seven o f the top ten) were based in Silicon Valley, another two in
Southern California, and the other nine in eight states aro u n d the country. Compaq
Com puter, the forty-sixth largest U S com pany in 2001 w ith $33.6 billion in sales and
70,950 employees, would have been high up on this list in 2002 had it not been
acquired by Hewlett-Packard.
M any o f these large New Econom y com panies have becom e im portan t contribu
tors to the patenting activity o f U S-based corp oration s. Sam u el K ortum and Josh
Lerner (2000) have shown that in the first h a lf o f the 1980s a sharp decline in
patenting by U S corporations was counterbalanced b y a m assive increase in early-
stage venture-capital disbursements. But from the last h a lf o f the 1980s patenting
picked up again, in part because it becam e im p o rtan t to the com petitive strategy of
high-growth N ew Econom y firms. In 2001 Intel was eighteenth in the num ber o f US
patents issued to all com panies, and seventh am o n g U S-b ased com panies. Ahead of
Intel were not only Old Econom y com panies such as IB M , Lucent Technologies,
General Electric, and H ewlett-Packard but also tw o m u ch sm aller, but still sizeable,
N ew Econom y sem iconductor com panies, M icro n Technology, founded in 1978
in Idaho, in fourth place, and Advanced M icro D evices (A M D ), founded in
Silicon Valley in 1969, in fourteenth place. In 2002 A M D w as the 535th largest US
com pany by sales and had 12,146 em ployees, w hile M icro n w as 554th and employed
18,700.
Innovative New Econom y com panies have tended to grow large b y upgrading and
expanding their product offerings within their m ain lines o f business, and thus far at
least have not engaged in the indiscrim inate diversification into unrelated technolo
gies and markets that characterized, and ultim ately und erm in ed the perform ance of,
THE INNOVATIVE FIRM 47
2.7 U n d e r st a n d in g t h e In n o v a t iv e
F ir m : Im p l ic a t io n s f o r T h eo r y
This chapter has illustrated that the social characteristics o f the innovative firm have
varied m arkedly over time and across institutional environ m en ts. To study the
innovative firm in abstraction from the particu lar social co n d itio n s that enable it
to generate higher quality, lower cost products is to forgo an understanding o f why it
became innovative in the first place and h ow its in n o vative capabilities may be
rendered obsolete. A com parative-h istorical analysis enables us to learn from the
past and provide w orking hypotheses for on goin g research.
First, the com parative-historical experience o f the in n o vative firm suggests that,
contrary to a com m on belief that has persisted since the tim e o f M arshall, the form of
firm ownership is not the critical issue for u nd erstand in g the type o f strategic control
that supports innovative enterprise. Critical are the abilities and incentives o f those
managers who exercise strategic control. W hether they are m a jo rity owners of the
firm, state employees, or employees o f pu blicly listed com pan ies, we need to know
where and how these strategic m anagers gained the experience to allocate resources
to the innovation process, and the conditions under w h ich their personal rewards
depend on the firm's innovative success.
Second, the most fundam ental, if b y no m eans the only, source o f financial
com m itm ent for the innovative firm is to be fo u n d in those fu nds that are generated
by the firm itself. W hen bank finance is used to leverage financial commitment, it
requires close relations between financial institutions and in n ovative firms, as for
example in the Japanese model. In certain tim es and places, the stock market can
provide some well-positioned firm s with financial co m m itm en t. But as a financial
institution, the fundam ental role o f the stock m arket is to p ro vid e liquidity, not
com m itm ent. It enables ow ner-entrepreneurs and venture capitalists to cash out of
their investments, and it enables households to d iversify their savings portfolios so
that they can (hopefully) tap into the yields o f the stock m arket w ithout having to
devote time and effort to understanding the innovative capabilities o f the companies
that have listed their securities on it.
Third, while strategic control and financial co m m itm en t are essential to the
innovative firm, it is organizational integration that determ ines the innovative
capability that the firm actually possesses. The types o f organizational integration
that result in innovation vary across industries and institu tion al environments as
well as over time. The hierarchical and fu nctional d ivisio n s o f lab or that, when
integrated into learning organizations, generated in n o vatio n in the past cannot
necessarily be expected to do so in the future when faced w ith changes in technology,
markets, and com petition changes which to som e extent successful innovation in
itself brings about,
THE IN N O V A TIV E FIRM 49
m an y lead in g O ld E co n o m y com panies in the 1960s and 1970s* At the same time,
N ew E co n o m y com pan ies have becom e less vertically integrated than Old Economy
com pan ies because equ ip m en t m anu factu rers such as C isco, Dell, and Sun M ic ro
system s have focused their investm ent strategies on activities that require o rgan iza
tional learn in g in their core com petencies, w hile o u tso u rcin g activities that, as is the
case w ith sem ico n d u cto r fab rication , are too expen sive and co m plex to be don e in-
house, or, alternatively, as is the case w ith prin ted circu it board assem bly, have
becom e routine. Som e o f the largest IC T com panies in the U nited States are
upstream electronics co m p o n en ts suppliers, m ost o f w hich are N ew Economy
firm s. A m o n g the top 10 0 0 U S com panies b y sales in 2002 w ere eleven sem ico n
d uctor com pan ies, w ith a total em p lo ym en t o f 212,354, ran gin g from Intel w ith its
78,700 em ployees to N v id ia (a specialist p ro d u cer o f grap h ics processors fo u n d ed
in 1993) w ith 1,513 em ployees. T h e w o r ld s five largest contract m an u factu rers—
Flextronics, Solectron , S an m in a -S C I, C elestica, and Jab il C ircu it— to w h o m e q u ip
m ent m an u factu rers o u tsou rce the m ass p ro d u ctio n o f printed circu it b oard s
and oth er co m p o n en ts, em ployed a total o f 2 6 0 ,0 0 0 -2 70 ,0 0 0 people at the b egin
n ing o f 2003.
T h e severe d o w n tu rn in the IC T in du stries in 20 0 1 and 2002 raised question s
ab ou t the su stain ab ility o f the N ew E c o n o m y m o d e l A m a jo r w eakness o f the N ew
E co n o m y m o d el lay in the huge person al gains, often am o u n tin g to tens o f m illion s
and even h u n d red s o f m illio n s o f dollars, that top executives co u ld reap fro m stock-
based rew ards in a volatile stock m arket. W h en stock prices w ere risin g, executives
h ad stron g p erson al incentives to allocate resources (o r give the app earan ce o f d o in g
so) in w ays that en cou raged the speculative m arket. M a n y o f these allocative d eci
sions u n d erm in ed the in n o vative capabilities o f the firm s o ver w h ich these execu
tives exercised co n tro l (C arp en ter et al. 2003). W h en stock prices b egan falling, the
sam e executives h ad stro n g person al incentives to cash in q u ick ly b y selling stock, so
that th ey m ad e im m en se fo rtu n es (in m o st instances w ith o u t b reakin g the law ) even
as their co m p an ies lost m o n ey and, in m an y cases, strag g led to su rvive (G im em et a l
2002).
A m a jo r p ro b lem fo r som e o f these co m p an ies w as the w a y in w h ich the use o f
stock as a co m b in atio n an d co m p en sation cu rren cy in the N ew E c o n o m y b o o m
affected the role o f the sto ck m arket as a source o f cash (O ’Su llivan 2003). Seven ty
years earlier, in the sto ck m arket b o o m o f the late 1920s, U S co rp o ratio n s had sold
stock at sp eculative prices to p a y d o w n debt o r b olster th eir treasuries, th u s making
th em less fin an cially vu ln erab le w h en the b o o m tu rn ed to bust. In the boom o f the
late 1990s co rp o ra tio n s d id n o t take advan tage o f the sp ecu lative market by selling
stock; i f an yth in g, these co m p an ies pu rch ased sto ck to su p p o rt th eir already inflated
sto ck prices* W h ile em ployees, an d p a rticu la rly h igh -level executives, benefited from
these sto ck p rice .increases, their co m p an ies w ere weakened financially as became
p a in fu lly evid en t fo r m a n y IC T co m p an ies fro m m id -a o o o when the sto ck maxkgt
tu rn ed d o w n .
52 W ILLIAM LAZONICK
i, For a com parison of managerial corporations of European nations such as Britain, France,
Germany, and Italy with the US model, see Lazonick 2003.
2** When in financial distress, a com pany m ight raise cash b y selling som e o f its cross
shareholdings to other com panies at the going m arket price but w ith an understanding
that the shares would be repurchased, also at the going m arket price, if and when its
financial condition improved.
R eferen ces
Sa x e n ia n , A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and Route
128 . C am b rid ge, M ass.: H arvard U n iversity Press.
Sc h u m p e t e r , J. (1934), The Theory of Economic Developmenty C am bridge, M ass.: H arvard
U niversity Press.
------ (1950), Capitalismy Socialism, and Democracy; 3rd edn., N ew York: Harper.
T e e c e , D., P isa n o , G ., and S h u e n , A . (1997), “ D yn am ic Capabilities and Strategic M an age
ment,” Strategic Management Journal 18(7): 509-33.
Y o n ekaw a , S. (1984), “ U n iversity G rad uates in Japanese Enterprises before the Second
W orld War,” Business History 26(3): 19 3-218 .
CHAPTER 3
N E T W O R K S OF
INNOVATORS
W A L T E R W. P O W E L L
STINE GRODAL
3.1 In t r o d u c t io n
I n February o f 2001, two rival consortia published rou gh d raft (rou gh ly 90 per cent
complete) sequences o f the hum an genom e in Nature and Science. The puDiic
H um an Genom e Project consisted o f five key institutions and eleven collaborators,1
supported by the U S N ational Institutes o f H ealth, D epartm en t o f Energy, and the
W ellcome Trust in the United Kingdom . The rival "p riv a te ” consortia, led by the
biotech firm Celera, included both com m ercial firm s and academ ic researchers from
the U niversity o f C alifornia, Penn State, Case W estern, J ohns H op kin s, Cal Tech, Yale,
Rockefeller, as well as scientists in Spain, Israel, and A ustralia. These projects have
been acclaimed for their rem arkable scientific achievem ent; they were also the
product o f considerable organizational innovation. In contrast to the Manhattan
Project or Project Apollo, both o f w hich were h ierarch ically organized, national
projects, the H um an Genom e Project (H G P ) and the C elera team were pluralist,
m ultiorganizational, m ultinational confederations. These tw o groups were intensely
rivalrous, but collaborated intensively w ithin their ow n grou p s (Lam bright 2002).
H G P involved m anagem ent by two governm ent agencies and a private British
We are grateful to David Mowery and Jan Fagerberg for their careful readings o f earlier drafts.
N E T W O R K S OF I N N O V A T O R S 57
Similarly, there is now ample research illustrating the growing links between US
firms and universities (Powell and Owen-Smith 1998), and greater involvement by
firms and government labs in research joint ventures (Link 1 9 9 6 , 1999). In the realm
of science, Hicks and Katz (1996) find that research papers are much more likely to
be co-authored and involve authors with multiple institutional affiliations that
span universities, government, and industry. Distributed networks of practice
are the organizing basis for many technical communities, suggesting both that
sources of knowledge are now more widely dispersed and that governance mechan
isms are emerging to orchestrate distributed knowledge. The open source software
movement is but one highly visible example of this trend (O’Mahony 20 02; Weber
20 0 3), which illustrates how advances in information technology have greatly
facilitated virtual networks, In short, as Mowery (19 9 9 9 ) observes, “the diversity
-
3.2 W h y h a v e N e t w o r k s G r o w n
in Im p o r t a n c e ?
The advantages o f a h eterogen eou s grou p o f contacts are well established in both
social th eo ry an d n etw ork analysis. A stron g trad ition o f th eo ry and research,
running fro m Sim m el (1954) to M erto n (1957) to G ran ovetter (1973) to B u rt
(1992), m akes ab u n d an tly clear that there are in fo rm atio n al, status, and resource
advantages to h avin g b ro ad and d iverse social circles. B elow w e review an array o f
recent em p irical studies that d em on strate h o w interorganization al relationships
lead to v ario u s benefits w ith respect to in fo rm atio n d iffu sio n , resource sharing,
access to specialized assets, and in terorgan ization al learning. In science and tech
n ology-based fields, the advantages that accrue fro m diverse sources o f in fo rm atio n
and resources are considerable. N o t surprisingly, then, as the co m m ercialization o f
know ledge has assu m ed greater im p o rtan ce in econ o m ic grow th, co llab o ration
across o rgan izatio n al b o u n d aries has b ecom e m o re com m on place. In terorgan iza
tional netw orks are a m eans b y w h ich organ ization s can p o o l or exchange resources,
and jo in tly d evelop n ew ideas and skills. In fields w here scientific o r tech nological
progress is d evelo p in g rapidly, and the sources o f know ledge are w id ely distributed,
no single firm has all the necessary skills to stay on top o f all areas o f progress and
bring sign ifican t in n o va tio n s to m arket (Pow ell and B ran tley 1992; Powell, K oput,
and S m ith -D o e rr 1996; H ag e d o o rn and D uysters 2002). In such settings, netw orks
can becom e the lo cu s o f in n o vatio n , as the creation o f know ledge is crucial to
im p ro vin g co m p etitive p o sitio n .
C o llab o rative n etw ork s have lo n g been central to the p ro d u ctio n process in craft-
based indu stries (Eccles 19 8 1), in in d u strial districts (B ru sco 1982; P iore and Sabel
1984), and in fields such as aerospace w h ere assem bly depended u p o n key inputs
from d iverse p articip an ts. T h e grow th o f kn o w led ge-in ten sive indu stries has
heightened the im p o rta n ce o f netw orks in R & D as w ell as p ro d u ct d evelopm ent
and d istrib u tion . A p ersisten t fin d in g fro m a diverse set o f em p irical studies is that
internal R & D in ten sity an d tech n o logical so p h isticatio n are p o sitively correlated
w ith both the n u m b e r an d in ten sity o f strategic alliances (Freem an 19 9 1; H aged o o rn
1995).
Fo r o rgan izatio n s in ra p id ly d evelo p in g fields, h eterogen eity in the p o rtfo lio o f
collaborators allow s firm s to learn fro m a w id e stock o f know ledge. O rgan ization s
w ith b ro ad er n etw o rk s are exp osed to m o re experiences, d ifferent com petencies,
and add ed o p p o rtu n itie s (B eck m an and H au n sch ild 2002). Su ch access creates an
en viron m en t in w h ic h “ creative abrasion,” the synthesis that is developed fro m
multiple p o in ts o f view , is m o re lik ely to occur. In this view , in n o vatio n occu rs at
the b o u n d a rie s betw een m in d sets, n o t w ith in the p ro vin cial te rrito ry o f one
fe o w le d g e a n d sk ill b ase” (L e o n a rd -B a rto n 1995: 62). B y h avin g access to a m o re
60 WALTER W. POWELL AND STINE GROP AL
varied set o f activities, experiences, and collaborators, com panies broaden the
resource and knowledge base that they can draw on. B y d evelopin g m ore multiplex
ties with individual partners, either through p u rsu in g m u ltiple collaborations or
expanding an existing R & D partnership into dow n stream developm ent, companies
increase the points o f contact between them . W hen relationships are deepened,
greater com m itm ent and m ore thorough know ledge sh arin g ensue. Organizations
with m ultiple and/or m ultifaceted ties to others are likely to have developed better
protocols for the exchange o f inform ation and the resolu tion o f disputes (Powell
1998). Parties that develop a broader bandw id th fo r co m m u n icatio n are, in turn,
m ore capable o f transferring com plex know ledge. In scien ce-driven fields such as
biotechnology, organizations that develop ties to different kinds o f organizations
and carry out m ultiple types o f activities w ith these o rgan izatio n s are central players
in industry networks (Powell et al. 2004). These centrally po sitio n ed organizations
are both capable o f pulling prom ising new entrants into the n etw ork and collabor
ating with a wide assortm ent o f incum bents. M oreover, research shows that in
biotechnology, organizations lacking such connections fail to keep pace and fall by
the wayside (Powell et al. 2005).
3.3 V a r i e t i e s o f N e t w o r k s
The literature on networks em phasizes that they are m ost p ro n o u n ced in the domain
between the flexibility and autonom y o f m arkets and the force and control of
organizational authority (Powell 1990). N etw orks thus co m b in e som e o f the incen
tive structures o f markets with the m on itorin g capabilities and administrative
oversight associated with hierarchies (M ow ery, O xley, and Silverm an 1996). For
our purposes, we include networks based on fo rm al contractual relations, such as
subcontracting relationships, strategic alliances or p articip ation in an industry-wide
research consortium , and inform al ties, based on co m m o n m em bership in a pro
fessional or trade association, or even a looser affiliation w ith a technological
com m unity.
One can differentiate networks with respect to their duration and stability, as well
as whether they are forged to accomplish a specific task or evolve out o f pre-existing
bonds of association. Networks vary from short-term projects to long-term rela
tionships, and the different temporal dimensions have important implications for
governance. Some networks are hierarchical, monitored by a central authority; while
others are more heterarchical, with distributed authority and strong self-organizing
features. Grabber and Powell (2004) focus on temporal stability and forms of
N E T W O R K S OF I N N O V A T O R S 6l
lack ties, such as when A knows B, and C know s B b u t n o t A . B is the bridge between
A and C, thus the gateway to a linkage between A and C . G ran ovetter (1973) argued
that bridges are the links that m ake w eak ties possible. B u rt (1992) deepened the
argum ent by m oving from the w ho question (i.e., w h ich p o sitio n in a network is best
situated) to the question o f how certain structural arrangem ents generate benefits
and opportunities. He coined the term structural holes as a potential connection
between clusters o f units that are not connected. The p o ssib ility o f m aking such a
connection provides leverage, or opportu nities fo r arbitrage. Those positioned to
take advantage o f structural holes can broker gaps in the social structure. See Figure
3.2 for illustration.
There is debate as to whether strong or w eak ties, o r brid ges o r structural holes,
offer greater opportunities for innovation (A h u ja 2000a; R u e f 20 0 2). Clearly, vari
ation in network structures is associated w ith d ifferent content in relationships.
Strong ties between two parties m ay restrict in fo rm atio n gathering in terms of the
breadth o f search, but the inform ation that is exchanged is “ thick,” or detailed and
rich. Weak ties are thinner and less durable, but p ro vid e better access to non-
redundant inform ation. There is also disagreem ent as to w hether networks can be
designed or “ pruned” to produce “ o p tim al” shapes, w ith o u t triggering repercus
sions. W hether location in a netw ork is h igh ly m alleable o r not, position in a
network both em powers and constrains opportu nities.
A third point o f contrast is between netw orks fo rm ed intentionally across a
market interface to accom plish a task and em ergent netw orks that grow out of
ongoing relationships. The form er m ay be considered an instrum en tal or strategic
relation, while the latter stems from m ore p rim o rd ial relations, such as common
ethnicity, friendship, or location. These different starting points m atter, but in the
fluid w orld o f networks, the point o f origin does not fix the evolu tion o f a relation
ship. Consider two cases. There is a global trend tow ard vertical disaggregation in
m anufacturing, as firm s are relying on suppliers for design and com ponent inputs
Structural hole
in a variety o f indu stries (see W om ack, Jones, and R oos 1990, for autos; M cK en drick,
D oner, and H aggard 2000, fo r disk drives). O ften these o u tso u rcin g decisions are
driven by the need to reduce costs, save tim e, and enhance flexibility, w hile the large
firm concentrates 011 those activities in w hich it has som e form o f com petitive
advantage. But as m an y analysts have noted, there is no natural sto pp in g p o in t in
this relation (Sabel 1994; H elper, M cD u ffie, and Sabel 2000), T h e su b co n tractor can
end up in volved in design issues, d o in g critical R & D , o r becom e central to efforts to
im prove quality. W hat began as a choice to ou tsou rce can, in som e circum stances,
becom e either a deep, m u tu ally dependent co llab o ration o r a h ig h ly am b ig u o u s and
o p p o rtu n istic p artn ersh ip. H elper et al. (2000) and D yer and N ob eoka (2000)
illustrate the m arked trend fo r au to m o b ile su b con tractin g to evolve into in terd e
pendent, bilateral relationships.
O r con sider the c o n tem p o ra ry life sciences, w here m an y R8cD partnerships
em erge out o f o n g o in g intellectual relation sh ips— co -au th o rsh ip s, m en to r-m e n tee
relationships, and co m m o n train in g (M u rray 2002). These in fo rm al person al rela
tionships m ay, how ever, co m e to involve significant intellectual p ro p e rty in the fo rm
o f patents, an d thu s b ecom e h igh ly form alized contractual agreem ents between
organizations. We o ffer these exam ples as illu strations that netw orks forged out o f
strategic p u rp oses can take on stron g relational elem ents, w hile m o re person al ties
can becom e co n tractu al and h ig h ly specified. W hile it is possible to assign netw orks
to either a tran scatio n al (i.e., based o n a con sideration o f business o p p o rtu n ities
w ith ou t regard to p rio r social relations) o r relational (i.e., em bedded in o n goin g
social relation sh ips) category, it is in ap p ro p riate to assum e that relations rem ain
fixed. A s netw orks evolve, there is considerable give and take. Intense co m p etitio n
can render calcu lative strategic alliances m ore em bedded, w hile the prospect o f great
financial rew ard can tu rn a “ h an d sh ak e55 relationship betw een in d ivid u als into a
form al legal lin kage betw een firm s.
Figure 3.3 p ro vid es a ty p o lo g y o f different fo rm s o f netw orks, w ith the h orizo n tal
axis rep resen tin g degree o f p u rp osiven ess, ran gin g fro m in fo rm al to contractual.
The vertical axis represents the extent o f em beddedness, v ary in g fro m open, e p i
sodic, o r flu id to recu rren t, dense con n ection s am o n g a fairly closed g ro u p (G ran -
ovetter, 1985). We illustrate each o f the fo u r cells w ith exam ples o f types o f
in n o vatio n n etw orks. In the lo w er left cell w e place in fo rm al netw orks, such as a
scientific in visib le college, that em erge o u t o f shared exp erien ce o r c o m m o n interest.
A lth o u gh these relation s ten d to be te m p o ra ry and sh ort-lived , the g ray arrow s are
intended to sh o w that these in fo rm al linkages can evolve into fo rm a l business
alliances o r m o re e n d u rin g p rim o rd ia l relations, w here p articip atio n is m o re c o n
stant a n d less flu id . T h e p rim o rd ia l n etw ork in the u p p er left cell is characterized by a
c o m m o n so cial iden tity, co n tin u o u s p articip atio n , an d close ties. A ll these features
are o ften fo u n d in p ro fe ssio n al netw orks, craft-b ased o ccu p atio n s, ethnic c o m m u
nities, an d in d u stria l districts. T h e u p p er rig h t cell is typ ified b y in volvem en t in a
common project. M e m b e rsh ip in such a network is typically restricted and o ften
64 W A L T E R W. P O W E L L A N D S T I N E G R O D A L
Embeddedness
Supply Chain
Primordial
Type:
Type:
Nodes dissimilar
Nodes similar
Common social identity Common work identity
Function:
Function:
*k0-% Multifunctional, craft based, Horizontal or vertical specialization
& Division of labor
J3 project-based
£ Structure:
Incremental innovation
<u
£ Dense, since the network exists Structure:
“O
<u prior to activity, work relationships Spider web
ІЛ are nested in ongoing personal The establishment of ties creates the
О
relations network The network can become a
Examples: form of social identification over
Film, construction, ethnic time, as for example in the Toyota
community, diamond trade supply chain network.
Examples:
Supply chain
Focal
firm
Caleulativeness
lnf0rmal Contractual
discussion o f the relatiom hfn her l th* low er ri§ht celL We turn now t0 a
Р ween these fo rm s and the in n o vatio n process.
3*4 E m p i r i c a l S t u d i e s o f t h e R o l e
of N e t w o r k s in I n n o v a t i o n
3 4 ,1 Formal Ties
M ost em p irical studies o f the relationship betw een netw orks and in n o vatio n focus
on fo rm al ties established am o n g organizations. T h is stream o f research d ocum en ts
a strong p o sitive relation sh ip betw een alliance fo rm atio n and in n o vation , across
such diverse indu stries as ch em icals (A h u ja 2 0 0 0 a), b io tech n o lo gy (Pow ell et aL
19 9 6 ,19 9 9 ; W alker, K o gu t, and Shan 1997; B au m , Calabrese, and Silverm an 2000),
telecom m u n icatio n s (G o d o e 2000), and sem icon d u ctors (Stuart 1998, 2000). The
d iversity o f the research contexts suggests the effects o f n etw ork structure m ay be
generalizable. N evertheless, m ost research has focused on high tech n o logy in d u s
tries, and uses patents as a p ro x y fo r in n o vation . M o re direct m easures o f in n o vative
outputs are needed. Som e o f the im p o rtan t them es that em erge fro m this research
h ighlight specific tie characteristics, tech nological uncertainty, and netw ork evo lu
tion. In ad d itio n , researchers have em phasized the increased benefits in the fo rm o f
resources and k n o w led ge that alliances p ro vid e to entrepreneurial firm s.
network ties also had a positive influence on rates o f patenting. Powell et al. (1999)
found that while network experience had a positive in flu en ce on patenting, the rate
o f increase dim inished with additional experience, suggesting possible declining
returns to network connectivity. The question o f w hether there are limits to con
nectivity needs to be investigated further.3 These results suggested a “ cycles of
learning” process in which R & D collaborations generate attention that attracts
other partners, who collaborate in developing n ovel ideas. T h is enhanced diversity
o f affiliations increases a firm's experience at m an agin g co llab o ration s and transfer
ring knowledge, and increases their centrality in the in d u stry network. Greater
centrality is associated with a higher rate o f patenting, and both centrality and
higher volum es o f patenting trigger subsequent R & D partnerships, restarting the
cycle for centrally placed firms.
M ost research has looked at the presence or absence o f a fo rm al collaboration,
Ahuja (2000a), however, developed а m ore nuanced analysis inclu ding both direct
and indirect ties, and the level o f indirectness. D raw in g data fro m a sample of 97
firm s in the global chem icals industry, he used the n u m b er o f patents as a measure of
innovative output, while collaboration was m easured th ro u gh form al ties. More
distant connections through affiliates o f partners w ere coded as w eak or indirect ties.
The results show that both direct and indirect ties have a positive influence on
innovation, though the im pact o f indirect ties is sm aller than the im pact o f direct
ties. The num ber o f direct ties also negatively m oderates the im pact o f indirect ties.
In contrast to B u rt’s (1992) argum ents about the arbitrage opportu nities available
through non-redundant contacts, A huja shows that a n etw ork w ith m any structural
holes can reduce innovative output, as m easured b y rates o f patenting. A key
advantage o f close-knit networks m ay be due to their su p erio r ability to transfer
tacit knowledge (Van W ijk, Van den Bosch, and V olberda 2003). In an analysis o f the
exchange o f inform ation across project team s in a large m u ltinational computer
company, Hansen (1999) also illustrates that com plex know ledge is transferred most
easily through tightly knit networks.
Entrepreneurial firms. One active area o f research concerns the effects o f networks
on survival chances o f newly founded firm s. Larson s (1992) ethnographic study of
how a startup firm grew and prospered by draw ing on external resources and
support for key business functions illustrates h o w relationships are forged and
sustained as startup firm s grow. W hile not explicitly lo o k in g at innovative output,
Larson added insight into the signal im portance o f netw orks in obtaining resources
necessary to fuel a startup firm s success. Shan, Walker, and K ogu t (1994) examined
whether biotechnology startup firm s' cooperative relationships w ith other firms had
a positive effect on patenting. Their results offer su p p o rt for the argum ent that
collaborative relationships increased innovation, because form al cooperative rela
tionships explained innovative output, while innovative ou tp u t did not account for
the pattern of alliances. The salience o f alliances for yo u n g and sm all firm s is further
N E T W O R K S OF I N N O V A T O R S 6/
em phasized in S tu a rts (2000) stu d y o f in n o vatio n in the sem icon d u ctor industry.
H is dataset includes 150 firm s, follow ed by the co n su ltan cy firm D ataquest over the
period i985“ 9i. D raw in g on sales figures, patterns o f strategic alliances, and
patenting activity, Stu art show s that firm s possessing tech n ologically sophisticated
alliance partn ers patented at a su bstantially greater rate than those that lacked such
ties. Firm s establishin g strategic alliances w ith large partners also grew at a h igher
rate than firm s w ith o u t access to such partners. T h e returns to netw orks w ith regard
to patentin g w ere greatest fo r both you n g and sm all firm s.
B au m , C alabrese, and Silverm an (2000) pu rsue a sim ilar question, asking h o w the
co m p o sitio n o f a startu p firm 's alliance p o rtfo lio affects its p erform an ce. U sing data
on 142 b io tech n o lo g y firm s fo u n d ed in C an ad a betw een 1991 and 1996, they find a
positive effect o f alliance fo rm atio n on startup in n o vation . N etw o rk efficiency,
defined as the d iversity o f in fo rm atio n and capabilities p er alliance, show ed a large
positive effect on the n u m b er o f biotech patents. A lliances w ith direct co m petitors
had a negative effect on in n o vation , how ever. These results w ere m o d erated w hen
the rival b io tech n o lo g y firm had a larger share o f the relevant m arket o r i f the rival
b io tech n o lo gy firm w as h ig h ly innovative. O f the vario u s p erfo rm an ce m easures
used, the n u m b er o f patents and the volu m e o f R & D expen ditu res w ere m ost
significantly in flu en ced b y rates o f alliance fo rm atio n .
than do larger firms* M o st notably, firm s w ith a central location w ith in netw orks
generate m o re in n o vative o u tp u t. B oth direct and indirect ties p ro vid e a positive
co n trib u tio n to in n o vation , but the effect o f indirect ties is m oderated by the
prevalence o f direct ties. T h e evidence for the benefits o f stru ctu ral holes is not
u n ifo rm ; w here stru ctu ral holes m igh t be beneficial is in the search fo r new in fo r
m ation, b u t the know ledge transfer process appears to be facilitated b y closer-kn it
netw orks. From the vie w o f the d yn am ics o f co llab o ration , successful external
relations app ear to beget m o re ties, w hich fuel firm grow th and in n o vation . Clearly,
there are lim its to this cycle, but research has not addressed this qu estion in depth
thus far.
The m a jo rity o f the studies review ed in this section have been carried out using
patents as the dep end en t variable and fo rm al relationships as the independent
variable. Patents p ro vid e a m easure o f n o velty that is extern ally validated through
the patent exam in atio n process, hence th ey are a useful in d icato r o f know ledge
creation (G rilich es 1990). B u t patents have som e lim itation s. Som e kinds o f in n o v
ations are not patented, and there is variatio n in the extent o f patentin g across
industries. (See C h ap ter 14 b y M alerb a on in ter-in d u stry variatio n in in n o vation
processes.) O n the other h an d, the focus o f m an y o f these studies— sem icon d u ctors,
chem icals, biotechnology-— is in fields w here paten tin g is co m m on place, and c o m
petitors in these sectors are active patentors. T h e attention to these h ig h -tech n o lo gy
industries raises qu estion s, how ever, as to the generalizability o f the results to oth er
less k n o w led ge-in ten sive industries.
O ne stu d y that speaks to differences across indu stries is Row ley, Behrens, and
K rackh ard t’s (2000) analysis o f stron g an d w eak ties in the steel and sem icon d u ctor
industries. T h is stu d y m ad e a notable effort to d istin gu ish betw een stron g affilia
tions, w here alliances entailed significant resource co m m itm en ts an d regular in ter
actions, and m o re “ a rm s-le n g th ” tran sactions, w here there w as a rap id exchange,
and the relatio n sh ip w as characterized b y less freq u en cy and depth. For exam ple,
equ ity alliances, jo in t ventures, and R & D partn ersh ips w ere categorized as stron g
ties, w h ile licensin g, patent agreem ents, and m arketin g relations were classified as
w eak ties. R ecogn izin g that w eak ties serve as brid ges to n ovel in fo rm atio n , w hile
strong ties are u sefu l fo r b o th social co n tro l and the exchange o f tacit know ledge,
they find d ivergen t results. In the steel in d u stry stron g ties are positively associated
w ith p erfo rm an ce; w h ile in sem icon d u ctors weak ties are m o re efficacious. T h e y
suggest these fin d in g s reflect the im p o rtan ce o f search and p ro d u ct in n o vatio n in
sem icon d u ctors, an d a fo cu s o n im p ro vem en ts in the p ro d u ctio n process fo r ste e l
Much of the research on buyer-supplier relations and subcontracting has
focused on more traditional industries, such as automobiles or textiles. To be sure,
these industries make considerable use o f technological advances, but they are less
science-driven. As a consequence, the sources of relevant knowledge are not as
widely dispersed. Strong ties thus tend to predominate over weak ties. But the
content of those ties can evolve, changing from contractual to relational. Consider
JO W A L T E R W. P O W E L L A N D S T I N E G R O D A L ___________ ______
relationships that connect technical profession als across organizations. A kera (20m )
uses archival data to m ap the im p o rtan ce o f the IB M user grou p Share in the early
days o f co m p u tin g , o fferin g a p o rtrait o f a large m u lti-p arty collaboration. Share was
form ed soon after IB M 's release o f the first m ain fram e co m p u ter in 1953, to enable
IB M 's costum ers to sw ap p ro gram s, and collaborate on p ro g ram m in g so that
duplicative effort w as avoided. Som e o f the m ain in n o vation s that cam e ou t o f
Share p ro vid ed the basis fo r both system s p ro g ram m in g and o peratin g system s,
w hich to d ay fo rm the b ackbon e o f m o d ern com pu ter use. Share conveyed in fo r
m ation about h ard w are changes that substantially im p ro ved the design o f IB M
com puters and periph erals. O ne o f the m ain co n trib u tio n s o f the Share netw ork
was the creation o f technical standards, but a secon d -ord er effect w as also the
d iffu sion o f kn ow ledge across com panies and betw een users.
A m odern an alogu e to Share is the L in u x co m m u n ity, fo u n d ed as a grou p o f users
tryin g to d evelop an alternative to the o p eratin g system su pplied b y M icro so ft. T h e
netw ork o f L in u x p ro g ram m ers has proven effective in d evelopin g softw are in a
h igh ly distrib u ted fashion . In the b egin n in g, m ost p ro gram m ers h ad never m et each
other and o n ly k n ew each o th er v irtu ally — b y the u sern am e they used w hen coding.
The L in u x c o m m u n ity has a v e ry m odest o rgan izatio n al structure, relying on a
co m b in atio n o f in terp erson al netw orks and an in d iv id u a ls rep u tation as a skilled
p ro gram m er to serve as the ad m issio n ticket to the n etw ork (O 'M a h o n y 2002; W eber
2003). O ne difference betw een the Share n etw ork and the L in u x n etw ork is that
corp orate interests d ro ve the Share netw ork, w h ile L in u x has been p rim a rily d riven
b y the end users. N evertheless, in both cases, the d eco m p o sab ility o f p ro g ra m m in g
tasks is an im p o rta n t facto r in facilitatin g distribu ted netw orks.
A n o th er large n etw o rk that has been w id ely studied is the n etw ork o f scientists,
often term ed in visib le colleges (C ran e 1972). A n invisible college is an in fo rm al
netw ork o f researchers w h o fo rm aro u n d a co m m o n p ro b lem o r p arad igm . B y
stud yin g in visib le colleges, C ran e (1972) h o p ed to u n d erstan d h o w know ledge
grow s and h o w the stru ctu re o f scientific co m m u n ities affects the exp an sion o f
know ledge. T h ere are n o w n u m ero u s studies o f scientific netw orks, m a p p in g the
structure o f co -a u th o rsh ip an d citation s (N ew m an 2003 pro vid es a go od o verview ),
th ou gh few attend exp licitly to the issue o f in n o vatio n . D av id (2 0 0 1), how ever,
develops a fo rm a l m o d el to sh o w that the liberal sh arin g o f know ledge w ith in the
scientific c o m m u n ity is a m a jo r d river o f sch o larly in n o vation . O ne o f the historic
characteristics o f scien tific co m m u n ities is that in fo rm a tio n an d research results
have been d istrib u ted o p e n ly am o n g m em b ers o f the relevant co m m u n ity. T h e shift
tow ard in creasin g research co m m ercializatio n b y universities has led som e scholars
to qu estio n w h eth er the in n o vative benefits o f in visib le colleges w ill persist, o r i f
co m m ercial interests w ill b lo ck in fo rm a l know ledge sh arin g a m o n g scientists
(Pow ell an d O w e n -S m ith 1998; O w en -Sm ith 2003). C h ap ter 8 b y M o w e ry and
Sampat o n u n iv e rsity -in d u stry interfaces offers a m o re d etailed d iscu ssio n o f the
ro fe-o f u n iversities in the in n o v a tio n process.
W A L T E R W. P O W E L L A N D S T I N E G R O D A L
74
Another line o f research that has attended to scientific and technological net
works, dubbed A cto r-N etw o rk T h eory (A N T ), exam ines h o w particular definitions
or configurations o f science and technology triu m p h o ver alternative conceptions,
A cto r-N etw o rk T heory is unique in its treatm ent o f artifacts and technologies, as
well as people and organizations, as m em bers o f a n etw ork (C allo n 1998; Latour
1987)* The p rim ary contribution o f A N T to the relationship betw een networks and
innovation is to show that not only can netw orks facilitate in n o vation , but they also
constrain it by determ ining the kind o f in n o vation s pro d u ced , their subsequent
interpretation, and their final use (C allon 2002).
A related line o f w ork that looks at netw orks as system s o f activity is the markets-
as-networks approach, developed by Scan dinavian m arketin g researchers (Hakans-
son and Snehota 1995), This approach exam ines the m u ltiple relationships among
organizations, and shows how these different aspects o f interorganizational rela
tionships transform and evolve over tim e. For exam ple, su p p lier networks may
change frequently, with different elem ents o f p ro d u ctio n b ein g either outsourced
or insourced (Waluszewski 1995). The resources that are exchanged among the
partners in a production netw ork are constantly changing. W hat determines
whether an entity is a resource depends on the situation , and its use in combination
with other resources. Resources are, thus, always p o lyvalen t in both use and value.
The participants in a production network, both in d iv id u ally and collectively, de
velop bonds characterized by trust and com m itm ent. These bonds also have an
organizing effect on networks, as they shape the identities o f actors, and account for
different levels o f com m itm ent am ong participants. T h is rich vein o f qualitative
research has not explicitly focused on innovation, how ever.
The various studies o f m u lti-p arty netw orks tend to em phasize the processual
aspects o f collaboration. This attention to content is w elcom e, but it sometimes
comes at the expense o f m easuring the output o f relationships, particularly how the
sharing and processing o f inform ation b y m em bers o f a netw ork can determine the
generation of novelty. A fuller understanding o f the in n o vatio n process needs to
exam ine the topic o f inform ation sharing, a subject to w h ich we now turn.
3-5 K n o w l e d g e T r a n s f e r
The role o f knowledge transfer is clearly central to the innovation process. Research
has highlighted two different aspects o f the know ledge-transfer process, each o f which
influences innovation, albeit in different respects. O ne explanation for the exchange
o f infoim ation through networks emphasizes the im portance o f complementary
N E T W O R K S OF I N N O V A T O R S 75
assets in the d ivisio n o f in n o vative lab or (M ow ery, O xley, and Silverm an 1996)* I f
firm A is good at p ro d u cin g a specific co m p o n en t and firm B is capable o f using that
com p o n en t to p ro d u ce an engine, they collaborate in a jo in t p ro d u ctio n in which
their capabilities reinforce one another. In biotech n ology, fo r exam ple, sm all firm s
w ith close ties to u n iversity scientists m ay excel at d ru g d evelopm ent, but lack the
skills and resources to m anage o r fu n d costly clinical trials. B y w o rk in g closely w ith a
research h osp ital and an established firm that has a lim ited pip elin e o f new m e d i
cines, the parties collab orate in a d ivisio n o f lab or that is m u tu ally rew arding, and
can result in the p articip an ts learn in g fro m one another, and accom p lish in g tasks
they could not do in d ivid u ally.
A second fo rm o f know ledge sh arin g occu rs w hen existing in fo rm atio n w ithin a
netw ork is reco m b in ed in novel w ays. Indeed, n o velty is often the unanticipated
result o f reco n figu rin g existin g know ledge, problem s, and solu tion s (N elson and
W inter 1982; Flem in g an d Soren son 2001). A s a consequence o f such collisions o r
tran spositions, firm s can generate so m eth in g they w ere unable to create on their
ow n. B oth fo rm s o f kn ow led ge tran sfer d epend on som e m an n er o f successful
exchange o f ideas, how ever.
A n o ft-u sed d istin ctio n is d raw n betw een tacit and explicit know ledge (C o w an ,
D avid, an d F o ra y 2000). Interest in tacit know ledge stem s fro m Polanyi ’s (1956)
argum ent that w e freq u en tly k n o w a go od deal m o re than w e can express verbally.
E xplicit kn ow led ge is h ig h ly codified , as in blueprin ts, recipes, m anuals, o r in the
form o f train in g. Tacit kn ow led ge lacks such extensive co d ificatio n (N on aka and
Takeuchi 1995). V aluable, p ro d u ctive know ledge often d em an d s considerable effort
to acquire, an d su ch kn ow led ge is freq u en tly altered in the process o f acq u isition and
application. Perh aps the m o st v iv id exam ple is the co n tin u in g effort o f U S a u to
m akers to acquire, u n d erstan d , and im plem en t the Jap an ese system o f lean p ro d u c
tion (W om ack, Jon es, an d R o o s 1990; D yer an d N o b eo k a 2000). K n ow led ge o f
co m p lex p ro d u c tio n tech nologies is rarely obtained in a fu lly digestible fo rm ;
un d erstan d in g in evitab ly entails learn in g b y using. T h e d istin ction betw een co d ified
and tacit is k e y becau se the latter d em an d s co n sid erab ly m o re trial-an d -erro r
learning to a p p ly the n ew k n o w led ge in a different setting.
M a n y studies p o in t to the relatively easy tran sferab ility o f exp licit know ledge in
contrast to tacit kn ow ledge. S im o n in (1999) show s that kn ow led ge tran sfer w ith in
alliances is n egatively affected b y b oth the nature o f know ledge an d differences in
organ izatio n al culture. H e observes im p o rtan t differences in know ledge exchange
betw een lo n g an d sh o rt-lived alliances. O ld er alliances d evelop a co m m o n language
and shared m en tal m o d els betw een partn ers, suggesting a learn in g cu rve w ith in
alliances w h ere the negative effects o f lack o f experien ce an d know ledge co m p le x ity
subside as the alliance m atu res. T h u s, as an alliance ages an d p articip an ts develop
relatio n sh ip -sp ecific u n d erstan d in g, there is the o p p o rtu n ity to co n vey m o re subtle
fo rm s o f in fo rm a tio n m o re effectively. C o m p le x tacit k n o w led ge can b eco m e m o re
exp licit as p a rtn e rs d evelo p a w id er b an d w id th o f co m m u n icatio n .
7б W A L T E R W. P O W E L L A N D S T I N B G R O P A L _ -------------------------------------_ _ _
The costs of
distinguished. H e dem onstrates that relationships between sender and receiver are
im portan t, in that b o th parties need m utual awareness o f state-of-th e-art practices.
O bviously, c o m m u n ica tio n is critical to in fo rm atio n exchange. But even when
relationships fu n ctio n well, som e know ledge is causally am b igu ou s, or stic k y and
thus not easily tran sferred. M oreover, in fo rm atio n exchange is hin dered w hen the
parties have differential levels o f ab sorp tive c a p a c ity that is, the ab ility to recognize
the value o f new in fo rm atio n , assim ilate it, and ap p ly it to com m ercial ends (C ohen
and Levinthal 1990). T h is cap acity is essential to in n o vative c a p a b ility F o r exam ple,
internal success w ith R 3 cD and R8cD expen ditures p o sitively affect a firm 's ab ility to
exploit the o p p o rtu n ities presented in external relations (C o h en and Levinthal
1990; Pow ell et al. 1996).
The p ro d u ctive tran sfer o f know ledge is also essential w hen tw o o r m o re o rgan iza
tions are able to co m b in e their d ifferent capabilities, and create a pro d u ct o r service
that th ey w o u ld not be able to construct on their ow n. A g o o d illustration is the
Italian m o torcycle in d u stry w h ere the locus o f in n o vation is b ro ad ly dispersed,
and th rough co o p eratio n , the particip an ts benefit fro m specialization an d variety
generation (L ip p a rin i, L o ren zo n i, and Z o llo 2001). Because all the participan ts
p ro vid e valuab le inpu ts, there is a h igh co m m itm en t to know ledge generation.
The lead firm s d evelop relational capacities aim ed at p o o lin g the skills o f specialized
participan ts, h elp in g the overall flo w o f in fo rm atio n and resources in the netw ork.
A parallel analysis o f the Italian packagin g m ach in ery in d u stry stresses the creation
o f a su p p lier n etw ork in w h ich specialized roles are h ig h ly co m p lem en tary
(Loren zon i an d L ip p a rin i 1999). O ver tim e, m an agers o f the core com pan ies d e
veloped a specialized su p p lier n etw ork and each p articip an t focused on a narrow ,
but h igh ly co m p etitive set o f core com petencies. T h is n etw ork structu re enabled
every step o f the su p p ly ch ain to specialize in im p ro vem en ts o f their specific
co m pon en t, th ereb y in creasin g the responsiveness o f the particip an ts to m arket
conditions.
3 .6 G o v e r n a n c e and In c e n t iv e Issu es
M any studies o f interfirm networks draw data from a single point in time and thus
do not examine how collaborations unfold over time. Even studies that do look at
dynam ics tend to do so from the perspective o f a dyad. Initially, the choice by a
young firm o f whom to partner with is often driven by resource needs. A s both firm s
and the network m ature, various dyadic choices increasingly reflect structural
properties o f the network. Thus, the existing network structure shapes search
78 W A L T E R W. P O W E L L A N D S T I N E G R O D A L
l The five that form ed the core o f the hum an sequencing com pon en t were the Whitehead
Institute, Washington University, Baylor College o f M edicine, the Joint Genome Institute
(a cluster ot three national laboratories at the D epartm ent o f Energy), and the Sanger
Institute in England.
2. Notable exceptions include Ahuja 2000#, Row ley et al. 2000, Powell et al. 2005.
3. Owen-Sm ith and Powell (2003) found that US universities w ith strong ties to a limited set
o f com m ercial partners had "fertile" patent portfolios, w ith fertility measured by the
impact o f patent citations. Those universities with few ties also had les fertile patents and
patented much less. The optim al strategy for research universities, w ith respect to patent
volum e and impact, appears to be one o f diverse ties to a w ide array o f industrial partners.
Diversity mitigates possible capture from too close relations w ith com m ercial firms.
4. Inkpen and Ross (2001) do not specify how to m easure the appropriate duration of
relationships. Instead, they assume that som e relationships becom e stale and dutiful
over time, and no longer generate benefits that outw eigh the costs o f sustaining them.
R e f e r e n c e s *
------ ( i 999)j “ Public/Private Partnerships in the United S tates” Industry and Innovation 6(2):
19 1-2 17 .
L ip p a r in i , A ., L o ren zo n i , G ., and Z ollo , M . (2001), “ D ual N etw ork Strategies: M anaging
K now ledge-based and Efficiency-based N etw orks in the Italian M otorcycle In d u stry ”
W orking paper,
Lo ren zo n i , G., and L ip p a r in i , A . (1999), “ The Leveraging o f Interfirm Relationships as a
D istinctive O rganizational C a p a b ility ” Strategic Management Journal 20: 317-38.
M c K e n d r ic k , D. G ., D o n n er , R. R, and H a g g a r d , S. (2000), From Silicon Valley to
Singapore: Competitive Advantage in the Hard Disk Drive Industry, Stanford, Calif.: Stan
ford U niversity Press.
M e r r il l , S. A ., and C oo per , R. S. (1999), “ Trends in Industrial Research and D evelopm ent:
Evidence fro m N ational D ata Sources,” in Securing America's Industrial Strength, W ash
ington, D C : N ation al A cadem y Press, 9 9 -116 .
M erto n , R. K. (1957), “ The R ole-Set: Problem s in Sociological Theory,” British journal of
Sociology 8 : 110 -2 0 .
M o w ery , D. C. (ed.) (1988), International Collaborative Ventures in US Manufacturing,
C am bridge, M ass.: Ballinger.
------ (i999)> “A m e rica s Industrial Resurgence? A n O verview ,” in U.S. Industry in 2000:
Studies in Competitive Performance. W ashington, D C : N ational A cadem y Press, 1-16 .
------ O x l e y , J. E., and S il v e r m a n , B. S. (1996), “ Strategic A lliances and Interfirm K n o w
ledge Transfer,” Strategic Management Journal 17: 77-9 1.
M u r r a y , F. (2002), “ In n o vatio n as C o -evo lu tio n o f Scientific and Technological N etw orks:
E xp lo rin g T issue Engineering,” Research Policy 3 1:13 8 9 -4 0 3 .
N ational Science B o ard (1998), Science and Engineering Indicators— 1998. A rlington, Va.:
N ational Science Fou nd ation .
N elson , R., and W in t e r , S. (1982), A n Evolutionary Theory of Economic Change, Cam bridge,
M ass.: H arvard U n iversity Press.
N ew m an , M . (2003), “ The Structure and Fun ction o f C o m p lex N etw orks,” SIAM Review 45:
167-256.
N o n aka , L, and T a k e u c h i , N . (1995), The Knowledge-Creating Company, N ew York: O xford
U niversity Press.
N oteboom , B. (1999), Inter-Firm Alliances: Analysis and Design, Lon don: Routledge.
O’M a h o n y , S. (2002), The Emergence of a New Commercial Actor: Community Managed
Software Projects, Ph.D . dissertation, D epartm en t o f M anagem ent Science and Engineer
ing, Stan ford U niversity.
O w en -S m it h , J. (2003), “ F ro m Separate System s to a H ybrid O rder: A ccum ulative A d v an
tage A cross P ublic an d P rivate Science at Research O ne Universities,” Research Policy 32(6):
10 8 1-10 4 .
-------and P o w ell , W. W. (2003), “ The E xp an d in g R ole o f U n iversity Patenting in the Life
Sciences,” Research Policy 32(9): 16 9 5 -7 11.
P io r e , M ., and S a b e l , C (1984), The Second Industrial Divide, N ew York: Basic Books.
Po lanyi , M. (1956), Personal Knowledge— Towards a Post-Critical Philosophy, Chicago:
University o f Chicago Press.
Po r tes , A ., and S e n s e n b r e n n e r , J. (1993), “ Em beddedness and Im m igration : N otes o n the
Social D eterm in an ts o f E co n o m ic A c tio n ” American Journal of Sociology 9 8 :13 2 0 -5 0 .
P o w b u , W. W* (1985), Getting Into Print: The Decision-Making Process in Scholarly Publish
ing, Chicago: University o f Chicago Press.
84 W A L T E R W. P O W E L L A N D S T I N E G R O D A L ____________________
S im o n in , B, L. (1999), "A m b igu ity and the Process o f Know ledge Transfer in Strategic
A llian ce s” Strategic Management Journal 20: 595-623.
St u a r t , T. E, (1998), "N etw o rk Positions and Propensities to C o lla b o ra te ” Administrative
Science Quarterly 43; 668-98,
------ (2000), "In tern rganizational A lliances and the Perform ance o f Firm s: A Study o f
G row th and In n ovation Rates in a H igh-tech nology In d u stry ” Strategic Management
Journal 21: 7 9 1-8 11.
S z u la n sk i , G. (1996), "E x p lo rin g Internal Stickiness: Im pedim ents to the Transfer o f Best
Practice w ith the F ir m ” Strategic Management Journal 17: 27-43.
T s a i , W., and G h o sh al , S. (1998), "S o cial C apital and Value Creation: The Role o f Intrafirm
N etw orks,” Academy of Management Journal 41: 464-76.
*U z zi, B. (1997), "S o cial Structure and C om petition in Interfirm N etw orks: The Paradox o f
E m b ed d ed n ess” Administrative Science Quarterly 42: 464-76.
V an W i j k , R ., van den B osch , F. A. J., and V o lberd a , H. W. (2003), "K n o w led ge and
N etw orks,” in M . Easterby-Sm ith and M . A. Lyles (eds.), Handbook of Organizational
Learning and Knowledge Management, O xford : Blackwell.
V in d in g ,A. L. (2002), "In tero rgan ization al D iffu sion and T ransform ation o f Know ledge in
the Process o f Product In n o vatio n ” Ph.D . Thesis, A alb o rg University.
V on H ip p e l , E. (1987), "C o o p eratio n between R ivals: In form al K n o w -h o w Trading,”
Research Policy 16: 29 1-30 2.
------ (1998), "E co n o m ics o f P rodu ct D evelopm ent by Users: The Im pact o f'S tic k y " Local
In form ation ” Management Science 44: 629-44.
*W a l k e r , G. B „ K o gu t , B „ and S h a n , W. (1997), "S o cial Capital, Structural H oles and the
Form ation o f an In d u stry N etw ork,” Organization Science 8 :10 9 -2 5 .
W a l u sz e w sk i , A. (1995), "G lu lam ,” in H. H akansson and I. Snethota (eds.), Developing
Relationships in Business Networks, Lon don: Routledge.
W e b e r , S, (2003), The Success of Open Source, C am bridge, M ass.: H arvard U n iversity Press.
W e n g e r , E. (1998), Communities of Practice, N ew York: C am bridge U niversity Press.
W o m ack , J. P., J o n es , D. T „ and R o o s , D. (1990), The Machine That Changed the World: The
Story of Lean Production, N ew York: H arp er 8c Row,
C H A P T E R 4
INNOVATION
PROCESSES
KEITH PAVITT
4.1 In t r o d u c t io n 1
4.2 C o r p o r a t e In n o v a t io n P r o c e s s e s
In order to understand this rich but poten tially co n fu sin g m osaic o f knowledge
about innovation processes, I suggest the fo llo w in g general fram ew ork:
(i ) Innovation processes involve the exp loratio n and exp loitatio n o f opportun
ities for new or im proved products, processes o r services, based either on
an advance in technical practice (“ k n o w -h o w ” ), o r a change in market
dem and, or a com bination o f the two. In n o vatio n is therefore essentially a
m atching process. The classic paper on this subject is M o w e ry and Rosenberg
(t 979 )*
(2) Innovation is inherently uncertain, given the im p o ssib ility o f predicting
accurately the cost and perform ance o f a new artifact, and the reaction of
users to it. It therefore inevitably involves processes o f learn in g through either
experim entation (trial and error) o r im pro ved un d erstan d in g (theory). Some
(but not all) o f this learning is firm -specific. T h e processes o f competition in
capitalist markets thus involve pu rposive exp erim en tation through competi
tion am ong alternative products, system s, processes, and services and the
technical and organizational processes that deliver them .
In organizing the evidence, it is useful to divid e in n o vatio n into three, partially
overlapping processes, consistent with the tw o features described above.7 Each
process is closely associated with contributions fro m particu lar academ ic discip
lines. Each process also has undergone m ajor h istorical transform ations as the
process o f innovation has evolved.
The production o f scientific and technological knowledge: a m ajo r trend, since the
industrial revolution, has been for the pro d u ction o f scientific and technological
knowledge to have becom e increasingly specialized, b y discipline, by function and
by institution. H istory and social studies o f science, tech n o lo gy and business have
contributed significantly to o u r understanding o f this tran sform ation .
The translation of knowledge into working artifacts: in spite o f the explosive growth
in scientific knowledge in recent years, th eo ry rem ains an insufficient guide to
technological practice. This reflects an underlying trend for grow in g complexity of
technological artifacts,8 and in the know ledge bases u n d erp in n in g them. Techno
logical and business history have m ade m ajo r co n trib u tio n s here as have the
cognitive sciences m ore recently.
Responding to and influencing market demand: this involves a continual process of
m atching working artifacts with users' requirem ents. T h e nature and extent o f the
opportunities to transform technological know ledge into useful artifacts vary
am ongst fields and over tim e, and determ ine in part the nature o f products,
users and m ethods o f production. In the com petitive capitalist system , corporate
technological and organizational practices со -evolve w ith m arkets. Social change
and Innovations in m arketing and m arket research have contributed to complex
problem s and equally com plex solutions to the challenge o f m atching techno
logical opportunities with m arket needs and o rgan ization al practices. These
INNOVATION PROCESS 89
processes are central concerns o f scholars in m anagem ent, econ om ics and
m arketing studies.
I n o w discuss the im p licatio n s o f change over tim e in the structure and nature o f
each o f these features o f in n o vatio n processes. Such change presents considerable
challenges to the m o d ern in n o vatio n m anager and to the co rp oratio n as a whole.
4.3 T h e P r o d u c t io n o f S c ie n t if ic a n d
T e c h n o l o g ic a l K n o w l e d g e
All the im provem ents in m achinery, however, have by no m eans been the inventions o f
those w ho had occasion to use the m achines. M a n y ...h a v e been m ade by the m akers
o f the m achines, w hen to m ake them becam e the business o f a peculiar trade: and som e
b y . . .th o se w ho are called philosophers, or m en o f speculation, w hose trade is not to do
anything but to observe everything: and w ho, u pon that account are often capable o f
com bining together the pow ers o f the m ost distant and dissim ilar o b je c ts.. . , Like every
other employment, . . it is subdivided into a number of different branches, each of which affords
occupation to a peculiar tribe or class of philosophers; and this subdivision of employmen t
in philosophy as well as in every other business, improves dexterity and saves time. (Sm ith
1937 (orig. 1776): 8, m y italics)
im possible) to m an u factu re, an d /or fail to take into account often elem entary user
requirem ents (Forrest 19 9 1).
From a co rp o rate strategy perspective, the im po rtan ce o f such in trafirm co llab o r
ation has increased the im p o rtan ce o f cross-fu n ction al integration sp an n in g d ep art
m ental b o u n d aries. Japanese au tom ob ile firm s pioneered the use o f “ h eavyw eigh t”
project m anagers, em po w ered to control resources across the firm , rep ortin g d ir
ectly to the sen ior m an agem en t team at the sam e level as the departm ental m anager
(C lark and F u jim o to 1992). These project m anagers were in turn linked to in n o v
ation processes w ith in cu stom ers and key suppliers, enabling fast, project-based
innovation. These “ h eavyw eigh t” project m anagers o ccasion ally clash w ith fu n c
tional bosses, som e o f w h o m are u n w illin g to “ give u p ” control over their resources
and object to project-led m anagem ent. M a n y large firm s n o w p ro vid e form al
training in p ro ject m an agem en t to their p rofession al project m anager, covering
such issues as m an agem en t o f fast-m o vin g project team s responsible fo r integrating
research o u tp u ts, con cep tu al and detailed design, and vario u s engineering fu n c
tions, w hile at the sam e tim e resp o n d in g to ch an gin g o r em erging cu stom er req u ire
m ents d u rin g the p ro d u c tio n process.
M an y w riters stress the im p o rtan ce o f person al contacts an d exchanges across
functions w ith in the firm to deal w ith tacit elem ents o f both pro d u ct design and its
successful tran sfer to m an u factu re and m arket. There is no perfect o r fo o lp ro o f
process fo r en su rin g effective co o rd in atio n . Indeed, so called “ best practice” can be
p ositively h arm fu l w h en its ap p lication is taken too far. Excessive use o f “ h eavy
w eigh t” project m an agers can lead to the loss o f such benefits from integration as
econom ies o f scale, and cost red u ction s fro m the use o f co m m o n co m pon en ts (e.g.
in au to m o b ile d evelopm en t; see L e o n a rd -B arto n 1995). Firm s can find it h ard to
decide w h at to d o w ith a h eavyw eigh t p ro d u ct m anager (and the associated staff)
w hen a p ro d u ct d evelo pm en t is failing. Failure to grasp this d ifficu lt nettle can lead
to the p ro b lem o f “ escalation” o r an in ab ility to term inate a failin g p ro je c t.12
M an agin g the trad e-o ffs betw een project and fu n ction al m anagem ent, and o ver
co m in g the inh eren t difficu lties in project-b ased m anagem en t, present m a jo r d iffi
culties to sen io r tech n o lo g y m anagers.
exam ple, logistics and IC T system s often d iffer betw een co m po n en t su ppliers and
integrator com pan ies, creating seriou s (albeit often technically sim ple) problem s for
co m m u n icatio n and transactions am o n g these firm s. M ore fundam entally, the
choice o f w h ich activities to ou tsou rce and w hich to retain in-h ouse defines the
“ core co m p eten ce” o f the m o d ern co rp o ratio n , m o ld in g the bou n d aries o f the firm
(H am el an d P rah alad 1994; D avies et al. 2001),
States, hom e o f a n u m ber o f MNEs w hich have a relatively low p ro po rtion o f their
R& D and patenting activity ab ro ad .5
On average, firm s fro m E U countries obtain a larger share o f patents from their
foreign subsidiaries than is true o f U S or Japanese com panies (Table 12.6). D uring
the 1969-95 p eriod, the share o f total patents o f EU firm s attributable to foreign
affiliates grew from 26.3 to 32.5 per cent. European firm s tend to concentrate a
considerable share o f their international R8cD activities in the U S (over 50 per cent
o f their foreign R & D investm ent on average, w ith G erm an, British, and Swiss firm s
showing the highest concen tration o f their foreign activities in the United States).
The foreign patenting activity o f U S firm s also increased during this period, but
rem ained below 10 per cent.6 A lth ough U S foreign R & D activities are relatively low
com pared to E U firm s, they are m uch larger than Japanese com panies, w hose
offshore patenting declined from 2.1 per cent in 19 6 9 -77 to approxim ately 1 per
cent o f their total patentin g activity in 1987-95.
O verall, M N E s have increasingly internationalized their innovative activities,
with a few relevant exceptions (m ost notably, Japanese M N E s). The im portance o f
R & D activities o f foreign affiliates has grow n in m ost host econom ies over the 1990s.
R & D b y foreign firm s is especially high in the U K , Ireland, Spain, H ungary, and
Canada, and low est in Japan , w ith other countries (including the U S, France, and
Sweden) in interm ediate positions. N evertheless, m ost R & D and patenting activities
are still largely concentrated in the M N E s' hom e countries, and in a few host
countries. Well over 90 per cent o f the R & D expenditures o f m ost M N E s is located
Notes*
a Germany, UK, Italy, France, Netherlands, Belgium, Luxembourg, Switzerland, Sweden,
Denmark, Ireland, Spain, Portugal, Greece, Austria, Norway, Finland.
b Total induces all the 784 world's largest firms recorded by the University of Reading
database, base year 1984.
Source: Cantwell and Jarme (2000).
96 KEITH PAVITT
role in innovation is likely to be low where large firms are selling to a mass market
of users who lack strong technological capabilities (e.g. many consumer products).
• Innovation processes will differ greatly between large and small firms in other
respects. Innovations in large firms involve a larger number of people in special
ized functions, with shifting responsibilities over time. Innovation processes in
large firms are also more likely to involve recognizable procedures, whether formal
or informal. In small firms, there are fewer resources to apply to such issues.
Decisions related to the recognition of opportunities, the allocation of resources,
and the coordination of functional activities, are more likely to reflect the compe
tencies and behavior of senior managers.16
The heterogeneity and contingent nature of innovation means that there can be no
simple “best practice” innovation model for firms or managers to follow. Each firm
proceeds on the basis of its prior experience and the technological trajectories
evident in the specific industry or product group. But the lack of global “best
practices” should not be taken to mean that innovation strategy does not matter,
nor that good management cannot make a difference to firms’ productivity, market
share, or profitability.
4.4 T r a n s f o r m a t i o n of Knowledge
into W orking A r t if a c t s
Scientific advances enable the creation o f artifacts o f increasing com plexity, em
bodying an increasing num ber o f subsystem s and com pon en ts, and drawing on a
broadening range o f fields o f specialized know ledge. T h is increasing system com
plexity is one consequence o f the grow ing specialization in know ledge production, it
has resulted in both better understanding o f cau se-effect relationships, and better
and cheaper m ethods o f experim entation. These advances have reduced the costs of
technological search, enabling greater com plexity in term s o f the num ber o f com
ponents, parts or m olecules that can be successfully em bod ied in a new product or
service, Developm ents within IC T are accelerating this trend: digitalization opens
options for m ore com plex systems, and sim ulation techniques reduce the costs of
experim entation (Pavitt and Steinm ueller 2001),
M anagers involved in transform ing S & T know ledge into products, systems, and
services need to be aware o f tour sets o f specific trends in their industries. These are:
(a) technology trajectories and scientific theories; (b) w here relevant, government-
funded R & D program s; (c) systems integration; and (d) techniques and approaches
to m anaging uncertainty. I deal with each o f these issues below.
INNOVATION PROCESS 97
4.5 M a t c h in g of A r t i f a c t s ,
O r g a n iz a t io n a l P r a c t ic e s a n d
M a r k e t D e m a n d
the exp loitatio n o f a rich, varied and rap id ly advan cin g b o d y o f technological
know ledge.23
The em p irical literatu re sum m arized in the first tw o colu m n s in Table 4.2 high-
lights the organ izatio n al and m arketin g practices that m ust be m ade consistent w ith
key features o f technologies:
* External linkages w ith potential custom ers, and w ith the im po rtan t sources o f
know ledge an d skills.
* Internal linkages in the key functional interfaces for experim entation and learning.
* The centralization o f resource allocation and m o n ito rin g activities needs to be
con son an t w ith the costs o f technological and m arket experim entation .
Inherent characteristics
1. Richness of la. Allocating resources for 1a. Greater opportunities not
opportunities exploring options matched by resources for
lb. Matching technologies exploring options
with product markets 1b. Matching opportunities missed
in the marketplace
2. Costs of specific 2. Degree of centralization 2. Reduced cost of experiments
experiments in decision making not matched by
decentralization or market
testing
* Criteria for resource allocation need to be con son an t w ith levels o f technological
and market opportunity*
* Alignm ent o f professional groups, w ho possess po w er an d control, with fields of
future opportunity.
The richness o f the technological and m arket o p p o rtu n ities and the scale of
technical experim ents should determ ine the app ro priate share o f resources allocated
to technological search, as well as the degree o f centralization and fluidity in
organization structures. Supporting skills and netw orks will define the specific
com petencies to be accum ulated, professional netw orks to be joined and key
functions and functional interfaces w ithin and across w h ich learning must take
place within the firm.
The particular circum stances o f the in d ivid u al firm and project will obviously
define the basic skills required for com m ercial in n o vation . But the discussion in this
chapter leads to a clear-cut conclusion: in add ition to specialist skills, “ gatekeeper”
skills and general com m unication skills n ow are m ore im p o rtan t alm ost everywhere.
People who are capable o f com m unicating across o rgan ization al barriers, disciplin
ary barriers, and professional barriers can be invaluable. In very sm all firms it maybe
satisfactory for one or two key individuals to possess the unique combination of
required skills. In larger firm s, the specific requirem ents m ay be hard to anticipate.
There is no single m anagerial planning prescription.
Differences am ongst technologies are reflected in differences in organizational
and m arketing practices. For exam ple, both ph arm aceutical and consum er electron
ics firm s see rich technological and m arket o p p ortu n ities, and thus devote sub
stantial resources to technological search. The m uch h igher costs o f experimentation
in pharm aceuticals mean that drug firm s tend to have centralized and formal
procedures for launching new products, whilst in con su m er electronics the situation
is m ore likely to be decentralized and inform al. Sim ilarly, as already mentioned,
both pharm aceutical and autom obile com panies have centralized decision struc
tures, but the form er stress interfaces between corporate R & D and public research
in biom edical fields, whilst the latter concentrate on links between R8cD and
production.
H ere a com parison between the conclusions o f tw o recent indu stry studies is
instructive, Klepper and Sim ons (2000) have show n that firm s already established
in m aking radios were subsequently the m ost successful in the new ly developing
colour T V market. O n the other hand, H o lb ro o k and his colleagues (2000) have
shown that none o f the firm s established in designing and m akin g therm ionic valves
was subsequently successful in sem iconductors.
W ith the benefit o f hindsight, we can see that success in sem iconductors required
m ore changes in technological com petencies, organizational practices, and market
experim entation am ongst incum bents, than d id success in co lo u r TV. The valve
firm s required new com petencies and netw orks in q u an tu m physics, a much
stronger interface between product design and v ery d em an d in g manufacturing
technology, and the ability to deal with new sorts o f custom ers (com puter makers
and the m ilitary, in addition to consum er electronics firm s). Fo r the radio firms, the
shift to color T V required basically the sam e technological com petencies, augmented
by w ell-know n screen technologies. O therw ise, the cu stom ers and distribution
channels rem ained unchanged, as did the key netw orks and linkages both inside
and outside the firm.
According to Chandler's (1997:76) so-called “ c o n tin u ity " thesis, the population of
incum bent large firm s has rem ained stable in recent tim es, because o f their accumu
lated skills and resources in adopting new technologies and adapting to them. This
thesis has been challenged by Lou^a and Mendon<;a (1999) and b y Freeman and
Lou^a (2001: 340-55), who argue that, on the contrary, a cohort o f new large firms
continues to join the population o f incum bent large firm s w ith each new wave of
technical change. O nly a m in o rity o f the largest firm s was able to rem ain at the top
through several waves. This suggests that the m icro-level evidence considered in this
Section, especially the factors listed in the third (righ t-h an d) colum n o f Table 4.2,
have had significant consequences for the evolu tion o f in d u stry structure.24 Radical
technological change, and the success or failure o f incum bent firm s in adapting to it,
thus m ay have im portant consequences for structural change in the econom y as a
whole.
Firm s in the vanguard o f developing and exploiting rad ically new technologies
must be distinguished from the m ore num erous firm s w ho adopt and integrate the
new technologies with their current activities. F o r these firm s, in-house competen
cies in the new technologies are background; in other w ords, necessary for the
effective adoption o f advances made outside the firm . Paradoxically, the very fact
that radically new technologies allow step-jum p reductions in the costs o f a key
input sim ultaneously makes their adoption a com petitive im perative and an un
likely source for sustained com petitive advantage am o n g adoptin g firms. For
exam ple, in the past m any factories had no choice but to adopt coal and steam—
and later electricity— as a source o f power, given their cost and other advantages. The
same is true today tor m any ICT-based m anagem ent practices. But in neither case
were these revolutionary advances by them selves a source o f sustainable competitive
I NNOVATI ON PROCESS 10J
It must be considered that there is nothing more difficult to carry out, nor more doubtful o f
success, nor more dangerous to handle, than to initiate a new order o f things. For the
reformer has enemies in all those who profit from the old order o f things, and only lukewarm
defenders in all those who would profit by the new order, this lukewarmness arising partly
from fear o f their adversaries. , . and partly from the incredulity o f mankind, who do not
truly believe in anything new until they have had actual experience o f it. Thus it arises that on
every opportunity for attacking the reformer, his opponents do so with the zeal o f partisans,
the others only defend him half-heartedly. (Machiavelli 1950: 21-2)
W ell-docum ented c o n tem p o ra ry exam ples o f this process include IB M ’s early re
luctance to enter the perso n al co m p u ter m arket and P o laro id ’s early com m itm en t
but subsequent failu re to d evelop a business based on d igital im agin g (Tripsas and
G avetti 2000). In b o th cases, a co m p an y w ith the technical resources to develop the
new tech nology, failed to d o so, due to resistance and scepticism fro m the established
p o w er structures. In these cases, it can p lau sib ly be argu ed that yesterday s "co re
com petencies,” becam e to d a y ’s "c o re rigid ities” (L e o n a rd -B arto n 1995).
B u t the n ew does n o t alw ays tu rn o u t to be better than the old. C o n servative
resistances in o il co m p an ies to investm ents in n uclear po w er in the 1970s tu rn o u t to
h ave been largely ju stified . So, m o re recently, w as scepticism ab ou t the d o t.co m
108 KEITH PAVITT
Despite spectacular im provem ents in the scientific know ledge base, and slower but
steady im provem ents in organizational know -how , in n o vation processes are neither
tidy, nor easy to delineate or m anage. Increasing specialization in the production of
artifacts and in knowledge has also increased levels o f co m p lexity— in artifacts
themselves, in the knowledge on w hich they are based, and in the organizational
form s and practices for their developm ent and com m ercial exploitation. As a
consequence— and contrary to som e o f the prediction s o f Schum peter (1962) and
Penrose (1959):
N otes
1. Keith Pavitt had prepared several drafts o f this chapter, but passed away before completing
it. Amendments to the original have been made by Christopher Freeman, Mike Hobday,
Ian Miles, and the editors.
2 This chapter focuses mainly on large firms within the USA, Europe, and Japan. It does not
consider issues arising in connection with entrepreneurial attitudes and motivations, such
as their willingness to take risks (Drucker (1985) examines some o f the links between
entrepreneurship and innovation, and Roberts (1991) and Oakey (1995) examine the
process o f innovation within SMEs). For studies o f innovation processes within firms
based in the industrializing countries (or “ latecomer firms” ) see Hobday (1995), Kim
(1997), and Fagerberg and Godinho in this volume.
3. Arguably, “ innovation managers” (in practice, if not formally identified by job title) can
be found at all levels o f the firm (for an overview o f innovation management in firms see
Tidd et al. 2001). Hamel (2000) examines “ strategic innovation” (or innovation in
strategy approach) while Schonberger (1982) and Robinson (1991) deal with kaizen (or
continuous improvements) to current vintages o f capital equipment and organization.
4. For critical assessments o f firm-level models o f innovation, ranging from the linear model
to chain-link models and more recent interactive/contingent models, see Rothwell (1992),
Forrest (1991) and Mahdi (2002).
5. The term “ stages” is avoided here, as it implies linearity. Research has consistently shown
that the process o f innovation within the firm is anything but linear (Kline and Rosenberg
1986; Tidd et al. 2001; Van der Yen et al. 1999). The three sub-processes o f innovation,
although distinctive, overlap considerably and often occur concurrently.
6. For a review o f models o f innovation and the importance o f contingency, see Mahdi (2002:
ch. 2).
7. As noted earlier, most innovation processes are overlapping and intertwined, terms like
“ stages” or “ phases” impose an unrealistic linearity on the various innovation processes.
8. Recall that by artifacts we mean services and systems as well as more tangible material
artifacts such as items o f equipment.
9. The classic texts on this are Rosenberg (1974b Price (1984). and Mowery and Rosenberg
(1989b
ПО KEI TH PAVI TT ______________ ______ ______
10. See PInchot (1997) for a discussion o f the intrapreneur, that is, an innovator operating
within a large corporation.
11. See Georghion et al. (1986).
12. For a discussion of escalation using examples from the UK stock exchange and other
major ICT project failures, see Flowers (1996). In the public sector, or state-dependent
industries, the problem can be acute. The experience o f many countries that established
agencies to develop nuclear power, for instance, suggests that these are often extremely
difficult to restrain.
13. Three very different lines o f argument have been mobilized over recent years. First is the
view that the output of university research is a free good available to everybody—as
assumed by orthodox economics. In contrast, a second view argues that publicly funded
university research is a form o f conspicuous intellectual consumption reflecting techno
logical and economic achievements, but not contributing to them (as suggested by Kealey
1996). A more nuanced argument is the “ mode 2” claim that the locus o f useful scientific
discovery is moving from universities to 4contexts o f application” (as persuasively
argued, if somewhat overstated, by Gibbons et al. 1994).
14. Universities also face management challenges here, o f course— not least in reconciling
the established structures of academic recognition with the contexts o f industrially
relevant research.
15. Georghiou et a l (1986).
16. Community Innovation Survey data (see the chapter by Smith in this volume for further
discussion) are now beginning to provide us with systematic data on many aspects of the
innovative effort and sources o f innovative ideas for firms o f different types and in
different sectors.
17. For example, in the absence of theory and/or cheap methods for constructing and testing
prototypes, the costs o f search and selection become prohibitively high— see Martin
(2000) on why Japanese swords did not improve over a period o f more than 500 years,
18. See e.g. Rosenberg and Nelson (1994) on the origins o f the engineering disciplines in US
universities.
19. A similar conclusion has been reached by Becker and Murphy (1992). They argue that the
degree of specialization in tasks is limited not by the extent o f the market, as in Smith's
famous formulation, but by the costs o f coordinating specialized activities. These
coordination costs are reduced by increases in general knowledge.
20. In addition, some prestigious academic institutions, such as Stanford University in the
USA, and the Ecole des Mines in Paris, are developing research programs in “ bio-inform
atics, growing out of the challenges to the complex results o f the Human Genome
project,
21. Sturgeon lists apparel and footwear, toys, data processing, offshore oil drilling, home
furnishings and lighting, semiconductor fabrication, food processing, automotive parts,
brewing, enterprise networking, and pharmaceuticals. In addition, Prencipe (1997)
demonstrates that outsourcing has increased in the production o f aircraft engine com
ponents.
22. Rosenberg (1994) has pointed out that in the nineteenth century, the Western Union
turned down an opportunity to purchase Bell's patent for the telephone, which it
regarded as an inferior product to the telegraph.
23. See e.g. the history of the UK General Electric Company under Arnold Weinstock (Aris
1998). ;
I NNOVATI ON PROCESS 111
24. Note also, the first point in this column (іа). This relates also to the attitude and behavior
o f the financial system. Some recent work on financial capital has returned to the original
Schumpeterian emphasis on “credit creation” for the finance o f innovation at various
stages o f the successive technological revolutions. These factors which affect the growth,
composition and fluctuation o f demand and hence the influence o f demand upon
innovation at the firm level are further considered in the chapters by Verspagen and
O’Sullivan in this volume.
R eferen c es*
A r i s , S. (1998), Arnold Weinstock and the Making of GEC, London: Aurum Press.
B a r n e y , J. (1991), “ Firm Resources and Sustained Competitive Advantage” Journal of
Management 17: 99-120.
B e c k e r , G., and M u r p h y , K. (1992), “ The Division o f Labour, Co-ordination Costs, and
Knowledge,” The Quarterly Journal of Economics 107:1137-60.
* B r u s o n i , S., P r e n c i p e , A., and P a v i t t , K. (2001), “ Knowledge Specialization, Organiza
tional Coupling and the Boundaries o f the Firm: Why Firms Know More Than They
Make” Administrative Science Quarterly 46 (4): 597-621.
C h a n d l e r , A. (1977), The Visible Hand: The Managerial Revolution in American Business,
Cambridge, Mass.: Belknap.
------ (1997), “ The United States: Engines o f Economic Growth in the Capital Intensive and
Knowledge Intensive Industries,” in A. Chandler, F. Amatori, and T. Hikino (eds.), Big
Business and the Wealth ofNationsyCambridge: Cambridge University Press.
C h r is t e n s e n , C. (1997), The Innovator's DilemmayBoston: Harvard University Press.
C l a r k , K., and F u jim o t o , T. (1992), Product Development Performance, Boston: Harvard
Business School Press.
C o n s t a n t , E. (2000), “ Recursive Practice and the Evolution o f Technological Knowledge,”
in J. Ziman (ed.), Technological Innovation as an Evolutionary Process, Cambridge: Cam-
bridge University Press.
C ooper, A., and Sch en d el , D. (1976), “ Strategic Responses to Technological Threats,”
Business Horizons (Feb.), 61-9.
D a v ie s , A., T a n g , P., Brady , T„ H obday , M., Rush , H „ and G a n n , D. (2001), Integrated
Solutions: The N ew Economy between M anufacturing and Services, Brighton: SPRU.
D rucker , P. F. (1985), Innovation and Entrepreneurship, New York: Harper & Row.
E ads , G „ and N elson , R. R. (1971), “ Government Support o f Advanced Civilian Technolo
gies: Power Reactors and the Supersonic Transport,” Public Policy (Summer): 405-28.
Fagerberg , J. (2003), “ Schumpeter and the Revival of Evolutionary Economics: an Ap
praisal of the Literature,” Journal o f Evolutionary Economics 13:125-59.
F loyd , C. (1997), M anaging Technology fo r Corporate Success, Hampshire: Gower.
F lowers , S. (1996), Software Failure: Management Failure: Amazing Stories and Cautionary
Tales, Chichester: Wiley.
M a c h ia v e l l i , N. (1950 edn.), The Prince, New York; Modern Library College Editions.
M ah d i , S. (2002), “ Search Strategy in Product Innovation Process: Theory and Evidence
from the Evolution o f Agrochemical Lead Discovery Process/' D.PhiL Thesis, Unpub
lished, SPRU, University o f Sussex,
M a n s f ie l d , E. (1995)? Innovation, Technology and the Economy, Aldershot; Edward Elgar.
M a r t i n , G. (2000), “ Stasis in Complex Artefacts," in J. Ziman (ed.), Technological Innovation
as an Evolutionary Process, Cambridge; Cambridge University Press.
M a r t in , B., and S a l t e r , A. (1996), “ The Relationship between Publicly Funded Basic
Research and Economic Performance: a SPRU Review," Report for HM Treasury, SPRU,
University o f Sussex,
M endon ^A, S, (2000), The ICT Component of Technological Diversification, M.Sc. disserta
tion, SPRU, University o f Sussex.
M eth e , D., Sw a m in a t h a n , A., and M it c h e l l , W. (1996), “ The Underemphasized Role o f
Established Firms as Sources o f Major Innovations," Industrial and Corporate Change 5:
ll8l-203.
*M it c h e ll , G., and H a m ilto n , W, (1988), “ Managing R&D as a Strategic Option," Research
Technology Management 31:15-24 .
M ow ery , D. (1995), “ The Boundaries o f the U.S. Firm in R&D," in N, R. Lamoreaux and
D. M. G. Raff (eds.), Coordination and Information: Historical Perspectives on the Organiza
tion of Enterprise, Chicago: University o f Chicago Press o f NBER.
*—— and R o se n b e r g , N. (1979), “ The Influence o f Market Demand upon Innovation: a
Critical Review o f Some Recent Empirical Studies," Research Policy 8:102-53.
------------- (1989), Technology and the Pursuit of Economic Growth, Cambridge: Cambridge
University Press.
N a r in , F., H a m ilt o n , K., and O l iv a st r o , D. (1997), “ The Increasing Linkage Between U.S.
Technology and Public Science," Research Policy 26:317-30.
N elso n , R. R. (ed.) (1982), Government and Technical Progress: A Cross-Industry Comparison,
New York: Pergamon.
N ig h t in g a l e , P. (2000), “ Economies o f Scale in Experimentation: Knowledge and Technol
ogy in Pharmaceutical R 8 c D Industrial and Corporate Change 9: 3U-59*
O a k e y , R. (1995), High-Technology New Firms, London: Paul Chapman.
Pa v it t , K., and S t e in m u e l l e r , W. (2001), “ Technology in Corporate Strategy: Change,
Continuity and the Information Revolution," in A. Pettigrew, H, Thomas and R. Whit
tington (eds,), Handbook of Strategy and Management, London: Sage.
P en ro se , E. T. (1959), The Theory of the Growth of the Firm, Oxford: Basil Blackwell.
Р ш сн о т, G. (1997), “ Innovation Through Intrapreneuring," ch. 26 in R. Katz (ed.), The
Human Side of Managing Technological Innovation, New York: Oxford University Press.
P o rter , M. (1996), “ What is Strategy?" Harvard Business Review, Nov./Dec,: 61-78.
P ra h a la d , C. K „ and H a m e l , G. (1990), “ The Core Competence o f the Corporation,"
Harvard Business Review, 90(3): 79-91.
P r e n c ip e , A. (1997), “ Technological Competencies and Product's Evolutionary Dynamics: a
Case Study from the Aero-engine Industry," Research Policy 25:1261-76.
P r ic e , D. d e S o lla (1984), “ The Science/Technology Relationship, the Graft o f Experi
mental Science, and Policy for the Improvement o f High Technology Innovation,
Research Policy 13(1), 3-20.
Q u in n , J. B, (2000), “ Outsourcing Innovation; The New Engine o f Growth," Sloan Manage
ment Review (Summer), 13-28.
114 KEITH PAVITT
ORGANIZATIONAL
INNOVATION
ALICE LAM
5.1 In t r o d u c t io n
5.2 O r g a n iz a t io n a l S t r u c t u r e
and I n n o v a t i o n
C o n ven tio n al research on o rgan ization al innovativeness has explored the d eterm in
ants o f an o rgan izatio n 's p ro p en sity to innovate. A lth o u gh researchers have analysed
the influence o f in d ivid u al, organ ization al, and environ m en tal variables (K im b erley
and Evan isko 1981; B ald rid ge and B u rn h am 1975), m ost o f the research has focused
on o rgan ization al structu re (W olfe 1994)* W ith in the field o f o rgan ization al design
theories, there has been a lo n g trad ition o f investigating the links betw een e n viro n
m ent, structures, and o rgan ization al perform an ce. Several studies have show n h ow
certain o rgan izatio n al structures facilitate the creation o f new products and
processes, especially in relation to fast ch an gin g environ m en ts. T h e w o rk o f
m icro -eco n om ists in the field o f strategy also em phasizes the su p erio rity o f certain
organizational fo rm s w ith in p articu lar types o f business strategies and p ro d u ct
m arkets (Teece 1998). M o re recently, there has been a significant shift in the focus
o f theoretical e n q u iry aw ay fro m p u rely fo rm al structures tow ards a greater interest
in organ izatio n al processes, relationships, and b ou n d aries (Pettigrew and Fenton
2000). T h e gro w in g influence o f econ om ic so cio lo g y and the in tro d u ction o f
“ n etw o rk " concepts into the o rgan ization al design field denotes such a shift. The
relationship betw een n etw ork structu re and in n o vation is dealt w ith b y Pow ell and
G rod al (C h . 3 in this vo lu m e).
Burns and Stalker ( 1961) set out to explore whether differences in the technological
and market environments affect the structure and management processes in firms.
They investigated twenty manufacturing firms in depth, and classified environments
into “ stable and predictable” and “ unstable and unpredictable ” They found that firms
could be grouped into one of the two main types, mechanistic and organic forms, with
management practices and structures that Bums and Stalker considered to be logical
responses to environmental conditions.
The Mechanistic Organization has a more rigid structure and is typically found
where the environment is stable and predictable. Its characteristics are:
(a) tasks required by the organization are broken down into specialized, functionally
differentiated duties and individual tasks are pursued in an abstract way, that is
more or less distinct from the organization as a whole;
( b) the precise definition o f rights, obligations, and technical methods is attached to
roles, and these are translated into the responsibilities o f a functional position, and
there is a hierarchical structure o f control, authority, and communication;
(c) knowledge o f the whole organization is located exclusively at the top of the
hierarchy, with greater importance and prestige being attached to internal and
local knowledge, experience, and skill rather than that which is general to the whole
organization;
{d) a tendency for interactions between members o f the organization to be vertical, Le,
between superior and subordinate.
The Organic Organization has a much more fluid set o f arrangements and is an
appropriate form to changing environmental conditions which require emergent and
innovative responses. Its characteristics are:
(a) individuals contribute to the common task o f the organization and there is
continual adjustment and redefinition o f individual tasks through interaction
with others;
(b) the spread o f commitment to the organization beyond any technical definition, a
network structure o f control authority and communication, and the direction of
communication is lateral rather than vertical;
(c) knowledge may be located anywhere in the network, with this ad hoc location
becoming the center of authority and communication;
(d) importance and prestige attach to affiliations and expertise valid in industrial and
technical and commercial milieu external to the firm.
Mechanistic and organic form s are polar types at the opposite ends o f a continuum
and, m some organizations, a m ixture o f both types could be found.
Source. Bums and Stalker (1961).
ORGANIZATIONAL INNOVATION 119
izational structu res and in n o vatio n m anagem ent. T h eir study found that firm s
could he grou p ed into one o f the tw o m ain types: the fo rm er m ore rigid and
hierarchical, suited to stable conditions; and the latter, a m ore flu id set o f arran ge
m ents, ad ap tin g to co n d itio n s o f rapid change and innovation. N either type is
inherently righ t o r w ro n g , b u t the firm 's environ m en t is the contin gency that
prom pts a structu ral response. Related is the w o rk o f Law rence and Lorsch (1967)
on prin ciples o f o rgan izatio n al differentiation and integration and h o w they adapt
to different en viron m en tal conditions, in clu din g the m arket, tech nical-econ om ic
and the scientific su b -en viro n m en ts, o f different industries. W hereas B urn s and
Stalker treat an organ izatio n as an undifferen tiated w hole that is either m echanistic
or organic, Law rence and Lorsch recognize that m echanistic and organic structures
can coexist in different parts o f the sam e o rgan ization ow in g to the different
dem ands o f the fu n ctio n al su b -en viron m en ts. T h e w ork o f these earlier authors
had a p ro fo u n d im p act on o rgan ization al th eo ry and pro vid ed useful design
guidelines fo r in n o vatio n m anagem ent. B u rn s and Stalker's m odel rem ains h igh ly
relevant fo r o u r u n d erstan d in g o f the co n tem p o rary challenges facing m any o rg an
izations in their attem pts to m ove aw ay fro m the m echan istic tow ards the organic
form o f o rgan izin g, as in n o vatio n becom es m ore im po rtan t and the pace o f en viro n
m ental change accelerates. Law rence and Lorsch ’s suggestion that m echanistic
and organic structu res can coexist is reflected in the co n tem p o rary debate about
the im p o rtan ce o f d evelo p in g hybrid m odes o f organizations— “ am b id extrou s
o rgan izatio n s"— that are capable o f copin g w ith both evo lu tio n ary an d revo lu tio n
ary tech n ological changes (T u shm an and O ’ R eilly 1996; see also Sections 5.4.2
and 5.4.3).
Another important early contribution is the work of Mintzberg (1979) who
synthesized much o f the work on organizational structure and proposed a series of
archetypes that provide the basic structural configurations of firms operating in
different environments. In line with contingency theory, he argues that the success
ful organization designs its structure to match its situation. Moreover, it develops a
logical configuration o f the design parameters. In other words, effective structuring
requires consistency of design parameters and contingency factors. T h e 4 configura
tional hypothesis" suggests that firms are likely to be dominated by one o f the five
pure archetypes identified by Mintzberg, each with different innovative potential:
simple structure, machine bureaucracy, professional bureaucracy, divisionalized
form, and adhocracy. The characteristic features of the archetypes and their innova
tive implications are shown in Table 5.1. The main thrust of the argument is that
bureaucratic structures work well in stable environments but they are not innovative
and cannot cope with novelty or change. Adhocracies, by contrast, are highly organic
and flexible forms o f organization capable o f radical innovation in a volatile envir
onment (see also Section 5.3*3)*
Contingency theories account for the diversity of organizational forms in differ
ent technological and task environments. They assume that as technology and
120 ALICE LAM
Simple stru ctu re An organic type centrally controlled Entrepreneurial and often
by one person but can respond quickly highly innovative, continually
to changes in the environment, e.g. searching for high-risk
small start-ups in high-technology. environments. Weaknesses ait
the vulnerability to individual
misjudgement and resource
lim its on g ro w th .
product m arkets becom e m o re co m p lex and uncertain, and task activities m ore
heterogeneous and u n predictable, organizations w ill adopt m ore adaptive and
flexible structures, and they w ill do so by m o vin g aw ay from bureaucratic to organic
form s o f organ izin g. T h e u n d erlyin g difficulties in achieving the “ m atch,” how ever,
are not addressed in this stran d o f research. C o n tin gen cy theories neglect the
possibility that the factors identified as m ost im po rtan t in this th eo ry are susceptible
to different in terpretation s b y o rgan ization al actors (D aft and W eick 1984), and
ignores the influence o f other factors such as m anagerial choice (C h ild 1972; 1997) o r
institutional pressures (Pow ell and D iM ag g io 1991). These aspects w ill be discussed
in Sections 5.3 and 5.4.
o u r attention to the social structure o f the enterprise and its internal cohesiveness
as a critical determ inant o f corporate strategy and in n o vative performance.
But this interpretation itself is insufficiently attentive to the contingency view-
point__the Japanese model of organizational integration works well in established
technological fields in which increm ental in n o vatio n is im portant* but not
necessarily in rapidly developing new fields w here rad ical in n o vation is vital for
com petitiveness.
Teece (1998) explains the links between firm strategy, structure, and the nature of
innovation by specifying the underlying properties o f tech nological innovation and
then proposing a related set o f organizational requirem ents o f the innovation
process. H is fram ew ork suggests that both the fo rm al (governance modes) and
inform al (cultures and values) structures, as w ell as firm s1 external network,
pow erfully influence the rate and direction o f their in n o vative activities. Based on
four classes o f variables including firm bou n daries, internal fo rm al structure, in
ternal inform al structure (culture), and external linkages, the au th or identifies four
archetypal corporate governance m odes: m u ltip rod u ct integrated hierarchy, high-
flex silicon valley type, virtual corporation, and con glom erate. H e argues that
different organizational arrangem ents are suited to different types o f competitive
environm ents and differing types o f innovation. Teece (1998: 15 6 -7 ) illustrates the
argum ent by distinguishing between two m ain types o f in n o vation , namely “ au
ton om ous” and “ system ic” innovation, and m atchin g them with different
organizational structures. A n autonom ous in n ovation is one that can be introduced
to the market w ithout m assive m odification o f related p ro d u cts and processes. An
exam ple is the introduction o f pow er steering w h ich did not initially require any
significant alternatives to the design o f cars or engines. T h is can often be advanced
rapidly by sm aller autonom ous structures, such as “ v irtu a l” firm s, accomplishing
necessary coordination through arm s-len gth arrangem ents in the open market. By
contrast, the m ove to front-w heel drive required the com plete redesign o f many
autom obiles in the 1980s. This type o f change is system ic in n o vation which favors
integrated enterprises because it requires com plex co o rd in atio n am ongst various
subsystems, and hence is usually accom plished under one “ roof.” These propos
itions, however, have yet to be em pirically verified (Teece 19 9 8 :14 6 -7 ).
The work by micro-economists highlights the interaction between market and
organizational factors in shaping innovative performance, although it devotes little
attention to the internal dynamics and social processes within organizations. Many
of the empirical predictions within this literature on the relationship between firm
strategy and structure, and innovative performance have yet to be verified, and pose
intriguing opportunities for future research.
ORGANIZATIONAL INNOVATION 1.23
The idea that organizations can think and act collectively, and serve as a repository
o f organized knowledge has stim ulated m uch research o n organizational learning
and knowledge creation. This w ork has sought to u n d erstan d h ow social inter
action and group dynam ics w ithin organizations sh ape collective intelligence,
learning, and knowledge generation, and yields im p o rtan t insights into the micro
dynam ics underpinning the innovative capab ility o f organizations. It has also
exam ined h ow shared m ental m odels o r interpretive schem es affect organizational
adaptiveness. On the positive side, som e argue that sh ared interpretative schemes
facilitate an organization's capacity to process and interpret inform ation in a
purposeful manner, prom ote organizational learning and collective problem solv
ing, and thus enhance its adaptive potential (Fiol 1993; B ro w n and D uguid 1991),
Other studies suggest that organizational interpretative schem es can create “ blind
spots" in organizational decision m aking and b lo ck organ izatio n al change (Shri-
vasta and Schneider 1984; Shrivastava et al. 1987). T h e p a ra d o x seem s to be that
organizational cognition can be at once enabling and crip p lin g, like tw o sides o f the
same coin.
View ing organizational innovation fro m the cognitive perspective shifts our
analysis from organizational structures and system s to the processes o f organiza
tional learning and knowledge creation. The analysis b elo w suggests that organiza
tions with different structural form s v ary in their patterns o f learning and knowledge
creation, giving rise to different types o f innovative capabilities. Organizational
boundaries and the social context o f learning influence an organization s cognitive
vision and its capacity for radical change and innovation.
interaction. C o llective know ledge exists between rather than w ithin individ uals. It
can be m ore or less than the sum o f the in d ivid u als' know ledge, d epending on the
m echanism s that translate in d ivid u al into collective know ledge (G lyn n 1996). Both
in divid u als and o rgan ization s are learning entities. A ll learning activities, however,
take place in a social context, and it is the nature and bou n d aries o f the context that
m ake a difference to learning outcom es.
M u ch o f the literature o n o rgan ization al learning p oin ts to the im po rtan ce o f
social interaction, context, and shared cognitive schem es for learning and kn ow -
ledge creation (N o n ak a 1994; A g y ris and Schon 1978; Lave and W enger 1991; Brow n
and D u gu id 1991, 1998). T h is bu ilds on P o lan yi’s (1966) idea that a large part o f
hum an kn ow ledge is subjective and tacit, and cannot be easily codified and tran s
m itted in dep en d en t o f the kn o w in g subject. H ence its transfer requires social
interaction an d the d evelopm en t o f shared u nd erstand in g and co m m o n in terpret
ative schem es.
N o n ak a s th eo ry o f organ izatio n al know ledge creation is rooted in the idea that
shared co gn itio n and collective learning constitute the fo u n d atio n o f organizational
know ledge creation (N o n ak a 1994; N o n ak a and Takeuchi 1995). At the heart o f the
th eo ry is the idea that tacit k n o w in g constitutes the origin o f all h u m an know ledge,
and o rgan izatio n al know ledge creation is a process o f m ob ilizin g in d ivid u al tacit
know ledge an d fo sterin g its in teraction w ith the explicit know ledge base o f the firm .
N on aka argues that know ledge needs a context to be created. H e uses the Japanese
w ord “ b a ” w h ich literally m eans “ place,” to describe such a context. “ B a” pro vid es a
shared social and m ental space fo r the interpretation o f in fo rm atio n , interaction,
and em erging relation sh ips that serves as a fo u n d atio n fo r know ledge creation.
P articipatin g in a “ ba” m eans tran scen ding one's lim ited cognitive perspective o r
social b o u n d a ry to engage in a d yn am ic process o f know ledge sh aring and creation.
In a sim ilar vein , the n o tio n o f “ c o m m u n ity o f p ractice” developed in the w o rk o f
Lave and W enger (19 9 1), W enger (1998), and B ro w n and D u gu id (1991; і 99&)5
suggests that o rgan izatio n al m em bers construct their shared identities and perspec
tives th ro u gh “ practice,” that is shared w o rk experiences. Practice pro vid es a social
activity in w h ich sh ared perspectives and cognitive repertoires develop to facilitate
know ledge sh arin g an d transfer. H ence, the w o rk g ro u p p ro vid es an im p o rtan t site
w here intense learn in g an d know ledge creation m ay develop. T h e grou p , placed at
the in tersection o f h o rizo n tal and vertical flow s o f know ledge w ith in the o rgan iza
tion, serves as a b rid ge betw een the in d ivid u al and organ izatio n in the know ledge
creation process. N o n ak a ’s th e o ry stresses the critical role o f the sem i-au to n o m o u s
project team s in kn ow led ge creation. M u ch o f the recent literature o n new and
in n o vative fo rm s o f o rg an izatio n also focuses on the use o f decentralized, group-
based structure as a key organizing principle.
Many organizational and management researchers regard the firm as a critical
social context where collective learning and knowledge creation takes place. Nonaka
and--Takeuchi- (1995) talk ab o u t the “ kn ow led ge creating c o m p a n y ” A rg y ris and
126 ALICE LA M
Schon (1978 ) suggest that an organization is, at its ro o t, a cogn itive enterprise that
learns and develops knowledge. "O rgan ization al k n o w led g e" essentially refers to the
shared cognitive schem es and distributed co m m o n u n d erstan d in g w ithin the i m
that facilitate knowledge sharing and transfer. It is sim ila r to N elson and Winter3!
{1982) concept o f "organizational routines * a kin d o f collective know ledge rooted in
shared norm s and beliefs that aids jo in t prob lem so lvin g an d capable o f supporting
com plex patterns o f action in the absence o f w ritten rules. T h e notion o f "cote
com petence" (Prahalad and H am el 1990) im plies that the learn in g and knowledge
creation activities o f firm s tend to be cum ulative and path -d epen dent. Firm s tend to
persist in what they do because learning and know ledge are em bedded in social
relationships, shared cognition, and existing w ays o f d o in g things (Kogut and
Zander 1992). Several authors have analyzed h o w collective learning in technology
depends on firm s' cum ulative com petences and evolves alo n g specific trajectories
(Dosi 1988; Pavitt 1991J. Thus, the shared context and social iden tity associated with
strong group-level learning and know ledge accu m u latio n processes m ay constrain
the evolution o f collective knowledge. Firm s m ay find it d ifficu lt to unlearn past
practices and explore alternative ways o f d oin g things. Levinthal and M arch (1993)
argue that organizations often suffer from "le a rn in g m yo p ia," and have a tendency
to sustain their current focus and accentuate their distinctive com petence, what they
call falling into a "com petency trap." The em pirical research by Leonardo-Barton
(1992) illustrates how firm s’ "core capabilities" can turn into "co re rigidities" in new
product developm ent.
A n inherent difficulty in organizational learning is the need to maintain an
external b oun dary and identity while at the sam e tim e keeping the boundary
sufficiently open to allow the flow o f new know ledge and ideas from outside.
M arch (1991) points out that a fundam ental tension in organizational learning is
balancing the com peting goals o f "th e exploitation o f old certainties” and the
exploration o f new p ossib ilities” W hereas know ledge creation is often a product
o f an organization s capability to recom bine existing know ledge and generate new
applications from its existing knowledge base, radically new learning tends to arise
from contacts with those outside the organization w ho are in a better position to
challenge existing perspectives and paradigm s. E m p irical research has suggested that
sources o f innovation often lie outside an organization ( vo n H ip pel 1988; Lundvall
1992). External business alliances and netw ork relationships, as w ell as using new
personnel to graft new knowledge onto the existing learning system s, are important
m echanism s for organizational learning and know ledge renew al in an environment
characterized by rapid technological developm ent and d isru p tive changes. The
dynam ic capability” perspective argues that the lon g-term com petitive perform-
ance o f the firm lies in its ability to build and develop firm -sp ecific capability and,
sim ultaneously, to renew and reconfigure its com petences in response to an environ
m ent m arked by creative destruction” (Teece and Pisano 1994}. T h u s, a fundamen
tal organizational challenge in innovation is not sim p ly to m aintain a static balance
ORGANIZATIONAL INNOVATION 12/
between exp lo itatio n and exp loration , o r stability and change, but a co n tin u ou s
need to balance and co ord in ate the tw o d yn am ically th rou gh ou t the organization.
The J-form organization relics on know ledge that is em bed ded in its operating
routines, team relationships, and shared culture. L earn in g and know ledge creation
within the J-fo rm takes place w ithin an “ o rgan izatio n al co m m u n ity” that in
corporates sh op-floor skills in problem solving, an d inten sive interaction and
knowledge sharing across different fu nctional units. T h e existence o f stable organ
izational careers rooted in an internal lab or m arket p ro vid e an incentive for
organizational m em bers to com m it to organizational goals and to develop firm-
specific problem -solving knowledge for co n tin u ou s p ro d u ct and process im
provem ent. N ew knowledge is generated th ro u gh the fu sion, synthesis, and
com bination o f the existing knowledge base. The J-fo rm tends to develop a strong
orientation towards pursuing an increm ental in n o vatio n strategy and to do well in
relatively m ature technological fields characterized b y rich possibilities o f combin
ations and increm ental im provem ents o f existing co m p o n en ts and products (e.g.
m achine-based industries, electronics com ponents, and autom obiles). But the
J-form 's focus on nurturing organizationally em bedded, tacit knowledge and its
em phasis on continuous im provem ent in such know ledge can inhibit learning
radically new knowledge from external sources. T h e d isap p o in tin g performance of
Japanese firm s in such fields as software and b io tech n o lo g y d u rin g the 1990s may
constitute evidence o f the difficulties faced b y “ J-fo rm firm s'" in entering and
innovating in rapidly developing new technological fields (Lam 2002; Whitley
2003; see also Box 5.2).
The adhocracy is an organic and adaptive form o f organ izatio n that is able to fuse
professional experts w ith varied skills and know ledge into adh oc project teams for
solving com plex and often highly uncertain problem s. L earn in g and knowledge
creation in an adhocracy occurs w ithin professional team s that often are composed
o f em ployees from different organizatons. C areers are u su ally structured around a
series o f discrete projects rather than advancing w ithin an in tra-firm hierarchy The
resulting project-based career system is rooted in a relatively flu id occupational
labor m arket which perm its the rapid reconfiguration o f h u m an resources to align
with shifting m arket requirem ents and technological changes. T h e adhocracy has a
m uch m ore perm eable organizational b o u n d ary that allow s the insertion o f new
ideas and knowledge from outside. This occurs th rou gh the recruitm ent o f new staff,
and the open professional networks o f the organizational m em bers that span
organizational boundaries. The adhocracy derives its com petitive strength from its
ability to reconfigure the knowledge base rapidly to deal w ith high levels o f technical
uncertainty, and to create new knowledge to produ ce n ovel in n ovation s in emerging
new industries. It is a very adaptive form o f o rgan ization capable o f dynamic
learning and radical innovation. H owever, the flu id structure and speed o f change
m ay create problem s in knowledge accum ulation, since the organization's compe
tence is em bodied in its m em bers' professional expertise and m arket-based know
how which are potentially transferable. The adh ocracy is subject to knowledge loss
when individuals leave the organization. Starbuck (1992: 725), for instance, talks
ORGANIZATIONAL INNOVATION 129
about the “ porous boundaries” of this type of organizations and points out that they
often find it hard to keep unique expertise exclusive.
The long-term survival of this loose, permeable organizational form requires the
support of a stable social infrastructure rooted in a wider occupational co m m u n ity
or localized firm networks. The example of high-technology firms in Silicon Valley
highlights the importance for the “ adhocracy” of supportive local labor markets and
other external institutions typically included in analyses of national, sectoral and
regional innovation systems (Saxenian 1996» Bahrami and Evans 2000; Angels 2000;
see also Ch. 7 in this volume by Edquist, Ch. 11 by Asheim and Gertler, and C k 14 b y
Malerba, as well as Box 5.3).
130 ALICE LAM
Silicon Valley has been an enormously successful and dynamic region characterisedЩ.
rapid innovation and commercalization in the fast growing technological fidd$/ j f c f
core industries o f the region include microelectronics, semiconductors, computer
networking, both hardware and software, and more recently biotechnology Firms'
operating in these industries undergo frequent reconfiguration and realignment in
order to survive in a constantly changing environment marked by incessant innov
ation. The availability o f a large pool o f professional experts with known reputations in
particular fields enables firms to quickly reconstitute their knowledge and sk i! base m
the course o f their innovative endeavours, The rapid creation o f new start-up firms
focusing on novel innovative projects, and the ease with which project-based firms are
able to assemble and reassemble their teams o f highly skilled scientists and engineers Ш
engage in new innovative activities are central to the technological and orpnkatioiM
dynamism o f the region. The high rate o f labor mobility and extensive hiring and firing
creates a permissive environment for entrepreneurial start-ups and flexible recon
figuration o f project teams and knowledge sources. Labor mobility within the context
o f a region plays a critical role in the generation o f professional networks and facilitates
the rapid transmission o f evolving new knowledge, a large part o f which may be tacit
Such a regionally based occupational labor market provides a stable social context and
shared industrial culture needed to ensure the efficient transfer o f tacit knowledge m an
interfirm career framework. The shared context and industry-specific values within the
regional community ensure that tacit knowledge will not be wasted when one changes
employers, and this gives the individual a positive incentive to engage in tacit “ know
how” learning (Deffillipi and Arthur 1996). A regionally based labor market and
networks o f firms create a stable social structure to sustain collective learning and
knowledge creation within and across firm boundaries. The creation o f a wider social
learning system amplifies the learning and innovative capability o f the individual firms
locating within the system. It provides an anchor o f stability for fostering and sustain
ing the innovative capability o f the adhocracy.
w ay to a m o re trad itio n al m a trix o rgan ization a decade later (Foss 2003; see Box 5.4).
Elsewhere, the m ost successful exam ples o f adhocracies are fou nd in region ally based
industrial co m m u n ities, as in the case o f Silicon Valley, and other h igh -tech n olo gy
clusters. There, the agglom eratio n o l firm s creates a stable social context and shared
Box 5,4 Oticon: the rise and decline of the "spaghetti organization"
Oticon, a Danish electronics producer, is one o f the world market leaders in hearing
aids. The company became world famous for radical organizational transformation in
the early 1990s, and has been treated as an outstanding example o f the innovative
benefits that a radical project-based organization may generate (Verona and Ravasi
2003). The “ spaghetti organization,” as it has come to be known, refers to a flat, loosely
coupled, project-based organization characterized by ambiguous job boundaries and
extensive delegation o f task and project responsibilities to autonomous teams. The
adoption o f the radical structure in 1990 represented a dramatic break from the
traditional hierarchical, functional-based organization that the company had relied
upon in the past.
The background to the implementation o f the spaghetti organization was the loss o f
competitive advantage that Oticon increasingly realized during the 1980s. Although
for decades the company had played a leading role in the hearing aids industry, at the
end o f the 1980s its products largely depended on a mature and declining technology.
The advent o f digital technology had gradually led to a shift in the technological
paradigm during the 1980s and Oticon was losing ground to its major competitors.
In 1990, the company underwent extensive restructuring in response to the crisis. The
spaghetti organization was introduced, aiming at developing a more creative and
entrepreneurial organization. The radical reorganization had immediate and strong
performance effects, resulting in a series o f remarkable innovations during the 1990s.
Despite this success, the spaghetti organization was partially abandoned from about
1996 and was gradually superseded by a more stable, traditional matrix organization.
The study by Foss ( 2003) suggests that the Oticon spaghetti organization had
encountered severe problems o f coordination and knowledge sharing between projects
because o f the fluid and adhocratic nature o f project assignments, and difficulties in
ensuring employee commitments to projects. M ore notably, Foss argues that the
spaghetti organization, as an “ internal hybrid” (Le. the infusion o f elements o f market
autonomy and flexibility into a hierarchy), was inherently unstable partly because
o f the motivational problems caused by “ selective intervention” Attempts by man
agement selectively to intervene in project selection and coordination became increas
ingly at odds with the official rhetoric that stresses self-organization. The mounting
frustration among employees eventually led to the retreat from the radical spaghetti
or^anBation.
Mthough the Oticon experiment is widely considered as a success story oforganiza-'
':>:':fionali'iffibvation, the partial retreat from the spaghetti organization iflustrates ffie
: \ihitei#:nt>diffic;ulties'. o f sustaining complete adhocracy.
.v
132 ALICE LAM
cognitive fram ew ork to sustain collective learning and reduce uncertainty associated
with swift form ation o f project team s and o rgan izatio n al changes* A n important
item for future research is a clear identification o f the p o p u latio n o f adhocracies in
different industries and regions o f the global “ kn ow ledge-b ased econom y/5 Current
w ork on this organizational type consists largely o f case studies and anecdotes,
5.4 O r g a n iz a t io n a l C h a n g e
a n d In n o v a t io n
M ost strategic adaptation theories assum e that organ izatio n al adaptation can
occur through increm ental and frequent shifts, and that n ew organizational forms
and discontinuous transform ation can be b ro u g h t ab ou t b y such processes. This
strand o f research highlights the im portance o f firm -level adap tation and internal
organizational processes in the creation o f new organ izatio n al form s. Once again,
however, m ost studies o f strategic adaptation present reti ospective studies o f suc
cessful organizational adaptation. T h ey tend to focus o n organizational restructur
ing and transform ation w ithin prevailing o rgan ization al fo rm s and are not
specifically concerned with the creation o f new organ izatio n al form s (Lewin and
Volberda 1999). We rem ain in need o f a th eo ry to account for h o w and under what
conditions m anagerial action and organizational learn in g is connected to the emer
gence o f new organizational form s.
5.5 C o n c l u s io n
The relationship between organization and in n o vation is com plex, dynamic, and
m u ltilevel The existing literature is volu m in o u s and diverse. T h is chapter has
sought to understand the nature o f the relationship fro m three different but inter
dependent perspectives: (a) the relationship betw een organizational structural
form s and innovativeness; (b) innovation as a process o f organizational learning
and knowledge creation; and (c) organizational capacity for change and adaptation.
A lthough there are potentially im portant overlaps and interconnections between
these different aspects o f the relationships, the different strands o f research have
rem ained separate and there is no single coherent conceptual fram ew ork for under
standing the phenom enon of “ organizational innovation.” T h is is partly due to the
great conceptual am biguity and confusion su rro u n d in g the term “ organizational
innovation. O ur review o f the existing literature reveals no consensus definition of
the term organizational innovation ” D ifferent researchers have used the term to
describe different aspects o f the relationships betw een o rgan ization and innovation.
Indeed, the concept has been used in a rather loose and slipp ery m anner in many
writings and som e authors are coy about stating definitions. Perhaps this conceptual
indeterm inacy reflects the fact that “ organizational in n o va tio n ” embraces a very
wide range o f phenom ena. M uch work rem ains to be done i f we are to understand
how the different dim ensions fit together.
I his large literature has advanced our understandin g o f the effects o f organiza
tional structure on the ability of organizations to learn, create knowledge, and
generate technological innovation. We know relatively less, however, about how
internal organizational dynam ics and actor learning interact w ith technological and
ORGANIZATIONAL INNOVATION 139
environm ental forces to shape o rgan ization al evolution. It rem ains unclear how and
under w hat co n d itio n s organ ization s shift fro m one structu ral archetype to another,
and the role o f tech nological in n o vation in d rivin g the process o f organizational
change is also obscure. Progress in these areas will require greater efforts to b rid ge the
different levels o f analysis and m u ltid iscip lin ary research to add insight and depth
beyond one n arro w perspective.
At present, research on organizational change and adaptation is fragmented: the
different levels o f analysis are disconnected and often rooted in different theoretical
paradigm s that use different research m ethods. T h us, ecology and evo lu tio n ary
theorists have sough t to u nd erstand the dynam ic relationship between in n o vation
and o rgan ization al evo lu tio n at the p o p u latio n o r in d u stry levels using retrospective
historical data, w hile o rgan izatio n al and m anagem en t researchers tend to exam ine
the process o f ad ap tation at the level o f in d ivid u al organizations, m o stly based on
cross-sectional case studies. The fo rm er is rooted in a structuralist determ inistic
paradigm w hereas the latter takes into account actor choice and intentionality. T h e
d isconnection betw een these tw o different levels o f analysis has m eant that w e
continue to treat selection and ad ap tation as tw o separate processes in o rgan iza
tional evolution , w hereas in reality new fo rm s o f o rgan ization em erge from the
dynam ic in teractio n betw een the tw o processes (Lew in and V olberda 1999), The
biggest challenge fo r researchers is to bridge the w id e g u lf betw een ecology/evolu-
tionary theories (dealing with organizational evolution and external forces o f
change) and strategic ch oice and learning theories (focu sin g on actor choice,
interpretation, and gro u p d yn am ics w ith in organizations). A useful avenue for
future research w o u ld con sider h o w o rgan ization al choice and evo lu tio n ary p ro
cesses interact to facilitate o rgan izatio n al change and inn ovation . T h is w ill require
lon gitud in al research on organ izatio n al ad ap tation in “ real tim e,” as distinct fro m
retrospective h isto rical case studies (Lew in et al. 1999).
A n oth er facto r that inh ibits m ajo r theoretical progress in the field is the failure o f
researchers in the fields o f in n o vatio n and o rgan ization al studies to w o rk m ore
closely together. A lth o u g h in n o vatio n scholars have lo n g recognized the im portan ce
o f the o rgan izatio n al d im en sio n o f in n o vation , m an y in n o vatio n studies continue to
be d om in ated b y an eco n o m ic app roach that allow s little ro o m for the analysis o f
creative change and in n o va tio n w ith in the o rgan ization itself. B y contrast, research
ers in the field o f o rgan izatio n al studies w h o have developed a rich literature on
organizational co gn itio n , learn in g, an d creativity rarely relate their w o rk exp licitly
to in n o vation , A s a result, this stream o f w o rk w h ich offers great poten tial fo r
un d erstan d in g the m icro -d y n a m ics o f o rgan ization al change and in n o vation
rem ains o u tsid e the m a in arena o f in n o va tio n studies. T h e b u lk o f the existing
research on the relatio n sh ip betw een o rgan ization and in n o vatio n con tin ues to
focus o n h o w tech n o lo g y an d m arket forces shape o rgan ization al ou tcom es and
treat o rgan izatio n s primarily as a vehicle or facilitator o f innovation, rather than as
innovation itself. For example, we tend to assume that technological innovation
140 ALIC E LAM
triggers o rg a n iz a tio n a l change because it shifts the com petitive environment and
forces organizations to adapt to the new set o f dem ands* T h is deterministic view
neglects the possibility that differences in organ izatio n al interpretations of, and
responses to, external stim uli can affect the ou tcom es o f organizational change. The
literature in organizational cognition argues that the en viron m en t is equivocal and
changes in the environm ent creates am b ig u ity and u n certain ty w hich prompts the
organization to em bark on a cycle o f environ m en tal scann in g, interpretation, and
learning (Daft and W eick 1984; G reve and Taylor 2000). T h e scanning and search
process m ay lead to new interpretative schem ata and o rgan ization al action which
could be an im portant source o f innovative organ izatio n al change. Treating the
organization as an interpretation and learning system directs o u r attention to the
im portant role o f internal organizational d ynam ics, actor cogn ition , and behavior in
shaping the external environm ent and outcom es o f o rgan izatio n al change.
Another prom ising direction for future research recognizes that organizational
innovation m ay be a necessary precon dition fo r tech n ological innovation, rather
than treating this process u niform ly as a response to extern al forces, and focuses on
the processes o f internal organizational reform and tran sfo rm ation that are neces
sary to create such preconditions. This requires that scholars take greater account of
the role o f endogenous organizational forces such as cap acity for learning, values,
interests, and pow er in shaping organizational evo lu tio n and technological change,
This is an area where organization and m anagem ent researchers could make a
significant contribution by placing a greater em phasis o n rigo rou s empirical re
search and theory building.
N otes
1. The term organizational innovation” is ambiguous. Some authors use it to refer to the
broad meaning of innovation or innovative behaviour in organizations” (Slappendel
1996; Sorensen and Stuart 2000), or “ organizational adoption of innovations” (Kimberley
and Evanisko 1981; Damanpour and Evan 1984; Damanpour 1996). Within these broad
meanings, the dependent variable “ innovation” is defined to encompass a range of types,
including new products or process technologies, new organizational arrangements or
administrative systems. Ihe main aim of these studies has been to identify a range of
individual, organizational, and environmental variables that affect an organization's
propensity to adopt an innovation. Others (e.g. Pettigrew and Fenton 2000) use the
term in a more restrictive way simply to refer to innovation in organizational arrange
ments. Here the dependent variable is new organizational practices or organizational
forms. Innovation may refer to the widespread adoption by organizational population of
an organizational innovation, or merely some novel combination of organizational
piocesses or structures not previously associated. There is a tendency for authors in this
camp to equate organizational innovation to organizational change or development,
ORGANIZATIONAL INNOVATION 141
assuming that change in itself is necessarily innovative, without making an explicit link
between organizational change and technological innovation,
2. For a detailed analysis o f the interaction between institutions and organizations in
innovation systems, see Edquist and Johnson (1992) and Hollingsworth (2000).
3. Institutional theorists accept that radical, innovative change would be possible in newly
emerging sectors (e.g. biotechology) where the organizational fields are “ informed” and
there is no stipulated template for organizing (Greenwood and Hinings 1996).
4. According to Tushman and O'Reilly (1996; 1999), ambidextrous organizations are ones
that can sustain their competitive advantage by operating in multiple modes simultan
eously— managing for short-term efficiency by emphasizing stability and control, and for
long-term innovation by taking risks. Organizations that operate in this way develop
multiple, internally inconsistent architectures, competences, and cultures, with built-in
capabilities for efficiency, consistency, and reliability on the one hand, and experimen
tation and improvisation on the other. During periods o f incremental change, organiza
tions require units with relatively formalized roles, responsibilities, functional structures,
and efficiency-oriented cultures that emphasize teamwork and continuous improvement.
By contrast, during periods o f ferment— times that can generate architectural and
discontinuous innovation— organizations require entrepreneurial “ skunkworks” types
o f units. These units are relatively small, have loose decentralized product structures,
experimental cultures, loose work processes, strong entrepreneurial and technical com
petences. Examples o f companies that have successfully developed ambidextrous organ
izations include Hewlett-Packard, Johnson and Johnson, and ABB (Asea Brown Boveri),
as well as such large Japanese companies as Canon and Honda.
R eferen c es
Employment Principle for a New Organizational Era, New York; Oxford University Press,
116-31.
D iM ag g io , P J., and P ow ell , W. W. (1983), “ The Iron Cage Revisited: Institutional Iso
morphism and Collective Rationality in Organizational Fields” American Sociological
Review 48; 147-60,
Dosi, G. (1988), Sources, Procedures, and Microeconomic Effects o f Innovation,” journal of
Economic Literature 2 6 :1120—71.
E dquist , C., and J ohnson , B. (1997), “ Institutions and Organizations in Systems o f Innov
ation ” in C. Edquist (ed.), Systems of Innovation: TechnologiesyInstitutions and Organiza-
tionSy London: Pinter, 41-63.
F iol , С. M. (1993), “ Consensus, Diversity, and Learning in Organizations,” Organization
Science 5: 403-20,
Foss, N. J. (2003), “ Selective Intervention and Internal Hybrids: Interpreting and Learning
from the Rise and Decline o f the Oticon Spaghetti Organization,” Organization Science
14(3): 331- 4 9 .
G a lu n ic , D. C., and E ise n h a r d t , K. M. (2001), “Architectural Innovation and Modular
Corporate Forms,” Academy of Management Journal 44(6): 1229-49.
G e r sic k , C. J. G. (1991), “ Revolutionary Change Theories: A Multilevel Exploration o f the
Punctuated Paradigm,” Academy of Management Review 16(1): 10-36.
G l y n n , M. A. (1996), “ Innovative Genius: A Framework for Relating Individual and Organ
izational Intelligence to Innovation,” Academy of Management Review 21(4): 1081-111.
G ra n t , R. M. (1996), “ Toward a Knowledge-Based Theory o f the Firm,” Strategic Manage
ment Journal 17:109 -22.
^G reenw ood , R., and H in in g s , C R. (1996), “ Understanding Radical Organizational
Change: Bringing Together the Old and New Institutionalism,” Academy of Management
Review 21(4): 1022-54.
G r e v e , H. R., and T a y l o r , A. (2000), “ Innovations as Catalysts for Organizational Change:
Shifts in Organizational Cognition and Change,” Administrative Science Quarterly 45:
54-80.
H a l l , P., and S o s k i c e , D. (eds.) (2001), Varieties of Capitalism: The Institutional Foundations
of Comparative AdvantageyOxford: Oxford University Press.
H a n n a n , M. X , and F r e e m a n , J. H. (1977). “ The Population Ecology o f Organizations,”
American Journal of Sociology 82(5): 929-63,
*------ — (1984), “ Structural Inertia and Organizational Change,” American Sociological
Review 49(2): 149-64.
H e d l u n d , G. (1994), “A Model o f Knowledge Management and The N-Form Corporation,”
Strategic Management Journal 15: 73-90.
H en derso n , R. M., and C l a r k , R. B. (1990), “Architectural Innovation: The Reconfigura
tion o f Existing Product Technologies and the Failure o f Established Firms,” Adminis
trative Science Quarterly 29: 26-42.
H in in g s , C R., T h i b a u l t , L., S l a c k , X , and K ik u l is , L. M. (1996), “ Values and Organiza
tional S t r u c t u r e Human Relations 49(7): 885-916.
H o d g k i n s o n , G. P. (2003), “ The Interface o f Cognitive and Industrial, Work and Organiza
tional Psychology” Journal o f Occupational and Organizational Psychology 76(1): 1—24.
H o l l i n g s w o r t h , J. R, (2000), “ Doing Institutional Analysis: Implications for the Study o f
Innovations,” Review o f International Political Economy 7(4): 595—644.
14 4 ALICE LAM
Kanter, R. M. (1983), The Change Masters, New York: Simon & Schuster.
K i m b e r l y . J. R„ and E v a n i s k o , M. J. (1981), “Organizational Innovation: The Influence
of Individual, Organizational, and Contextual Factors on Hospital Adoption of
Technological and Administrative Innovations,” The Academy o f Management Journal
24(4): 689-713.
Kodama, M. (2003), “Strategic Innovation in Traditional Big Business: Case Studies ofTtoo
Japanese Companies,” Organization Studies 24(2): 235-68.
K o g u t , B „ and Z a n d e r , U. (1992), “Knowledge of the Firm, Combinative Capabilities, and
the Replication of Technology,” Organization Science 3(3): 383-97.
L a m , A. (1996), “Engineers, Management and Work Organization: a Comparative Analysis of
Engineers’ Work Roles in British and Japanese Electronics Firms,” Journal of Management
Studies 33(2): 183-212.
------(1997)1 “ Embedded Firms, Embedded Knowledge: Problems o f Collaboration and
Knowledge Transfer in Global Cooperative Ventures,” Organization Studies 18(6): 973-96.
*------ (2000), “ Tacit Knowledge, Organizational Learning, Societal Institutions: an Inte
grated Framework,” Organization Studies 21(3): 487-513.
------ (2002), “Alternative Societal Models o f Learning and Innovation in the Knowledge
Economy,” International Social Science Journal 17(1): 67-82.
L a v e , J., and W e n g e r , E. (1991), Situated Learning: Legitimate Peripheral Participation. New
York: Cambridge University Press.
Law ren ce, P. R., and L o r s c h , J. W. (1967), “Differentiation and Integration in Complex
Organizations,” Administrative Science Quarterly 12:1-47.
L a z o n i c k , W., and W e s t , J. (1998), “Organizational Integration and Competitive Advan
tage,” in G. Dosi et al. (eds.). Technology, Organization, and Competitiveness, Oxford:
Oxford University Press.
L e o n a r d - B a r t o n , D. (1992), “Core Capabilities and Core Rigidities: A Paradox in Man
aging New Product Development,” Strategic Management Journal 13:111-25.
L e v i n t h a l , D. A., and M a r c h , J. G. (1993), “The Myopia of Learning,” Strategic Manage
ment Journal 14: 95-112.
* L e w i n , A. Y., and V o l b e r d a , H. W. (1999), “Prolegomena on Coevolution: a Framework
for Research on Strategy and New Organizational Forms,” Organization Science 10(5):
519- 3 4 *
L e w i n , A. Y, Long, С. P, and C a r r o l l , T. N. (1999), “The Co-evolution of New Organiza
tional Forms,” Organization Science 10: 535— 50.
L u n d v a l l , B.-A. (ed.) (1992), National Systems of Innovation: Towards a Theory ofInnovation
and Interactive Learning, London: Pinter.
M a r c h , J. G. (1991), Exploration and Exploitation in Organizational Learning,” Organiza
tion Science 2: 71— 87.
M ezia s , S. J., and G lynn , M. A. (1993), “The Three Faces of Corporate Renewal: Institution,
Revolution, and Evolution,” Strategic Management Journal 14: 77-101.
M i l e s , R. E„ S n o w , C. C„ M a t h e w s , J. A., M i l e s , G„ and C o l e m a n , H. J. Jr. (1997).
“Organizing in the Knowledge Age: Anticipating the Cellular Form,” Academy of Manage-
ment Executive 11(4): 7-20.
D„ and F r i e s e n , P. H. (1982), “ Structural Change and Performance: Quantum
M il l e r ,
versus Piecemeal-Incremental Approaches,” Academy of Management Journal 25(4):
M intzberg, H. (1979), The Structuring of Organization, Englewood Cliffs, NJ: Prentice Hall.
ORGANIZATIONAL INNOVATION Ц5
T e e c e , D.J. (1998), “Design Issues for Innovative Firms: Bureaucracy, Incentives and Indus
trial'Structure,” in A. D. Chandler Jr., P. Hagstrom, and O. Solvell (eds.). The Dynamic
Firm, Oxford: Oxford University Press, 134-65.
_____ and P i s a n o , G. (1994), “The Dynamic Capabilities of Firms: an Introduction,” Indus
Wom ack , J. P., J o n e s , D. T „ and R o o s , D. (1990), The Machine that Changed the World, N ew
York: Rawson Associates,
Woodman , R. W., S a w y e r , J. E., an d G r i f f i n , R, W. (1993), "'Toward a T h eo ry o f O rganiza
tional Creativity/5 Academy of Management Review 18(2): 293-321,
Woodward , J. (1965), Industrial Organization, Theory and Practice, London: O xford U n i
versity Press.
Z u cker , L. G, (19B7), “ In stitution al T heories o f O rganizations,55 Annual Review of Sociology
13 :443-64.
C H A P T E R 6
MEASURING
INNOVATION
KEITH SMITH
6.1 In t r o d u c t io n 1
It is som etim es suggested that in n ovation is in h eren tly im p o ssib le to quantify and
о m easure. This chapter argues that w hile this is tru e fo r so m e aspects o f innovation,
its overall characteristics do not preclude m easu rem en t o f key dimensions of
processes and outputs. A n im portan t d evelopm ent h as been the emergence o f new
ti , rS ° mi*ovat*on in P uts and outputs, in clu d in g econ o m y-w id e measures
e som e egree o f international com parability. F o llo w in g sections discuss
: .C *ss^ies *n construction and use o f science, technology, and
Г г ,™ , H- У m dlcators' the" tu ™ (briefly) to the strengths and weaknesses of
t- fn r • 1Ca ° Г k Part ctdarty and patents. Final sections cover recent initia-
innovation ^ ° n * 6 conceP tua^ zat *on, collection, and analysis o f direct measures of
which has been carried o u t three tim es in all EU M em ber States. The basic form at o f
CIS has diffused to m an y o th er cou n tries (in clu d in g C anad a, A ustralia, H ungary,
Brazil, A rgentina, and C h in a ). H as this effort been justified? In answ ering this
much depends on the q u ality o f analysis these surveys m ake possible, so the final
section discusses the rap id ly grow in g research and pu blication efforts deriving
from CIS.
6.2 T h e C o n c e p t u a l B a c k g r o u n d :
M ea su r em en t Issues
Measurement im plies co m m en su rab ility : that there is at least som e level on w hich
entities are qu alitatively sim ilar, so that co m p arison s can be m ade in quantitative
terms.
An im m ediate p ro b lem is that in n o vatio n is, b y definition, novelty. It is the
creation o f so m eth in g q u alitatively new, v ia processes o f learning and know ledge
building. It involves ch an g in g com petences and capabilities, and p ro d u cin g q u ali
tatively new p erfo rm an ce o u tco m es. T h is m ay lead to new prod u ct characteristics
that are in trin sically m easu rab le in so m e w a y — new lift/drag aspects o f an aircraft
wing, for exam ple, o r im p ro ve d fu el efficien cy o f an engine. H ow ever, such technical
measurement co m p ariso n s are o n ly rarely m ean in gfu l across products. M ore
generally, in n o vatio n in volves m u ltid im en sio n al n ovelty in aspects o f learning
or knowledge o rg an izatio n that are d ifficu lt to m easure or intrinsically n on-
measurable. K ey p ro b lem s in in n o vatio n in dicators therefore concern the und er
lying conceptualization o f the object being m easured, the m eaning o f the
measurement concept, an d the general feasib ility o f d ifferent types o f m easurem ent.
Problems o f c o m m en su ra b ility are n o t necessarily insoluble, b u t a m ain point
arising from recent w o rk is the need fo r care in distin gu ish in g betw een w hat can
and what cannot be m easu red in in n o vatio n .
Quite apart from the problem o f whether novelty can be m easured, a fundamental
definitional issue is w h at w e actu ally m ean b y “ new” (see Ch. i b y Fagerberg in this
volume). D oes an in n o v a tio n h ave to co n tain a basic n ew p rin cip le that has never
been used in the w o rld b efo re, o r d oes it o n ly need to be new to a firm ? D oes an
innovation have to incorporate a radically novel idea, or only an incremental
change? In general, what kinds o f novelty count as an innovation? These issues o f
commensurability and novelty are basic p ro b lem s for all S & T indicators R & D in
particular—-but have been most exp licitly addressed in the development of direct
innovation indicators.
150 KEITH SMITH
6 ч T h e o r i e s of I n n o v a t i o n a n d t h e i r
U se , n I n d i c a t o r D e v e l o p m e n t
A lthough statistics are often treated as though their m eanings are transparent, they
always rest on som e kind o f (usually im plicit) conceptual fou n d ation s. The system of
national accounts, for exam ple, derives from K eynesian m acroecon om ic concepts
that seek to identify com ponents o f aggregate dem and . R & D data has a complex
background in the scientification o f in n o vation — the n o tio n that acts o f research
and discovery underpin innovation (Laestadius 2003). These conceptual founda
tions are rarely considered when indicators are used. Such issues are complicated by
the fact that som e key S8cT indicators are b y-p rod u cts o f other processes— legal
procedures (as with patents), or academ ic institution s (as w ith bibliom etrics, which
rest on publishing conventions).
W hat kinds o f ideas have form ed the conceptual fo u n d atio n s o f innovation
indicators? A n im portant figure here has been N ath an R osenberg, whose work
quite explicitly affected the O E C D s Innovation M anual (O E C D 1992,1997). (This
m anual is usually called the Oslo M anual because m u ch o f the drafting and expert
meetings on it occurred there.) First, R osenberg challenged the notion o f research-
based discovery as a prelim inary phase o f in n ovation . Second, he challenged the
idea o f separability between innovation and d iffu sio n processes, pointing out
that m ost diffusion processes involve long and cu m u lative program s o f post
com m ercialization im provem ents (see R osenberg 1976 and 1982). Perhaps his
best-know n contribution, w ith Steven Kline, has been the so-called chain-link
m odel o f innovation, which stresses three basic aspects o f in n ovation (Kline and
Rosenberg 1986):
* innovation is not a sequential (linear) process but one in vo lvin g m any interactions
and feedbacks in knowledge creation
* innovation is a learning process involving m ultiple inputs
* innovation does not depend on invention processes (in the sense o f discovery of
new principles), and such processes (in volving fo rm al R8cD) tend to be under
taken as problem -solving w ithin an on goin g in n o vatio n process rather than an
initiating factor
The w ork o f Rosenberg alone, and o f R osenberg and Kline, has at least two
im portant im plications for indicator developm ent. The first is that novelty implies
not just the creation o f com pletely new products o r processes, but relatively small-
scale changes in product perform ance which m ay— over a long p eriod — have major
technological and econom ic im plications. A m eaningful in n ovation indicator
should therefore be able to pick up such change. T h e second is the im portance of
n o n -R & D inputs to innovation— design activities, engineering developm ents and
experim entation, training, exploration o f m arkets for new products, etc. So there is a
MEASURING INNOVATION 151
need for input in d icators that reflect this inpu t variety and its diverse d istributions
across activities.
The C IS effort has in general been in form ed b y ideas from recent innovation
research. O ne in p articu lar sh ou ld be m entioned, especially because it has had a
strong im pact on research u sing the new data. This is the idea that innovation relies
on collaboration an d interactive learning, in volvin g other enterprises, organ iza
tions, and the science and tech n o logy infrastructure. D ata gatherers have been
concerned to exp lore the n etw orkin g d im en sio n o f in n ovation , and this has been
an im portant con cep tu al issue in su rvey design (see H ow ells 2000, for an o verview o f
research on this top ic).
research, tooling up, etc. Expenditure on such activities can in principle be measured
(of course the practice may be difficult, since some o f these innovation-related
activities are not straightforwardly reflected in the accounting procedures of
firms). On the output side, the question is whether capability outcomes can be
measured by some tangible change in physical or economic magnitudes. Once again
there are also potential measurement areas— experience (with pilot or experimental
surveys in the 1980s) showed that firms can identify changes in their product mixes,
and can estimate sales from new or changed products (Smith 1992)- So it is possible
to define product change, in terms o f construction, use o f materials, technical
attributes, or performance characteristics, and then to look at the place of (differ
ently) changed products in the sales o f the firm. These considerations lead to
expenditure measures o f inputs to innovation, and sales measures o f outputs of
innovation. These economic measures o f innovation are clearly analogous to the
measurement o f research. This similarity in approach incidentally suggests that it
makes no sense to use R&D data while rejecting the use o f more direct innovation
data.
6.4 C u r r e n t M a j o r I n d i c a t o r s
T his section outlines the m ajor established in d icators that have been used for
innovation analysis, and provides a b rie f guide to fu rth er analysis o f them. There
are three broad areas o f indicator use in ST I analysis: first, R & D data; second, data
on patent applications, grants and citations; and third, b ib liom etric data (that is
data on scientific publication and citation).
In addition to this there are three other im p o rtan t classes o f indicators:
• technom etric indicators, which explore the technical perfo rm an ce characteristics
o f products (see e.g. Saviotti 1996 and 2001 for a theoretical view o f this, and Grupp
1994 and 1998 for analysis and em pirical specifications);
* synthetic indicators developed for scoreboard pu rp oses m ain ly by consultants (see
W orld Econom ic Forum 2003);
databases on specific topics developed as research tools b y individuals or groups
(such as the large firm database used b y Pavitt and Patel, o r the M ERIT-CATI
database on technological collaboration developed b y Jo h n H agedoorn, or the
D ISKO surveys on technological collaboration em anating from the University of
A lborg (see Patel and Pavitt 1997 and 1999, H aged oo rn and Schakenraad i 99°>
and— for extensi ve reporting on the use o f collab oration data— O E C D 2001).
Box 6.1 Bibliometric data
The fo llo w in g d iscu ssion concentrates on R & D and patents, since b ib liom etric
analysis relates p rim a rily to the d ynam ics o f science rather than in n o vation (see
M oed et al. 1995, and K alo u d is 1997 fo r review s o f the state o f the art).
stock o f commonly used knowledge and techniques in the area concerned” (OECD
2002:33). Education and training in general is not counted as R & D , Market research
is excluded. There are also many other activities with a scientific and technological
base that are kept distinct from R & D . These include such industrial activities related
to innovation as acquisition o f products and licenses? product design? trial produc
tion, training and tooling up, unless they are a component o f research, as well
as the acquisition o f equipment and machinery related to product or process
innovations.
R & D is often classified according to m ultiple criteria, an d data is collected in
highly detailed form s. Beyond the distinction betw een basic research, applied
research and developm ent the data is classified into sector o f perform ance: business
enterprise, governm ent, higher education, and p rivate n o n -p rofit. It also distin
guishes between sources o f finance, both dom estic an d in te rn a tio n a l Then there
is classification by socio-econom ic objectives, and a fu rth er classification by fields
o f research. These detailed classifications are u su ally ignored both by policy
analysts and researchers, who tend to focus on gross expenditure only (at industry
or country level), thereby m issing m ost o f the really interesting detail in the
data. For exam ple, a m ajor issue is that, w hen lo o k in g at R & D by fields of
research, IC T (inform ation and com m u n ication s technologies) turns out to be
the largest single category in all countries that classify R & D data in this way.
H ow ever m ost o f the IC T research is actually p erfo rm ed outside the IC T sector, in
the form o f systems and software developm ent b y users.3 O n the one hand, this raises
interesting questions about the cross-in d u stry significance o f the IC T sector;
but there are also questions about the extent to w h ich such activity should be
classified as R & D at all. C oncerns have also been expressed about whether
the R & D definitions are com prehensible to firm s (especially SM E s), and whether
or not there is system atic undercounting o f sm all-firm R&D (Kleinknecht,
M ontfort, and Brouw er 2002).
R & D data is always constrained as an in n ovation in d icator b y the fact that it
measures an input only (Kleinknecht et al. 2002). H ow ever, R & D also has funda
m ental advantages. These include the long period over w hich it has been collected,
the detailed subclassifications that are available in m any countries, and the relatively
good harm onization across countries. U n fo rtu n ately a great deal o f the literature
consists essentially o fa n attempt to m atch aggregate R & D m easures across time and
across sectors or countries to som e m easure o f p ro d u ctivity (see G riffith, Redding,
and Van Reenen (2000) for a very thorough recent exam ple; D o w rick (2003) is a
recent survey o f this very large literature). H ow ever this research effort is limited in
two senses on the one hand it tends to im ply (along w ith the new grow th theory,
incidentally) that R & D is the prim ary source o f p ro d u ctivity grow th, and on the
other it fails to exploit the basic com plexity o f the data that is actually available. The
disaggregation processes that are possible with R & D data contin ue to offer rich and
unexploited opportunities for researchers.
6.4*2. Policy Pitfalls: The Use and Misuse of R&D Indicators
It is w orth sayin g som eth in g about the pitfalls o f R & D as a policy indicator,
especially via the m ost w idely-u sed indicator, that o f “ R & D In te n sity ” T h is is the
ratio o f R & D exp en d itu re to som e m easure o f output. For a firm , it is usually the
R & D /Sales ratio. For an in d u stry o r a co u n try it is the ratio o f business expendi ture
on R & D (often k n ow n as B E R D ) to total p ro d u ctio n or value added. F o r a co u n try it
is usually gross exp en d itu re on R & D (G E R D ) to G D R
The R & D /G D P ratio is used in two prim ary ways. First, it is used to characterize
industries— high B E R D /G D P ratios for an indu stry are held to identify high-
technology activities. Second, a h igh G E R D /G D P ratio for a country is often believed
to indicate technological progressiveness and com m itm ent to knowledge creation
(see G o d in 2004 for an account o f the historical background to these notions).
Fo r countries, there is a d istrib u tion o f G E R D /G D P intensities, as Table 6.1
indicates. Both analysts and p o licy m akers often treat a particu lar place in the
ranking, o r the O E C D average, o r som e particu lar G E R D /G D P ratio as desirable
in itself. So C an ad a, for exam ple, has the objective o f raising its ran kin g to fifth in the
O E C D table; N o rw a y has the target o f reaching the O E C D average fo r G E R D /G D P ;
and the E U as a w h o le has a target o f reaching a G E R D /G D P ratio o f 3 per cent (it
could be argu ed that this target d om inates E U tech n ology po licy m aking at the
present tim e). B u t w h at is the in d icato r really telling us?
A basic p ro b lem is that R & D in ten sity depends on the indu strial m ix. C urren tly
the O E C D uses a fo u r-tier m odel to classify industries, in w hich the basic criterion is
the B E R D /P ro d u ctio n ratio:
3.65(1999) 1.40
Sweden
3.40 1.15
Finland
'■ '7;:
Japan 2.98 0.73
United States 2.72 0.47
.1;:•'•
$№■■;:}
h igh -R & D firm s in lo w -R & D industries and vice versa (H u ghes 1988 discusses the
intra-industry distributions using U K data).
An im portant recent m odification o f this in d icator has been the addition of
“ acquired technology,” calculated as the R & D em b od ied in capital and intermediate
goods used b y an industry, and com puted via the m ost recent in p u t-o u tp u t table.
The m ethod for calculating acquired R & D is to assum e that the R & D em bodied in a
capital good is equal to the capital good's value m u ltiplied b y the R & D intensity of
the supplying industry. The m ost recent year for w hich relevant in p u t-o u tp u t data is
generally available is 1990. The overall structure o f the classification as currently used
can be seen in Table 6.2, which shows direct R & D intensities for the m ain industrial
groups for 1997, plus the proportion o f acquired to direct R & D for 1990, the last year
for which it was calculated.
Table 6,2 Classification of industries based on RFtD intensity
M e d iu m -h ig h -te ch n o lo g y in dustries
Electrical machinery and apparatus 31 3.8 42
Motor vehicles and trailers 34 3.5 29
Chemicals 24 exc 2.6 18
2423
Railroad and transport eqpt n.e.c. 352+359 2.8 88
Machinery and eqpt n.e.c. 29 1.9 104
Sources: OECD, Science, Technology and industry Scoreboard 1999: Benchmarking Knowledge-Based
Econom ies (Paris:OECD 1999), Annex 1, p. 106; OECD, Science , Technology and Industry Scoreboard 20G1:
( Towards a Knowledge-Based Economy, Annex 1.1, pp, 13-139.
i? Mate: The JSiC elasslficatten was revised in 1996, though changes were relatively minor. 1990 data has been
reassigned to the most relevant Rev 3 category.
158 Keith smith
Table 6.2 shows that “ acquired tech n ology5' as a p ro p o rtio n o f direct R&D rises
dram atically as we m ove from high- to lo w -tech n o lo gy indu stries. This suggests that
technology intensity is likely to be very sensitive to h o w the measurement of
acquired technology is carried out. For exam ple, su p p o se we assum e that when a
firm buys a m achine it acquires not a p ro p o rtio n o f the R8dD that went into the
m achine (corresponding to the R & D /o u tp u t ratio) but all o f it? In other words,
purchasing a com puter gives the custom er access to all o f the R & D that was used to
produce it— this assum ption seems to be com patible w ith the know ledge externality
ideas o f the new grow th theory (for an o verview see Verspagen 1992, see also
Verspagen in this volum e). M aking this assu m ption w o u ld significantly alter the
rankings o f technology intensity in Table 6.2 by im p ro vin g the po sitio n o f industries
with substantial use o f R8cD em bodied in capital goods. A n o th er poin t to make here
is that so-called low -technology industries do not create o r access knowledge via
direct R8cD, and the classification is in effect biased against all industries that employ
n o n -R & D m ethods o f knowledge creation (H irsch -K rein so n et al. 2003). So the
indicator has drawbacks at the levels o f countries, indu stries and firm s; there are
therefore pitfalls in the uncritical use o f this apparently sim ple indicator.
The m ajo r sources o f patent data are the records o f the U S Patent O ffice and the
European Patent O ffice. Recent years have seen m ajo r increases in patenting activity,
as Figure 6 л show s. T h e causes o f this rise are an im p o rtan t issue: there does seem to
be grow th o f paten tin g extend in g back at least fifteen years, po ssib ly signifying
acceleration o f in n o vatio n efforts, or changes in strategic b eh avio r b y firm s; h o w
ever, the rise m ay also be shaped b y significant reductions in patent costs. (A n
analysis o f the issues here can be fo u n d in H all and Z ied o n is 2001; see also K o rtu m
and Lerner 1999.)
Patents also o f course have weaknesses, the m ost notable o f w h ich is that they are
an indicator o f invention rather than inn ovation : th ey m ark the em ergence o f a new
technical principle, not a com m ercial inn ovation . M a n y patents refer to inventions
that are intrinsically o f little technological or econ o m ic significance. M ore generally
Kleinknecht et a l have argued that
It is obvious that the patent indicator misses many non-patented inventions and innovations.
Some types of technology are not patentable, and, in some cases, it is still being debated
whether certain items (e.g. new business formulae on the internet) can be patented. On the
other hand, what is the share o f patents that is never translated into commercially viable
products and processes? And can this share be assumed to be constant across branches and
firm size classes? Moreover in some cases patent figures can be obscured by strategic behavior:
a firm will not commercialize the patent but use it to prevent a competitor patenting and
using it. (Kleinknecht et al. 2002:112)
But taking such qualifications into account, the analysis o f patent data has proven
very fruitful. Im portant achievem ents include the m ap p in g o f inventive activity over
long time periods (M acleod 1988; Sullivan 1990); assessing the im pacts o f economic
factors on the rate o f invention (Schm ookler 19 71); the elu cid ation o f the complexity
o f technological knowledge bases in large firm s (Patel and Pavitt 1999); the use and
roles o f science in industrial patenting (N arin and N o m a 1985; M eyer 2000); the
m apping o f inter-industry technology flows (Scherer 1982); the analysis o f spillovers
o f knowledge using patent citations (Jaffe, H enderson, and Trajtenberg 1993) and the
analysis o f patent values (Hall, Jaffe, and Trajtenberg 2001).
6.5 N ew In n o v a t io n In d ic a t o r s
Recent years have seen attem pts to create new and better-designed indicators
focused directly on innovation: for exam ple, the E uropean C o m m issio n has sup
ported large-scale efforts to overcom e the absence o f direct data on industrial
innovation and there have been other attem pts to im p ro ve o u r knowledge of
outputs, sources, instrum ents and m ethods o f in n o vation (recent discussions are
Hansen 2001, Guellec and Pattinson 2001: Sm ith 2002).
jou rn als or the general press— also im parts a sam ple selection bias to the exercise* In
effect what these surveys cover is an im po rtan t subset o f the pop u lation o f innov
ations: those that are new to an industry* W hat gets lost is the population of
innovation outputs w hich are “ routine,” increm ental, p art o f the norm al competi
tive activity o f firm s, yet not strikingly new en ou gh to be reported.
Fig. 6.2 The SPRU innovation database: The intersectoral flow of innovations
Source: Geroski (1 9 9 4 ).
Scope and impact of technological innovation and innovation a ctiv ity o f the enterprise
Technological innovations comprise implemented technologically new products and processes and
significant technological im provem ents in products and processes.
An innovation has been implemented if it has been introduced on the m arket (product innovation) or
used within a production process (process innovation). The product or process should be new (or
significantly improved) to the enterprise (it does not necessarily have to be new to the relevant market).
A technologically improved product is an existing product w hose perform ance has been
significantly enhanced or upgraded. A sim ple product m ay be im proved (in terms of better
performance or lower cost) th rou gh use of h igher-p erform ance com p onents or materials, or
a complex product w hich consist of a num ber of integrated technical subsystem s may be
improved by partial changes to one o f the subsystem s.
Descriptive overviews o f data results at national level. These studies are usually
written for policy makers, and typically consist o f tables and charts, accompanied
by commentary, showing results such as the distribution o f innovation expenditures
and their differences across industries, proportions o f firms introducing product or
l66 KEITH SMITH
Did your enterprise engage in the following innovation activities in 1996? txpeWiiart
Total expenditure
The expenditure items should cover current (labor If you have any RftD expenditure
costs, acquisition of services, materials etc.) and mentioned above, please
capital expenditure (instruments and equipment, in d icate...
computer software, land and buildings). If it is not
possible to estimate all expenditure items involved,
please at least indicate if your enterprise has been — percentage of RftD contracted out RftD
engaged in a particular innovation activity or not, — personnel in full tim e eqivaients in 1996
did your enterprise engage in R&D on a
~ continous basis (opposite to occasional)
between 1994 and 1996?
R esea rch and d evelop m en t o f p ro d u cts and p ro ce sse s (R ftD J com prises creative w ork undertaken on a
system atic basis in order to increase the stock of knowledge, and the use o f this stock o f know ledge to devise
new applications, Construction and testing of a prototype is often the m ost im portant phase o f REtD.
Softw are development is included as well. REtD can be carried out w ithin the enterprise or R8t0 services can
be acquired.
A cq u isitio n o f m a ch in ery and equipm ent lin ked to p ro d u c t a n d p ro c e ss in n o v a tio n s (including integrated
software) implemented by the enterprise.
A cq u isitio n o f extern a l tech n o lo g y in the form of patents, non-patented inventions, licenses, know -how ,
trademarks, draw ing plans and other consultancy services (excluding REtD), related to the im plem entation of
technological innovations, plus the acquisition of packaged softw are that is not classified elsewhere.
In du strial design and o th e r prod u ctio n p rep a ra tio n s fo r n ew p ro d u c ts include plans and d ra w in gs aimed at
defining procedures, technical specifications and operation features necessary to the production of
technologically new products and the im plem entation of new processes. D esign o f prototypes is a part of
RftD. This item also include changes in production and quality control procedures, m ethods and standards
and associated softw are required to produced the technologically new or im proved product or to use the
technologically new or improved process, Product or process m od ifications needed to start production,
including trial production (not included in R&D) is also included.
Training d ire ctly linked to in n o vatio n s is training for the im plem entation o f a technologically new or
improved product. Expenditure for training m ight include acquisition of external services and expenditure for
in-house training.
M a rk e t in tro d u ctio n o f in n ovation s includes activities in connection with the lau n ch in g of a technologically
new or improved product. These m ay include preliminary market research, m arket tests and launch
advertising, but will exclude the building of distribution netw orks to m arket innovations.
process innovations, the distribution o f different types o f new product sales across
industries, m ajor patterns o f technological collaboration, perceptions o f obstacles to
innovation, and data on objectives o f innovation. These studies tend to be important
not just in reaching p o licy m akers, but in em phasizing som e robust results w hich
em erge from this data in particu lar the conclu sion that in n o vation is pervasively
distributed across m o d ern econom ies, and that n o n -R & D inputs to in n ovation are
particularly im p o rtan t in n o n -h igh -tech sectors. In som e cases these reports are
sophisticated p ro d u ctio n s— the G erm an reports, for exam ple, rest on a substantial
panel dataset, and the C an ad ian analytical effort (sim ilar to but not identical
with C IS) is v e ry w ide ran gin g indeed (Jan z et al. 2002, and Statistics C anada:
ww w .statcan.ca ).6 M o st E U countries p rodu ce these reports and Eurostat in a d d
ition produces a E u ro p e-w id e o verview (Eurostat 2004).
6.6 C o n c l u s io n
W hile the C IS is clearly a step forw ard in term s o f the type and volu m e o f innovation
data that is available, it is ot course open to criticism . M o st criticism s focus on the
definitional restriction s in C IS w ith respect to in n ovation inputs and outputs, and
on whether an ap p ro ach that was o rigin ally adopted tor m an u factu rin g is extendable
to services. O n the ou tp u t side, the decisions m ade concerning the technological
definitions ot change o b vio u sly lim it the form s o f in n ovation that can be studied: it
seems to be the case that C IS w orks well fo r m anufactures, but not tor the extrem ely
heterogeneous services sector and its often intangible outputs. The analyses o f
Djellal and G allo uj (2001) and Tether and M iles (2001) suggest the need for quite
different approaches to data gathering on services. In defence o f the C IS approach it
can be argued that it is, and w as intended to be, m an u factu rin g-specific and that
extension to services w ou ld alw ays be problem atic. Sim ilar problem s arise w ith
other n on -tech n olo gical aspects o f in n o vation , such as organizational change (see
Lam, this volu m e, for an o verview o f organizational innovation ). It is very unclear
whether C IS, o r indeed any other su rvey-based m ethod, can grasp the dim ensions o f
this. The challenge fo r those w ho w ould go beyond this is w hether they can generate
definitional concepts, su rvey instrum ents, and collection m ethodologies that m ake
sense for other sectors o r other aspects o f innovation.
On the side ot R8cD and n o n -R & D in n ovation inputs, it is generally unclear just
how m uch o f a fir m s creative activity is captured by the types o f in n o vation outputs
that C IS m easures. A ru n d el has pointed ou t that “ W hen we talk about a firm
expending a great deal o f effort on in n o vation , we are not o n ly speaking o f financial
investm ents, b u t o f the use o f h u m an capital to think, learn and solve com plex
problem s and to p ro d u ce qu alitatively different types o f in n o vatio n s” (A rundel
1997: 6). T h is p o in t can n ot be argued w ith, but again the question arises as to w hat
can be done w ith su rvey qu estionnaires and w hat cannot. I f we w ant to explore
com plex p ro b lem so lvin g, fo r exam ple, then it is d ou btfu l w hether a su rvey in stru
ment is the righ t research to o l at all. Perhaps an u n d erlyin g issue here is the lo n g
standing ten sion betw een statistical m ethods, w ith their advantages o f generality but
lack o f depth, versus case stu d y m ethods, w h ich offer richness at the expense o f
generalizability.
N evertheless it is reasonable to conclu de that this data source is p ro vin g itself with
researchers. Both formal evaluations o f C IS as well as data tests by researchers have
been b ro ad ly p o sitive to the q u ality o f the data flo w in g from the su rvey (A alborg
U niversity 1995). O ne o f the p o sitive features o f C IS is that su rvey d efin ition and
construction, co llectio n m eth o d o logies, and general w o rk ab ility have been su b
jected to a degree o f evaluation, critique, and debate that goes far beyond anything
that has been carried out with other indicators (see Arundel et al, 1997, for one
contribution to the critical development o f C IS). This process is continuing, with
both positive and negative potential outcomes. On the positive side, the data source
m ay continue to be improved; on the negative, too much m ay be asked o f this
approach. But the real achievement is that C IS has produced results that have not
been possible with other data sources, and there is no doubt more to come as
researchers master the intricacies o f the data. In fact empirical studies using C IS
170 KEITH SMITH
data m ay well be the m ost rap idly grow ing su b -field o f pu b licatio n within innov
ation studies at the present tim e. A n interesting feature o f the pu blications using CIS
is the breadth o f w ork being done— the data is being used fo r pu blic presentations,
tor policy analyses, and for a w ide range o f sch olarly research. It w as argued above
that researchers have yet to m ake full use o f the richness o f R & D data, and this
applies even m ore to the existing survey-based in n o vatio n data. T h is source will
continue to offer considerable scope to researchers in years ahead; issues such as
innovation and firm perform ance, the use o f science b y in n o vatin g firm s, the roles of
non-R8cD inputs, and the em ploym ent im pacts o f in n o vatio n are am on g likely areas
o f developm ent.
This chapter has concentrated on the C o m m u n ity In n o vatio n Survey, but future
developm ents are unlikely to rely on this source alone. O ne possible trend is for
greater integration o f existing data sources, and this can alread y be seen in multi-
indicator approaches to such issues as national com petitiveness. Another likely
trend is for the continued developm ent o f new su rvey in stru m en ts aligned to specific
needs, along the lines o f the D ISK O surveys on in terfirm collaboration (OECD,
2001). Such developm ents are m uch to be w elcom ed as In n o vatio n Studies seeks to
generalize its propositions beyond the lim its o f the case stu d y m ethod.
A P P E N D IX 6 .1
A P P E N D I X 6.2
Regional Patterns o f Innovation: the Analysis o f CIS 2 R esu lts and Lessons from other
Innovation Surveys,
STEP SA S (Italy), 2000
Use of Multivariate Techniques to Investigate the Multidimensional Aspects of Innovation,
University of Newcastle Upon Tyne (ISRU) (UK), 2000
Statistics on Innovation in Europe,
European C om m ission, 2 0 0 1
1. I would like to thank Ian Miles, Bart Verspagen, and Richard Nelson for comments on an
earlier draft, and in particular Bronwyn Hall for comments and advice. None are impli
cated in the outcome, o f course.
2. The question o f what can be measured is an issue with all economic statistics. For example,
the national accounts do not cover all economic activity (in the sense o f all human activity
contributing to production or material welfare). They incorporate only activity that leads
to a measurable market outcome or financial recompense. This tends to leave out
economic activity such as domestic work, mutual aid, child rearing, and the informal
economy in general. Those services that are measured not by the value of output but by the
compensation o f inputs also provide problems for measurement o f output and product
ivity.
3. In both Australia and Norway, each o f which collects data by field o f research for all
industrial sectors, roughly 25 per cent o f all R&D is in ICT.
4. An excellent overview o f the literature on these and other patent issues can be found on the
website o f Bronwyn Hall: http://emlab.berkely.edu/users/bhhall See also Granstrand in
this volume.
5. For analyses using the SPRU database, see e.g. Pavitt 1983,1984; Robson et al. 1988; the
most recent sustained analytical work using the SPRU database is Geroski 1994.
6. Canada is a leading site o f policy-related indicator work at the present time— see e.g. the
outstanding work o f the Canadian Science and Innovation Indicators Consortium which
can be found at the website given above.
7. On innovation in low-tech industries, see Ch. 15 by von Tunzelmann and Acha in this
volume.
R eferen ces*
B iff l , Go and K n ell , M. (2001), “Innovation and Employment in Europe in the 1990s”
WIFO Working Paper 169/2001.
B rouw er , E,, and K lein kn ech t , A, (i997)> Measuring the Unmeasurable. A Country $
Expenditure on Product and Service Innovation, Research Policy 2 5 . 1235~ 4^*
________ (1999), "Innovative Output and a Firm's Propensity to Patent: An Exploration of
CIS Micro-Data;5 Research Policy 28: 615-24.
Cox, H., F ren z , M., and P r e v eze r , M. (2002), “ Patterns o f Innovation in UK Industry:
Exploring the CIS Data to Contrast High and Low Technology Industries;5 Journal of
Interdiciplinary Economics 13(1—3): 267-304.
*D je lla l , E, and G allo uj , R (2001), “ Innovation surveys for Service Industries: A Review,55
in Thurieaux, Arnold, and Couchot 2001: 70-6.
D ow rick , S. (2003), “A Review o f the Evidence on Science, R8cD and Productivity,55Working
Paper, Department o f Economics, Australian National University.
E uropean C ommission (1994), The European Report on Science and Technology Indicators
1994 (EUR 15897), Luxembourg.
—— (1997), Second European Report on Science and Technology Indicators 1997 (EUR 17639),
Luxembourg.
E u r o p e a n C o m m issio n (2001), Statistics on Innovation in Europe (EUR KS~32-oo~895-
EN-i), Luxembourg.
------ (2003), Third European Report on Science and Technology Indicators, Luxembourg.
eurosrat (2004), Innovation in Europe. Results for the EU, Iceland and Norway: Data 2998-
2ooi, (Luxembourg: Office for one official publications o f the European Communities)
------ San d ven , X , S ir il l i , G., and S m it h , K. (1998), “ Measuring Innovation in European
Industry,55 International Journal of the Economics of Business 5(3): 311-33.
------ and Savona , M. (2002), “ The Impact o f Innovation on Employment in Services:
Evidence from Italy,55 International Review of Applied Economics 16(3): 309—18,
E v a n g elist a , R. (1999), Knowledge and Investment, Cheltenham: Edward Elgar.
G au lt , E (ed.) (2004), Understanding Innovation in Canadian Industry, Montreal and
Kingston: McGill-Queens University Press.
G er o sk i , P. (1994), Market Structure, Corporate Performance and Innovative Activity, Oxford:
Clarendon Press.
G odin , B. (2004), “ The Obsession with Competitiveness and its Impact on Statistics: The
Construction o f High Technology Indicators,55 Research Policy (forthcoming).
G r if f it h , R., R edding , S., and V an R een en , J. (2000), “ Mapping the Two Faces of RM 3:
Productivity Growth in a Panel o f OECD Industries,55LSE Centre for Economic Performance
Discussion Paper, 2457: 1-74.
G r ilic h e s , Z. (1987), Comment,55 Washington, DC: Brookings Papers on Economic Activ
ity, 3.
G r u p p , H. (1994), The Measurement o f Technical Performance o f Innovations by Techno-
me^rics and its Impact on Established Technology Indicators,55 Research Policy 2 3 :175“ 93-
* (Д998), Foundations of the Economics of Innovation, Theory, Measurement and Practice,
Cheltenham: Elgar.
G u ellec , D„ and Pattin so n , B, (2001), “ Innovation Surveys: Lessons from OECD Coun
tries5 Experience,55 STI Review 27: 77—102.
H agedoorn , K., and Sch aken raad , J. (1990), “ Inter-firm Partnerships and Co-operative
Strategies in Core Technologies;5in C. Freeman and L. Soete (eds.), New Explorations in the
Economics of Technological Change, London: Pinter.
H all , В. H. and Ziedonis, R. H. (2002), “ The determinants o f patenting in the U.S,
semiconductor industry, 1980—1994 Rand Journal of Economics? 3 2 :10 1—28.
-----Jaffe, A., and Trajtenberg, M., 2001, “ The NBER Patent Citations Daa file: Lessons,
Insights and Methodological Tools^, NBER Working Paper No. 8498.
♦ H a n s e n , J. A. (2001), “ Technological Innovation Indicators: A Survey o f Historical Devel
opment and Current Practice ” in M. R Feldmann and A. Link (eds.), Innovation Policy in
the Knowledge-Based Economy; Dordrecht: Kluwer, 73-103.
H esslem a n , L. (2002), A Description o f Responses to the UK Community Innovation
Survey 2,” The Journal of Interdiciplinary Economics 13(1-3): 243-66.
H inloopen , f. (2003), “ Innovation Performance across Europe," Economics of Innovation
and New Technology 12(2): 145-61,
H irsc h -K r e in se n , H., J a c o bsen , D., L a est a d iu s , S., and S m it h , K. (2003), “ Low Tech
Industries and the Knowledge Economy: State o f the Art and Research Challenges,"
Working Paper 2003:10? Dept o f Industrial Economics and Management, Royal Institute
ofTechnology (KTH), Stockholm.
Holbrook , J. (1991), “ The Influence o f Scale Effects on International Comparisons o f R&D
Expenditure," Science and Public Policy 18(4): 259-62.
H owells , J. (2000), “ Innovation Collaboration and Networking: A European Perspective,"
in Science Policy Support Group, European Research? Technology and Development Issues
for a Competitive Future? London.
H u gh es , K. (1988), “ The Interpretation and Measurement o f R&D Intensity: A Note,"
Research Policy 17: 301-7,
Inzelt , A. (2002), “Attempts to Survey Innovation in the Hungarian Service Sector," Science
and Public Policy 29(5): 367-83.
Iv er sen , E. (1998), “ Patents," in K, Smith (ed.), Science? Technology and Innovation Indica
tors—a Guide for Policymakers? IDEA Report 5, STEP Group Oslo,
J aefe , A., and Henderson, R., and Trajtenberg, M., 1997, “ University Versus Corporate
Patents: A Window on the Basicness o f Invention", Economics of Innovation and New
Technology? 19-50.
J a n z , N., et al. (2002), Innovation Activities in the German Economy Report on Indicators from
the Innovation Survey 2000, Mannheim: ZEW.
K aloudis , A. (1998), “ Bibliometrics," in K. Smith (ed,), Science? Technology and Innovation
Indicators— a Guide for Policymakers? IDEA Report 5, STEP Group Oslo.
*K l ein k n ec h t , A. (ed.) (1996), Determinants of Innovation: The Message From New Indica
tors? London: Macmillan.
------and В а ш , D. (eds.) (1993), New Concepts in Innovation Output Measurement? London:
Macmillan.
------and M o h n en , P. (eds.) (2002), Innovation and Firm Performance. Econometric Explor
ations of Survey Data? Hampshire and New York: Palgrave.
------V an M o n tfo rt , K., and B ro uw er , E. (2002), “ The Non-Trivial Choice Between
Innovation Indicators," Economics of Innovation and New Technology 11(2): 109-21.
Kline , S., and R osenberg , N. (1986), “An Overview o f Innovation," in R. Landau (ed.), The
Positive Sum Strategy: Harnessing Technology for Economic Growth?Washington: National
Academy Press, 275-306,
Kortum, S„ and L e r n e r , J. (1999), “ What is Behind the Recent Surge in Patenting?" Research
Policy 28: 1-22,
I76 KEITH SMITH
THF SYSTFMTC
td ^ L» «ѵ іів тМ шт* ЛШшштИ * Jb ▼ ,я Д ь hJ H L
N A T U R E OF
ялкь *A * JKm » « А » «Л ^ т^шттш
INNOVATION
In t r o d u c t i o n to P a r t II
S Y S T E M S OF
INNOVATION
PERSPECTIVES
AND CHALLENGES
CHARLES EDQUIST
7.1 In t r o d u c t i o n 1
7.2 T he E m e r g e n c e a n d D e v e l o p m e n t
of t h e SI A p p r o a c h
The chapter b y Fagerberg in this volum e highlights the system ic nature o f innov
ation processes, noting that firm s do not n o rm ally in n ovate in isolation, but in
collaboration and interdependence w ith other organizations. These organizations
m ay be other firm s (suppliers, custom ers, com petitors, etc.) o r non-firm entities
such as universities, schools, and governm ent m inistries. T h e beh avior o f organiza
tions is also shaped by institutions— such as laws, rules, n orm s, and routines— that
constitute incentives and obstacles for innovation. These organizations and insti
tutions are com ponents o f systems for the creation and com m ercialization of
knowledge. Innovations emerge in such "system s o f in n o v a tio n "
The innovation concept used in this chapter is w id e and include product innov
ations as well as process innovations. Product in n o vation s are new — or better—
m aterial goods as well as new intangible services. Process in n o vation s are new ways
producing goods and services. I hey m ay be tech nological or organizational
(Edquist, H om m en, and M cK elvey 2001),
S Y S T E M S OF I N N O V A T I O N 183
M alerba (1997) sim ilarly focuses on a grou p o f firm s that d evelop and manufacture
the products o f a specific sector and that generate an d utilize the technologies of
that sector. The concept o f “ regional in n o vation system ” has been developed and
used by C o oke et al. (1997) and Braczyk et al. (199^)» C o o k e (20 0 1), an d Asheim and
Isaksen (2002).
The three perspectives— national, sectoral and region al— m a y be clustered as
variants o f a single generic “ system s o f in n o vatio n ” ap p ro ach (Edquist 1997b: 3)
11- 12 ). M uch o f the discussion in this chapter is relevant fo r the generic approach,
and is based on the prem ise that the different varian ts o f system s o f innovation
coexist and com plem ent each other. W hether the m ost ap p ro p riate conception of
the system o f innovation, in a certain context, sh o u ld be national, sectoral or
regional, depends to a large extent on the questions one w ants to ask.7
7. 3 S t r e n g t h s a n d W e a k n e s s e s
o f t h e SI A p p r o a c h
new ways. T h is focus distinguishes the SI approach from other approaches that
regard technological change and other innovations as exogenous.
T he S I a p p ro a c h c a n e n c o m p a s s b o th p r o d u c t a n d process in n o v a tio n s , as w e ll as
s u b ca te g o rie s o f these typ e s o f in n o v a t io n . Traditionally, in n ovation studies have, to
a large extent, focused u p o n tech nological process innovations and to som e extent
upon pro d u ct in n o vatio n s, b u t less on n on-tech nological and intangible ones, i.e.
service p ro d u ct in n o vatio n s and organizational process innovations (as specified in
Section 7.2). A s argu ed in this H an d b o ok , there are go od reasons to use a com pre-
hensive in n o vatio n co n cep t,8 an d the system s o f in n ovation approach is well suited
to this co m preh en sive perspective, since all the categories o f in n ovation s specified in
this chapter can be analyzed w ith in it. T h at n on-tech nological form s o f in n ovation
deserve m o re atten tion is also argued in O E C D (2002a: 24.d).
influence innovation processes* There is, how ever, no agreem ent about what the
term “ institutions” m eans (see Section 7.3*3)*
7. 4 T h e C o n s t i t u e n t s , F u n c t i o n ,
A c t i v i t i e s a n d B o u n d a r i e s of S i s
apply. In D en m ark and G erm any, new laws have recently transferred ow nership
from the teachers to the universities, w hile in Italy, a transfer has occurred in the
opposite d irection. M an y O E C D governm ents are currently experim enting with
changes in the ow n ersh ip o f know ledge created in universities, in the b elief (based on
little evidence— see the chapter b y M o w ery and Sam pat) that such changes will
influence the p ro p en sity to patent and accelerate the com m ercialization o f eco n o m
ically useful know ledge.
In sum m ary, there seem s to be general agreem ent that the m ain com ponents in Sis
are organizations— am o n g w h ich firm s are often considered to be the m ost im p o rt
ant ones— and institution s. H ow ever, the specific set-ups o f organizations and
institutions v a ry am o n g system s.
(1) Provision o f Research and D evelopm ent (R8cD), creating new knowledge,
prim arily in engineering, m edicine, and the natural sciences.
(2) Com petence building (provision o f education and training, creation of
hum an capital, production and reproduction o f skills, in d ivid u al learning)
m the labor force to be used in innovation and R & D activities.
(3) Form ation o f new product markets.
S Y S T E M S OF I N N O V A T I O N 191
(4) A rticu latio n o f q u ality requirem ents em anating from the dem and side with
regard to new produ cts.
(5) C reating and ch an gin g organizations needed for the developm ent o f new
fields o f in n o vation , e.g. enhancing entrepreneurship to create new firm s and
intrapreneu rship to diversify existing firm s, creating new research organ iza
tions, p o licy agencies, etc.
(6) N etw orkin g th rou gh m arkets and other m echanism s, including interactive
learning betw een different organizations (potentially) involved in the in
n ovation processes. T h is im plies integrating new know ledge elem ents d e
veloped in different spheres o f the SI and com in g from outside w ith elem ents
already available in the in n ovatin g firm s.
(7) C reating and ch an gin g institutions— e.g. IP R laws, tax laws, environm ent
and safety regulations, R8cD investm ent routines, etc— that influence in n o v
ating organ izatio n s and in n o vation processes b y pro vid in g incentives o r
obstacles to in n ovation .
(8) In cub atin g activities, e.g. p ro vid in g access to facilities, adm inistrative su p
p ort, etc. fo r new in n ovative efforts.
(9) Fin an cin g o f in n o vatio n processes and other activities that can facilitate
co m m ercialization o f know ledge and its adoption.
(10) P ro visio n o f consu ltancy services o f relevance for in n ovation processes, e.g.
tech n ology transfer, co m m ercial in form ation , and legal advice.
This list is p ro visio n al and w ill be subject to revision as o u r know ledge about
determ inan ts o f in n o vatio n processes increases. In add ition to a set o f activities that
is likely to be im p o rtan t in m ost Sis, there are activities that are v e ry im po rtan t in
som e kinds o f Sis an d less im p o rtan t in others. For exam ple, the creation o f technical
standards is critically im p o rtan t in som e (sectoral) system s, such as m obile telecom
m u n icatio n s.17
The system atic app roach to S is suggested here does not im p ly that they are or can
be consciously designed o r planned. O n the contrary, just as in n ovation processes
are evolutionary, S is evolve over tim e in a largely unplanned m anner. Even i f we
knew all the determ inants o f in n o vatio n processes in detail (w hich we certainly do
not now, and perhap s never w ill), w e w o u ld not be able to control them and design
or “ b u ild ” S is o n the basis o f this know ledge. C entralized control over Sis is
im possible and in n o vatio n p o licy can o n ly influence the spon tan eous developm ent
o f Sis to a lim ited extent.
* Innovation (in new products as well as processes) takes place m ain ly in firms and
leads to the creation o f “ structural capital* w hich is a know ledge-related asset
controlled by firm s (as opposed to * h u m an capital ); it is a m atter o f organiza
tional learning.
. Research and Development (R & D ) is carried ou t in universities and pu blic research
organizations as well as in firm s and leads to p u b licly available know ledge as well as
know ledge ow ned by firm s and other organizations, as w ell as b y individuals.
• Competence Building (e.g. training an d edu cation) w h ich o ccu rs in schools and
universities (schooling, education) as well as in firm s, and leads to the creation of
“ hum an capital.” Since hum an capital is controlled b y in d ivid u als, it is a matter of
individual learning.
A n im portant area for further research in the SI tradition concerns the relationship
am ong these three kinds o f learning, which appear to be closely related to one another.
One objective o f such studies would be to address w hat types and levels o f education
and training are m ost im portant for specific kinds o f innovations— e.g. process
innovations and product innovations, or increm ental and radical innovations.
As exem plification, I w ill now discuss tw o central learning activities— R& D and
com petence building— in som e greater detail. T h is d iscu ssion constitutes the be
ginning o f an exam ination o f the relations am o n g activities and constituents in Sis
that is continued in section 7.4.4. It also serves as a guide to and synthesis o f some of
the recent w ork on these issues w ithin the system s o f in n o vatio n literature.
countries such as the N etherlands, Sweden, and Sw itzerland, G overnm ents m ay also
provide grants to encourage research for the advancem ent o f know ledge” or grants
to obtain the know ledge needed fo r governm ent m issions such as defense or health
care. In m ost countries, b lock grants have declined and direct su p port has grow n in
im portance (O E C D 1998a: ch. 3).
In certain countries, universities fall under the responsibility of the national
government. In others, such as Germany, they are the responsibility of the regional
governments. Whatever the form of organization, a growing regional influence can
be observed in most countries. In Germany, the universities are financially very
autonomous. In the UK, financial support is provided by research councils to
individual projects selected on a competitive basis.
In m any countries, the science system also includes pu blic research institutes (or
“ national lab o rato ries” ) w h ich carry ou t the sam e type o f R & D activities as un iver
sities, as w ell as m o re applied research and technical developm ent w ork. A lthough
the relative im p o rtan ce o f universities in term s o f p erform in g R & D has increased in
most countries (see M o w e ry and Sam pat in this volu m e), pu blic research organ iza
tions rem ain im p o rtan t, These organizations m ay be linked to the universities and
included in the h igh er edu cation sector, or they m ay be independent o f them. The
largest single case in the O E C D area is the Centre N ation al de la Recherche Scien-
tifique (C N R S ) in France, w h ich receives the lion s share o f direct fu nding o f R & D in
the higher edu cation sector. T h e C N R S provides su p p o rt for projects that are
norm ally carried out in co llab oration w ith u n iversity researchers. In this regard,
the C N R S can be d e a rly d istinguished from its counterparts in G erm an y (M ax
Planck G esellschaft), Italy (C N R ) and Spain (C S IC ) (Laredo and M u star 2 0 0 1c $02).
In the U nited States, the h igh er education sector contains a large n um ber o f public
research laboratories. (O E C D 1998a: 8 3-4 ). O ther countries w ith а large institute
sector include N orw ay, Taiw an and G erm an y (e.g. M a x Planck Gesellshaft and
Fraunhofer). A n u m b er o f national governm ents have tried to change these o rgan
izations and p ro m o te their links w ith the rest o f the econ o m y and society. T h is has,
for exam ple, been d o n e in quite different w ays in France and the U K (Laredo and
M ustar 2 0 0 1c 503).
As this sh ort d iscu ssion suggests, different kinds o f pu blic organizations (such as
universities and p u b lic research institutes) can perform the sam e activity (R & D ) in
an N SI. N SIs d iffer sign ifican tly w ith regard to w hich organizations that perform
public R & D and w ith regard to the institutional rules that govern o r influence these
organizations (Lared o an d M u star 2001b: 6 -7 ).
In most NSIs, especially in low- and medium-income nations, only modest sums
are invested in R & D and most of the R & D is performed by public organizations. The
few countries that invest heavily in R & D are all rich, and much of their R & D is
carried out by private organizations. This group includes some large countries, such
as the U S A an d Jap an , b u t also som e sm all and m ediu m -sized countries, such as
Sweden, Sw itzerland, an d So u th Korea. There are also som e rich countries that do
194 CHARLES EDQUIST
rather little R & D , e.g. D enm ark and N orw ay. A s m en tion ed, a considerable part of
the R & D in m any rich countries is carried out and financed b y the private sector,
prim arily firm s (although there are also pu blic financial su p p o rt schem es to stimu
late firm s to perform R & D ). The p ro p o rtion o f all R & D p erfo rm ed in high-income
O E C D m em ber states that is financed b y firm s ranged betw een 21 per cent (Portugal)
and 72 per cent (Japan) in 1999 (O E C D 2002 b). A ck n ow led gin g such differences may
help to distinguish between different types o f N SIs.
M ost o f the R & D carried out b y private organ ization s m ay be characterized as
developm ent w ork rather than research. In n o vatio n certain ly does not depend solely
on R & D results, but requires also other actions, such as technical experimentation,
technology adoption, m arket investigations, and entrepreneurial initiative. R&D
and innovation activities are norm ally driven b y d ifferent rationales and motiv
ations— i.e. the advancem ent o f know ledge and the quest fo r profits, respectively
O ne im plication o f the com plex interface betw een “ research” an d “ innovation” is
that links between universities/public research o rgan ization s and innovating firms
are especially im portant to the perform ance o f N S Is.18 In n o vatin g firm s often need
to collaborate w ith public research organizations o r universities. Here, publicly
created institutions are im portant. G overnm ents m ay, fo r exam ple, support collab
orative centers and program s, rem ove barriers to co op eration , and facilitate the
m obility o f skilled personnel am ong different kinds o f organizations. This might
involve the creation o f institutional rules, such as those in Sw eden stating that
university professors should perform a “ third task” in ad d itio n to teaching and
doing research— i.e. interact w ith the society su rro u n d in g the university, including
firm s. H owever, such “ linkage activity” is carried o u t in different ways and to
different extents in different N SIs.
B. Competence building. The early w ork w ithin the SI ap p ro ach largely neglected
learning in the form o f education and train in g .19 H ow ever, com petence building is
increasingly considered to be an im portan t activity in system s o f innovation,
reflecting the im portance o f skilled personnel fo r m ost innovative activities
(Sm ith 2001: 8).20 But no rigorous analyses o f com petence b u ild in g have, to my
knowledge, been conducted as part o f the analysis o f in n o vatio n systems.
Nevertheless, there is a large literature on variou s aspects o f com petence building
outside the SI context. Com petence building (e.g. train in g and education) is the
sam e as enhancem ent o f hum an capital and is carried ou t largely, though not
exclusively, in schools and universities, C om petence b u ild in g also occurs in firms
(in the form o f training, learning-by-doing, learn in g-b y-u sin g, and individual
learning) often throughout w orking life,
A recent O E C D study analyzed vocational and technical education and training in
som e detail in Australia, Austria, D enm ark, England (in clu d in g W ales and Northern
Ireland), France, G eim any, Italy, the N etherlands, Q uebec, and Switzerland (OECD
1998b). This study pointed out m any differences across countries with regard
S Y S T E M S OF I N N O V A T I O N 195
These practices coexist in various countries, but their relative importance varies
considerably; frequently, one of them dominates and determines training policy.
The m odels for transition fro m education to em ploym ent also differ across coun
tries. Apprenticeship is im portan t in som e countries— e.g. in Germ any, where it caters
to about tw o-thirds o f the age group (O E C D 1996; 48). In other countries, school-
based learning and productive w o rk are com bined in alternative ways— e.g. in
Sweden, Australia, France, the U nited K ingdom , and Korea (O E C D 1996; 146).
The organizational and institutional contexts of competence building thus vary
considerably among NSIs. There are particularly significant differences between the
systems in the English-speaking countries and continental Europe. However,
scholars and policy makers lack good comparative measures on the scope and
structure of such differences. There is little systematic knowledge about the ways
in which organization of education and training influences the development, diffu
sion and use of innovations. Since labor, including skilled labor, is the least mobile
production factor, domestic systems for competence building remain among the
most enduringly “national” elements of NSIs.
* C om petition, which is an interactive process w herein the actors are rivals, and
which creates or affects the incentives for innovation.
* Transaction, which is a process by w hich goods and services, including technol
ogy-em bodied and tacit knowledge, are traded betw een econ o m ic actors.
* N etw orking, which is a process by w hich know ledge is transferred through collab
oration, cooperation and long term netw ork arrangem ents, (O E C D 2002a: 15)
W ith regard to interactions am ong organizations in their pu rsu it o f innovations,
em pirical w ork inspired by and designed on the basis o f the SI approach has been
carried out in m any countries. O ne exam ple is the C o m m u n ity In novation Surveys
(CIS) that have been coordinated by Eurostat o f the E u rop ean U n io n and carried out
in all EU countries and in several additional countries (see Sm ith , C hapter 6 in this
volum e, for a detailed discussion of C IS). The C IS results include data on collabor
ation am ong innovating organizations, and indicate that such collaboration is very
com m on and im portant. This result is supported by oth er surveys which have
shown that between 62 per cent and 97 per cent o f all p ro d u ct innovations
were achieved in collaboration between the in n o vatin g firm and some other
S Y S T E M S OF I N N O V A T I O N
organizations in A u stria, N orw ay, Spain , D enm ark, and the region o f East G othia in
Sweden (Christensen et al. 1999; O rstavik and Nas 1998; Edquist, Ericsson, and
Sjogren 2 0 0 0 : 4 7 )*
These findings constitute em p irical su p p o rt for one o f the m ain tenets o f the SI
approach, i.e. that interactive learning am o n g organizations is crucial for innovation
processes. This also illustrates the d ynam ics o f this field o f research over tim e, The
emergence o f the SI ap p ro ach in its D anish version (Lundvall 1992) took inspiration
from case studies in d icatin g that u se r-p ro d u ce r interaction was very im portan t for
innovations, e.g. in the D an ish d airy in du stry; the SI approach was form ulated
partly on this basis. O ne o f its central elem ents— the im portance o f relations o f
interactive learning am o n g organizations— has since been verified by system atic
empirical research in D en m ark and elsewhere. T h is is a good exam ple o f fruitful
interaction between theoretical and em pirical w ork.
Another exam ple o f em p irical w o rk partly based on the N S I approach is Furm an,
Porter, and Stern (2002), w h o introduce the concept o f national innovative capacity,
which is the ability o f a c o u n try to p ro d u ce and com m ercialize a flow o f new -to-the-
world technologies o ver the lo n g term . T h is concept is explicitly based u p on ideas-
driven endogenous grow th th eo ry a la R o m er (see Verspagen, Ch. 18 in this volum e),
the cluster-based ap p ro ach a la Porter (1990), and the N SI approach. O n this basis,
they estim ate the relation sh ip betw een international patenting (patenting by foreign
countries in the U SA ) an d observable m easures o f national innovative capacity.
Their results suggest that a sm all n u m b er o f observable factors describe a co u n try’s
national innovative cap acity— i.e. they identify determ inants o f the produ ction o f
new -to-the-w orld technologies. T h ey find that a great deal o f variation in patenting
across countries is due to differences in the level o f inputs devoted to innovation
(R&D m an p ow er and sp en din g). T h ey also find that an extrem ely im portan t role is
played by factors associated w ith differences in R & D productivity, e.g. po licy choices
such as the extent o f p ro tectio n o f intellectual p ro p erty and openness to inter
national trade, the share o f research p erfo rm ed b y the academ ic sector and funded
by the private sector, the degree o f technological specialization, and each in divid ual
country’s know ledge “ sto ck ” (Fu rm an et al. 2002).
The relations betw een o rgan izatio n s an d institutions are im portant for in n o v
ations and fo r the o p eratio n o f Sis. O rganizations are strongly influenced and
shaped b y in stitution s, so that o rganizations can be said to be “ em bedded in an
institutional en viron m en t o r set o f rules, w hich include the legal system , norm s,
routines, standards, etc. B u t in stitu tion s are also em bedded in and develop w ithin
organizations. E xam p les are firm -sp ecific rules w ith regard to bookkeeping or
concerning n o rm s w ith regard to the relations between m anagers and em ployees.
Hence, there is a co m p licated relationship o f m u tu al em beddedness between in sti
tutions and o rgan izatio n s (E d q u ist and Jo h n so n 1997* 59 "6 o).
Some organizations create institutions that influence other organizations.
Examples are organizations that set standards and public organizations that
Ifi C H A R L E S EDQUIST
formulate and implement those odes that we might call Innovation, policy. Sam ples
are the N M T 450 and the GSM mobile telecom standards. The NM T 4501«as created,
by the Nordic public telephone operators, which were state-owned monopolies at
the time. The development and implementation o f NM T 450 was an example oft he
importance o f user-producer relations in innovation processes, which is stressed 90
strongly in the SI approach. The public organizations provided a technical frame
work for private equipment producers and thereby decreased uncertainty,. The
Nordic equipment producers, Ericsson and Nokia, greatly benefited from tills, and
it was an important factor contributing to their leading role In mobile telecommuni
cations equipment production today. In essence, the NM T 450 provided the cradle
for the development of mobile telecommunications in Europe (Edquist 2003; 21-3).
Institutions may also be the basis for the creation o f organizations, as Mien a
government makes a law that leads to the establishment o f an organization.
Examples of such organizations include patent offices or public innovation policy
agencies.
There may also be important relations between different institutions, for example,
between patent laws and informal rules concerning exchange o f information
between firms. Institutions o f different kinds may support and reinforce each
other, but may also contradict and be in conflict with each other, as discussed in
some detail by Edquist and Johnson (1997). This work has- teen carried forward by
Coriat and Weinstein (2002), who discuss different levels o f institutions and focus
on the principle o f a hierarchy among rules themselves (Coriat and Weinstein
2002: 2B0).23
Our knowledge about the complex relations—characterized by reciprocity and
feedback—between organizations and institutions Is limited. Since the relations
between two phenomena cannot be satisfactorily analyzed if they are not conceptu
ally distinguished from each other, it is important to make a clear distinction
between organizations and institutions when specifying the concepts.24
O ne reason is the fact that the variou s case studies in N elson (1993) show that there
are sharp differences am o n g vario u s national system s in such attributes as in sti
tutional set-up, o rgan ization al set-up, investm ents in R & D , and perform ance. For
exam ple, the differences in these respects between D en m ark and Sweden are
rem arkable— in spite o f the fact that these tw o sm all countries in northern Europe
are very sim ilar in m an y other respects (Edquist and Lu ndvall 1993; 5-6 ).
A n oth er reason to focus on national Sis is that m ost public policies influencing
innovation processes o r the econ o m y as a w hole are still designed and im plem ented
at the national level. F o r very large countries, the national SI approach is less relevant
than for sm aller countries, but institutions such as laws and policies are still m ainly
national, even in a co u n try such as the U SA . In other w ords, the im portance o f
national Sis has p artly to do w ith the fact that they capture the im portance o f the
policy aspects o f inn ovation . It is not o n ly a m atter o f geographical delim itation, as
the state, and the po w er attached to it, is also im portant.
Sis m ay be supran ation al, national, or subnational (regional, local)— and at the
same tim e they m ay be sectoral w ith in any o f these geographical d em arcation s.25 A ll
these approaches m ay be fruitful, but for different pu rposes or objects o f study.
Generally, the varian ts o f the generic SI approach com plem ent rather than exclude
each other and it is useful to consider sectoral and regional S is in relation to— and
often as parts o f — n ation al ones.
There are three ways in which we can identify boundaries of Sis:
(1) spatially / geograph ically;
(2) sectorally; and
(3) in term s o f activities.
1. To define the spatial b ou n d aries is the easiest task, although it also has its
problem s. Such b ou n d aries have to be defined for regional and national Sis, and
som etim es also for sectoral ones. The problem o f geographical boun daries is so m e
what m ore com plicated for a regional than for a national SI. O ne question is w hich
criteria sh ould be used to id en tify a “ region.”
F o r a region al SI, the specification o f the bou n daries sh ou ld not only be a question
o f ch oosin g o r u sing adm inistrative b ou n daries betw een regions in a m echanical
m anner (alth ough this m igh t be useful fro m the p o in t o f view o f availability o f d ata ).
It sh ould also be a m atter o f ch oosing geographical areas for w hich the degree o f
“ coherence” o r “ inw ard o rien tation ” is high w ith regard to in n ovation processes.
O ne possible operatio n alization o f this criterion could be a sufficient level o f
localized learning sp ill-overs (am on g organizations), w hich is often associated
w ith the im p o rtan ce o f transfer o f tacit know ledge am on g (in divid u als and) o rg an
izations. A second cou ld be localized m o b ility o f skilled w orkers as carriers o f
know ledge, i.e. an o p eration alization w hich show s that the local lab or m arket is
im portant. A th ird p o ssib ility could be that a m in im u m p ro p o rtio n o f the in n o v
ation-related co llab o ration s am o n g organizations should be w ith partners w ithin
200 CHARLES EDQUIST
the region. This is a m atter o f localized netw orks, i.e. the extent to which learning
processes am ong organizations are contained w ith in regions.
For a national SI, the country's borders n o rm ally p ro vid e the boundaries. How
ever, it could be argued that the criteria fo r regional S is are as valid fo r national ones.
In other w ords, if the degree o f coherence o r inw ard o rien tatio n is very low, the
country m ight not reasonably be considered to have a n ation al SI. It was also
m entioned above that the national SI ap p ro ach is less relevant for large than for
sm aller countries. In Germ any, for exam ple, the ap p ro p riate u n it o f analysis maybe
“ Lander." The choice o f approach m ay not o n ly be a q u estio n o f size o f the country,
but also whether it is federally organized o r not.
2. Leaving the geographical dim ension, w e can also talk about “ sectorally"
delim ited Sis, i.e., systems that include o n ly a p art o f a regional, national or
international system. These are delim ited to specific tech n ological fields (generic
technologies) or product areas. The “ tech nological system s" approach belongs to
this category (although it did not initially use language associated w ith systems of
innovation) (Carlsson and Stankiewicz 1995; 49).
A ccording to Breschi and M alerba, “ a Sectoral In n o va tio n System (SIS) can be
defined as that system (group) o f firm s active in d evelo p in g and m aking a sector s
products and in generating and utilising a se c to rs tech n o lo gies" (Breschi and
M alerba 1997; 131; see also Ch. 14 by M alerba in this v o lu m e ). Specific technologies
or product areas are used to define the boun daries o f sectoral system s, but they must
also norm ally be geographically delim ited ( if they are n o t glo b al). However, it is not
self-evident what a sector is, i.e., the sectoral b ou n daries are p a rtly a theoretical— or
social— construction, which m ay reflect the specific p u rp o se o f the study. It should
also be noted here that the specification o f sectoral b o u n d aries is particularly
difficult with regard to new sectors or sectors goin g th ro u gh radical technological
shifts.
3, W ithin a geographical area (and perhaps also lim ited to a technology field or
product area), the whole socio-econ om ic system cannot, o f cou rse, be considered to
be included in the SI. The question is, then, which parts sh ou ld be included? This is a
m atter o f defining the boundaries o f Sis in term s o f activities. These have to be
defined for all kinds o f Sis: national, regional, and sectoral. T h is is m ore complicated
than in the cases o f spatial and sectoral boundaries.
Early w ork in the SI approach did not address the activities in S is in a systematic
way, and therefore failed to provide clear guidance as to w h at sh ou ld be included in a
system of innovation. N or have the boundaries o f the system s in term s o f activities
been defined in an operational w ay since then.
In Section 7.2, a system of innovation was defined as in clu d in g “ all important
econom ic, social, political, organizational, institution al, and other factors that
influence the developm ent, diffusion, and use o f in n o vation s." I f the concept of
innovations has been specified (e.g., as in the beginn in g o f Section 7.2), and if we
know the determ inants o f their developm ent, d iffu sion , and use, we will be able to
S Y S T E M S OF I N N O V A T I O N 201
7.5 R e s e a r c h G a p s a n d Opportunities
In innovation studies, there has trad ition ally been a tendency to focus m uch m ore
on technological process in n ovation s and goods produ ct innovations than on
organizational process in n ovation s and service prod u ct innovations. There are
strong reasons to use a com prehensive innovation concept and give m ore attention
to non -tech n ological and intangible kinds o f in n ovation (as proposed in Section
7.3.2). Such an o rien tation is im plicit in the fact that we talk about system s o f
innovation and not system s o f technological change.
M ore research sh ou ld be done on the activities in Sis, i.e., on the determ inants o f
the developm ent, d iffu sio n , and use o f innovations. O ne particu lar task m ay be to
revise and restructure the p relim in ary list o f im portan t activities in S is presented in
Section 7.4.3л. Su ch a list can p ro vid e an im po rtan t point o f en try for em pirical
innovation studies.
A stronger focus on activities w o u ld increase o u r know ledge of, and capacity for,
explaining in n o vatio n processes. G iven o u r lim ited system atic know ledge about
determ inants o f in n o vation s, case studies o f the determ inants o f specific innovations
o r specific (and n arrow ) categories o f in n ovation s w o u ld be very useful. In p articu
lar, I believe that comparative case studies have great potential, co m p arin g in n o v
ations system s o f v ario u s kinds as w ell as the determ inants o f in n o vation processes
w ithin them . Relevant questions to ask w ou ld include: W hich activities o f w hich
organizations are im p o rtan t fo r the developm ent, d iffu sion , o r use o f specific
innovations? Is it possible to distinguish between im p o rtan t activities and less
im p o rtan t ones? W h ich institu tion al rules influence the organizations in carryin g
out these activities? Such w o rk cou ld fu rth er develop the SI approach and contribute
to the creation o f p artial theories about relations between variables w ith in Sis. Such
theories w o u ld also im p ro ve o u r ab ility to sp ecify the bou n daries o f in n ovation
system s.
In this chapter, I have accounted fo r many o f the existing empirical studies that
claim to have been carried out w ith in a SI fram ew ork. T h e result has, on the whole,
202 CHARLES EDQUIST
been rather d isappointing in the sense that m an y o f the studies cited have not been
related to the SI approach in a p ro fo u n d way, alth o u gh there are exceptions. The SI
approach has often been used m ore as a label than as an analytical tool. It has not
influenced the em pirical studies in depth; fo r exam p le, it has not been used to
form ulate hypotheses to be confronted to em p irical o bservation s. T his has made a
virtuous fertilization between conceptual and em p irical w o rk , that is so important
to scholarly progress in this and other fields o f research, d ifficu lt to achieve. The state
o f the art o f the SI approach is partly responsible fo r this: it is often presented in too
vague and unclear a way.
C learly defined concepts are necessary in o rd er to id en tify em pirical correspond
ents to theoretical constructs and to identify the data that should be collected.
Therefore, conceptual specifications are crucial fo r em p irical studies and it is
im portant to increase the rigor and sp ecificity o f the SI approach. This can be
done by clarifying the m eaning o f key concepts such as innovation, function,
activities, com ponents, organizations, and in stitu tion s, as w ell as the relations
am ong them. M ovin g in this direction does not m ean tran sfo rm in g social science
into som ething sim ilar to natural science. Fo r exam ple, on e cannot abstract from
tim e and space, since there are no universal laws in the social sciences. It is also
Im portant to continue the w o rk o f specifying the b o u n d aries o f S is o f various kinds.
There are strong reasons to integrate conceptual and theoretical work with
em pirical studies in an effort to identify d eterm inants o f the development,
diffusion, and use o f innovations. Such integration can be expected to lead to
cross-fertilization— just as in the case o f w o rk on in teractive learning referred to
in Section 7.4A* The SI approach should be used as a conceptual framework in
specific em pirical analyses o f concrete conditions. Testable statem ents o r hypotheses
should be form ulated on the basis o f the app roach and these sh ou ld be investigated
em pirically, b y using qualitative as w ell as quantitative observation s. Theoretically
based em pirical w o rk is the best w ay to straighten up the SI app roach conceptually
and theoretically; the em pirical w ork w ill, in this way, serve as a “ d isciplining” device
in an effort to develop the conceptual and theoretical fram ew o rk. Such work would
increase o u r em pirical knowledge about relations betw een the m ain function,
activities, organizations, and institutions in Sis. T h is kn ow led ge could then be a
basis for further em pirical generalizations to develop the fram ew ork— including
theory elements. In other words, em pirically based theoretical w o rk is also very
fruitful. Independently o f where one starts, the im p o rtan t th in g is that there should
be a close relationship between theoretical and em pirical w ork.
The array o f determ inants o f innovations and the relations am o n g them can be
expected to vary over time and space, i.e. between in n o vatio n system s, as well as
am ong different categories o f innovation. For exam ple, the determ inants will
probably vary between process and product in n o vation s as w ell as between incre
m ental and radical innovations (and between subcategories o f these). It is therefore
im portant to pursue the explanatory w ork at a m eso- o r m icro -level o f aggregation.
S Y S T E M S OF I N N O V A T I O N 203
N o t e s
1. For comments on previous versions o f this chapter, I want to thank the editors o f this
Handbook, my discussant at the Lisbon workshop (John Cantwell), and my discussants at
the Roermond workshop (Jan Fagerberg, Bill Lazonick, and Rikard Stankiewicz). I have
also greatly profited from comments by other participants in the workshops and from
Pierre Bitard, Susana Borras, David Doloreux, Leif Hommen, Bjorn Johnson, Rachel
Parker, Lars Mjoset, and Annika Rickne. I also want to thank The Swedish Agency for
Innovation Systems (VINNOVA) for supporting my work with this chapter. However,
I remain responsible for the contents.
2. The regional and sectoral versions are dealt with in more detail in Ch. 11 by Asheim and
Gertler and in Ch. 14 by Malerba in this volume.
3. In this sense, this chapter is a continuation along the same trajectory as earlier attempts,
e.g. Edquist (1997b), Edquist and Johnson (1997) and Edquist (2001).
4. “ [T]he orientation o f this project has been to carefully describe and compare, and try to
understand, rather than to theorise first and then attempt to prove or calibrate the theory”
(Nelson and Rosenberg (1993: 4).
5. They mention organizations such as firms, industrial research laboratories, research
universities, and government laboratories.
6. Their definitions o f NSIs do not include, for example, consequences o f innovation—
which does not, o f course, exclude the fact that innovations, emerging in innovation
systems, have tremendously important consequences for socio-economic variables such
as productivity growth and employment. To distinguish between determinants and
consequences does not, o f course, exclude feedback mechanisms.
7. It should also be mentioned that the publications mentioned in Section 7,2 by no
means exhaust the stock o f literature addressing or using the SI approach. Edquist and
204 CHARLES EDQUIST
R eferen ces
UNIVERSITIES
IN N A T I O N A L
INNOVATION
SYSTEMS
D A V I D C. M O W E R Y
B H A V E N N. S A M P A T
8.1 In t r o d u c t io n
perhaps incorrectly) credited w ith im p ro vin g u n iv e rsity -in d u stry collaboration and
technology transfer in the U S national in n ovation system .
This chapter exam ines the roles o f universities in in d u strial-eco n o m y national
innovation system s, the com plex institutional landscapes that influence the cre
ation, developm ent, and dissem ination o f in n o vation s (fo r fu rth er discussion see
Edquist, Ch. 7 in this volum e). The inclusion o f a chapter on u n iversity research in a
volum e on innovation is itself an in n ovation — it is likely that a sim ilar handbook
published tw o decades ago w ou ld have devoted far less attention to the role of
universities in industrial in n o vatio n .1 But scholarship on the role o f universities in
the innovation process, as opposed to their role in basic research, has grow n rapidly
since 1970. One im portant them e in this research is the reconceptualization of
universities as im portant institutional actors in n ation al and regional systems of
innovation. Rather than “ ivo ry tow ers” devoted to the p u rsu it o f knowledge for its
ow n sake, a grow ing num ber o f in d u strial-econ o m y and developing-econom y
governm ents seek to use universities as instrum ents for know ledge-based economic
developm ent and change.
Governm ents have sought to increase the rate o f tran sfer o f academ ic research
advances to indu stry and to facilitate the app lication o f these research advances by
dom estic firm s since the 1970s as part o f b road er efforts to im prove national
econom ic perform ance. In the “ know ledge-based e c o n o m y ” according to this
view, national systems o f higher education can be a strategic asset, if links
w ith industry are strengthened and the transfer o f tech n o lo gy enhanced and
accelerated. M any if not m ost o f these “ tech n o logy-tran sfer” initiatives focus on
the codification o f p ro p erty rights to in d ivid u al inventions, and rarely address the
broader m atrix o f in d u stry-u n iversity relationships that span a broad range of
activities and outputs.
U niversities throughout the O E C D also have been affected b y tighter constraints
on public funding since 1970. G row th in pu blic fu n d in g fo r h igh er education has
slowed in a num ber o f O E C D m em ber states. In the U n ited States, Cohen et al
(1 998) note that federal research funding per fu ll-tim e academ ic researcher declined
by 9,4 per cent in real term s during i979“ 9L in the face o f significant upw ard pressure
on the costs o f conducting state-of-the-art research in m an y fields o f the physical
sciences and engineering. Financial su p port from state govern m en ts for US public
universities operating budgets (which o b v io u sly include m o re than research)
declined from nearly 46 per cent o f total revenues in 1980 to slightly more than
40 per cent in 1991 (Slaughter and Leslie 1997: Table 3.2), w hile the share o f federal
funds in U S public university operating budgets declined fro m 12.8 to 10 per cent
during the sam e period (the share o f operating revenues d erived from tuition and
fees rose from 12.9 to 15 per cent). The U K govern m en t reduced its institutional
f unding o f universities (as opposed to targeted, com petitive p ro gram s for research)
during the 1980s and 1990s, as did the governm ent o f A ustralia (Slaughter and
Leslie 1997).
U N I V E R S I T I E S IN N A T IO N AL INN O V AT IO N S YS T EM S 211
Faced w ith slow er grow th in overall public funding, increased com petition for
research fu n d in g, and con tin u in g cost pressures w ithin their operating budgets
during the past tw o decades, at least som e universities have becom e m ore aggressive
and “ e n tre p re n eu ria l" in seeking new sources o f funding. U niversity presidents and
vice-chancellors have p ro m oted the regional and national econom ic benefits flo w
ing from academ ic research and have sought closer links w ith indu stry as a m eans o f
expanding research support.
Both internal and external factors thus have led m any nations" universities to
prom ote stron ger linkages w ith in d u stry as a m eans o f publicizing and/or
strengthening their contribu tions to in n ovation and econom ic grow th. In som e
cases, these initiatives b u ild on lo n g histories o f collaboration between u niversity
and in d u stry researchers that reflect unique structural features o f national u niversity
system s and their in du strial environm ent. In other cases, however, these initiatives
are based on a m isu n d erstan d in g o f the roles played b y universities in national
innovation system s, as well as the factors that u nderpin their contributions to
industrial in n ovation .
A lth ough universities fu lfill b roadly sim ilar fu nctions in the in n ovation system s
o f m ost indu strial and indu strializing econom ies, the im portance o f their role varies
considerably, and is influenced b y the structure o f d om estic industry, the size and
structure o f oth er p u b licly fu nded research perform ers, and nu m erou s other factors.
Follow ing a discu ssion o f the (lim ited) evidence on the contrasting im portance o f
universities w ith in R8cD perfo rm an ce and em ploym ent in national innovation
systems, we exam in e other evidence on the contribu tions o f universities to industrial
innovation. Based on this discussion, we critically exam ine recent initiatives b y
governm ents in a n u m b er o f O E C D nations to enhance the contributions o f
universities to in n o vatio n and econom ic grow th. We conclude w ith a discussion
o f the b road agenda fo r future research.
8.2 W h a t F u n c t io n s d o U n iv e r s it ie s
P e r f o r m w i t h i n N a t io n a l
In n o v a t io n S y s t e m s ?
dem ands from students and their prospective em ployers fo r relevance in the
curriculum can strengthen links between the acad em ic research agenda and
the needs o f society.
The econom ically im portant "o u tp u ts” o f u n iversity research have come in
different form s, varyin g over tim e and across in d u stries.3 T h ey include, among
others; scientific and technological in fo rm atio n 4 (w h ich can increase the efficiency
o f applied R & D in indu stry by guiding research tow ards m o re fru itfu l departures),
equipm ent and instrum entation5 (used b y firm s in their p ro d u ctio n processes or
their research), skills or hum an capital (em b o d ied in students and faculty
m em bers), networks o f scientific and tech nological capabilities (which facilitate
the diffusion o f new know ledge), and prototypes fo r new p ro d u cts and processes.6
Universities are w idely cited as critical in stitu tion al actors in national innovation
system s (see Nelson 1993; Edquist, C h. 7 in this vo lu m e, an d n u m erou s other works).
As Edquist notes in his chapter, the precise d efin ition o f “ national innovation
system s” rem ains som ew hat hazy, but m ost o f the large literature on the topic
defines them as the institutions and actors that affect the creation, development,
and diffusion o f innovations. The literature on n ation al in n o vatio n systems empha
sizes the im portance o f strong linkages am on g these v a rio u s institution s in improv
ing national innovative and com petitive perform an ce, and this em phasis applies in
particular to universities w ithin national in n o vatio n system s.7 The “ national”
innovation system s o f the industrial econom ies app ear m o re and m o re interdepend
ent, reflecting rapid grow th during the post-1945 p erio d in cross-border flows of
capital, goods, people, and knowledge. Yet the u n iversity system s o f these economies
retain strong “ national” characteristics, reflecting sign ifican t contrasts among na
tional university system s in structure, and the influence o f historical evolution on
contem porary structure and policy.
O ne influential conceptualization o f the role o f academ ic research w ithin national
innovation systems and econom ies was the so-called “ lin ear m o d el” o f innovation
w idely associated w ith Vannevar Bush and his fam ou s “ b lu ep rin t” fo r the US post-
1945 R8cD system, Science: The Endless Frontier. B ush argu ed for expanded public
funding for basic research w ithin U S universities as a critical contributor to eco
nom ic growth, and argued that universities w ere the m ost app ro p riate institutional
locus for basic research. This “ linear m o d el” o f the in n o vatio n process asserted that
funding o f basic research was both necessary and sufficient to pro m ote innovation.
Bush s argum ent anticipated parts o f the “ m arket failu re” ration ale for the funding
o f basic academ ic research subsequently developed b y N elson (1959) and Arrow
(1962), This portrayal o f the innovation process has been w id ely criticized (see Kline
and Rosenberg 1986, for one such rebuttal o f the lin ear m o d el). M any US policy
m akers during the 1970s and 1980s cited the Japanese eco n o m y as evidence that basic
research m ay not be necessary or sufficient for a nation to im p ro ve its innovative
perform ance.
U N I V E R S I T I E S IN N A T I O N A L I N N O V A T I O N S Y S T E M S 213
Yet another view o f the role o f university research focuses on the contrasting
“ norm s o f academ ic and industrial research, M erely contrasting the “ fu ndam ental”
research activities o f academ ics w ith the applied research o f industrial scientists and
engineers obscures as m uch as it illum inates— after all, there are abundant exam ples
o f university researchers m aking im portan t contributions to technology develop
ment, as well as n u m erou s cases o f im portan t basic research advances in industrial
laboratories. Paul D avid and colleagues (D asgupta and D avid 1994; D avid, Foray,
and Steinm ueller 1999) argue that the norm s o f academ ic research differ significantly
from those observed w ithin industry. For academ ic researchers, professional recog
nition and advancem en t depend crucially on being first to disclose and publish their
result. P rom pt disclosure o f results and in m ost cases, the m ethods used to achieve
them, therefore is central to academ ic research. Industrial innovation, b y contrast,
relies m o re h eavily on secrecy and lim itation s to the disclosure o f research results.
The significance o f these “ cultural differences” for the conduct and dissem ination o f
research m ay assum e greater significance in the face o f closer links between un iver
sity and industrial researchers (see below ).
But these contrasts also can be overstated, as D avid et al. (1999) acknow ledge. The
h istory o f science is replete w ith exam ples o f fierce com petitions (“ discovery races” )
between team s o f researchers in a given field that system atically seek to m islead one
another through the d isclosure o f false inform ation. A n d recent research b y H en
derson and colleagues (H enderson, O rsenigo, and Pisano 1999; H enderson and
C ockburn 1998) o n ph arm aceu tical in d u stry R8cD highlights the increased em phasis
by a n um ber o f large p h arm aceu tical firm s on pu blication b y industrial researchers
as a m eans o f im p ro vin g their basic science capabilities. N evertheless, the potential
for clashes betw een the disclosure norm s o f academ ia and industry, and in p a rticu
lar, the poten tial risks posed b y m ore restrictive disclosure norm s for the educational
functions and the b ro ad er pace o f advance in scientific understanding, rem ains
significant.
Still another conceptual fram ew ork that has been applied recently to descriptions
o f the role o f academ ic research in “ p o st-m o d ern ” industrial societies is the “ M od e
2” concept o f research identified w ith M ichael G ib b on s and colleagues (G ib bon s
et al. 1994). “ M o d e 2” research is associated w ith a m ore interdisciplinary, pluralistic,
“ netw orked” in n o vatio n system , in contrast to the previou s system in w hich m ajo r
corporate o r academ ic research institutions w ere less closely linked w ith other
institutions. G ib b o n s and other scholars argue that the grow th o f “ M ode 2” research
reflects the increased scale and d iversity o f know ledge inputs required fo r scientific
research, a p o in t echoed in the chapter b y Pavitt in this volum e. Increased diversity
in inputs, in this view , is associated w ith greater interinstitutional collaboration and
m ore in terd iscip lin ary research. Because “ M o d e 2” involves the interaction o f m any
m ore co m m u n ities o f researchers and other actors w ithin any given research area,
21 4 D A V I D C. M O W E R Y A N D B H A V E N N . S A M P A T
purely academ ic research norm s m ay prove less in flu en tial even in such areas of
fundam ental research as biom edical research.
The “ M ode 2” fram ew ork assuredly is consistent w ith som e characteristics of
m odern innovation systems, notably the increased in terin stitu tion al collaboration
that has been rem arked upon by n um erous scholars. B u t this fram ew ork's claims
that the sources o f knowledge w ithin m odern in n o vatio n system s have become more
diverse need not im ply any decline in the role o f u niversities as fundam ental research
centers. Several studies (G od in and G ingras 2000; H icks an d H am ilton 1999; see
below for further discussion) su p port the “ M o d e 2 " assertion that cross-
institutional collaboration and diversification in know ledge sources have grown,
but indicate no such decline.
Still another conceptual framework for analyzing the changing position of uni
versities within national innovation systems is the “Triple Helix” popularized by
Etzkowitz and Leytesdorff (1997)* Like the “Mode 2” framework, the triple helix
emphasizes the increased interaction among these institutional actors in industrial
economies' innovation systems. Etzkowitz and co-authors (Etzkowitz et al. 1998)
further assert that
In addition to linkages am ong institutional spheres, each sphere takes the role o f the other.
Thus, universities assum e entrepreneurial tasks such as m arketing knowledge and creating
com panies even as firm s take on an academ ic dim ension, sharing knowledge among each
other and training at ever-higher skill levels, (p. 6)
The “ triple h elix” scholarship devotes little attention to the “ transform ations” in
industry and governm ent that are asserted to com plem ent those in universities. The
helix's em phasis on a m ore “ industrial” role for universities m a y b e valid, although it
overstates the extent to w hich these activities are o ccu rrin g th ro u gh o u t universities,
rather than in a few fields o f academ ic research. B u t the “ triple h elix ” has yet to yield
m ajor em pirical or research advances, and its value as a guide for future empirical
research appears to be lim ited.
The national systems ” “ M ode 2,” and “ triple h elix ” fram ew o rks for conceptual
izing the role o f the research university w ith in the in n o vatio n processes o f know
ledge-based econom ies em phasize the im po rtan ce o f stron g links between
universities and other institutional actors in these econom ies. A n d both “ Mode 2”
and the Triple H elix argue that interactions between universities and industry, in
particular, have grow n. According to the “ Triple H elix ” fram ew ork, increased
interactions are associated with change in the internal cu ltu re an d norm s o f univer
sities (as noted, this fram ew ork has m uch less to say about the change in the
characteristics of industrial and governm ental research institution s). What is
lacking in all o f these fram ew orks, however, is a clear set o f criteria by which to
assess the strength o f such linkages and a set o f in dicators to guide the collection of
data.
U N I V E R S I T I E S IN N A T IO N AL INN O V AT IO N S YS T EM S 215
8.3 T h e R o le o f U n iv e r s it ie s in
N a t io n a l In n o v a t io n S y s t e m s :
C r o s s -N a t io n a l D a ta
United K ingd om , Singapore, Finland, South K orea, and France. T h e share o f 24-
year-olds in the United States w ith “ first degrees” fro m universities in natural
sciences and engineering also lags well behind these and o th er n a tio n s.10
The lim ited data on the role o f national h igh er ed u catio n system s as R&D
perform ers highlight other cross-national contrasts, in clu d in g differences in their
significance w ithin the overall national R & D enterprise, th eir scale, their roles as
em ployers o f researchers, and their relationships w ith indu stry. A s Figure 8.1 shows,
the role o f universities as R & D perform ers (m easured in term s o f the share of
national R & D perform ed w ithin higher education) is greatest in Italy, the Nether
lands, and Canada, all o f which sh ow universities p e rfo rm in g m o re than 25 per cent
o f total national R & D b y 1998-2000 (Figure 8.1). T h e share o f national R&D
perform ance accounted for b y U S and Japanese universities, b y contrast, was slightly
m ore than 14 per cent during the sam e period.
-♦—USA
-•-C a n a d a
France
Germany
Italy
Netherlands
— Norway
— Sweden
UK
-•-Japan
-*-EU
C ro ss-n atio n al data high lighting differences in the “ division o f lab o r” between
universities and governm ent laboratories in basic research indicate that the higher
education sector’s share o f basic research perform ance is sim ilar in m ost W estern
European econom ies and the U nited States, although higher than in m ost o f the
Eastern E uropean and A sian countries for w hich data are available (O E C D 2001b;
A n nex Table A .6 .4 .1). B u t a key difference between the U nited States and m ost
European countries for w hich data are available is that a relatively low share o f
basic research outside the academ ic sector in the United States is perform ed b y the
governm ent, and a relatively high share b y industry. n
The data also reveal considerable variation am ong O E C D m em ber nations in the
scale o f the higher education research enterprise. A lth ough the U S higher education
system is larger in absolute term s than those o f other O E C D m em ber states, U S
universities’ perform an ce o f R & D in fact accounts for a sm aller share o f G D P than is
true o f Sw eden, France, C anad a, the N etherlands, and N o rw ay (Figure 8.2). Indeed,
Figure 8.2 indicates that U S universities’ R & D as a share o f G D P has in fact declined
slightly d u rin g the 1989-99 period. A t least a p o rtio n o f this decline reflects the rapid
growth in indu strially funded R & D perform ed w ithin U S industry, especially d urin g
the 1995-9 period.
C o m p ariso n o f the share o f “ em ployed researchers” in various n ation s’ R & D
system s that w o rk in universities reveals that the U nited States and Jap an rank
very low, reflecting the fact that a m uch higher share o f researchers in both nations
— Sweden
— France
.....N etherlands
Norway
— Japan
—*■Canada
G erm any
— EU
USA
--U K
— - -Italy
Fig. 8.2 R&D performed by the higher education sector as a percentage of GDP
S o u rc e : OECD, M a in S c ie n c e and T ech n o lo g y In d ica to rs, 2001
218 DAVID €. MOWERY AND BHAVEN N. S A MP A T
are em ployed b y in d u stry rather than h igh er ed u cation . In 1997, the last year for
w hich reasonably com plete data are available, 82.5 p e r cent o f researchers were
em ployed b y in d u stry in the United States (O E C D 2 0 0 1c Table 39), significantly
higher than in any other O E C D nation. K orea ranks secon d (68.1 per cent) and
Jap an third (64.6 per cent), w hile the overall average fo r E U countries is m uch lower
(4S.4 per cent).
Figure 8.3 depicts the share o f R8cD fu n d in g w ith in n ation al higher education
system s that is provided by industry. D espite the w id ely rem arked closeness o f US
u n iversity-in d u stry research ties and co llab oration (see R o sen b erg and Nelson 1994;
M o w ery et al. 2004), the share o f R & D in high er ed u catio n that is financed by
in du stry is higher for C anada, G erm any, and the U n ited K in g d o m than for the
U nited States in the late twentieth century.
* 4 - USA
-« -C a n a d a
France
Germany
—'"Ж..Italy
Netherlands
— Norway
—— Sweden
— UK
—4 - Japan
-*~EU
year *
O ther qualitative data from the O E C D 2002 study o f “ scien ce-in d u stry relation
ships” (2002: 37) com pare the labor m o b ility and other “ netw ork relationships”
linking universities and in d u stry for A ustria, Belgium , Finland, Germ any, Ireland,
Italy, Sw eden, the U K , the U S, and Japan. “ R8cD consulting w ith firm s by university
researchers” is greater than the EU average (the basis for these characterizations is
not pro vid ed b y the O E C D study) in Austria, Germ any, the U K , U S, and Japan; such
consulting is rated as “ lo w ” in Belgium , Finland, Ireland, and Italy. The annual flow
o f university researchers to industrial em ploym ent, another potentially im portant
channel for know ledge exchange, is significantly higher than the E U average in
Belgium , Finland, G erm any, Sweden, the U K , and the U nited States. Finally, the
“ significance o f n etw orks” linking universities and indu stry is rated as above the EU
average for Finland, G erm any, Sweden, the U K , the U S, and Japan.
Surprisingly, in view o f the frequency w ith w hich the U nited States is cited
ap p ro vin gly fo r the close links between university and industrial researchers, the
evidence that u n iversity^ in d u stry relationships are “ stronger” in the U S than
elsewhere is m ixed: the qualitative data on labor m o b ility support this characteriza
tion, w hile the data on industrial su p port o f academ ic research do not. A n im portant
gap in research on the role o f universities in national innovation system s and a
corresponding research o p p o rtu n ity is the developm ent o f better quantitative
m easures or indicators o f the scope and im portance o f this role. I f the stereotypical
view o f U S universities as m ore closely linked w ith industrial research and in n o v
ation is indeed valid (and we believe that it is), it is striking that the available
indicators shed so little light on the dim ensions o f these closer links.
A lth o u gh universities serve sim ilar fu nctions in m ost industrial econom ies, these
indicators suggest that their im portance in training scientists and engineers and in
research perfo rm an ce differs considerably am on g O E C D m em ber nations. These
differences reflect cross-nation al differences in in d u stry structure, especially the
im portance o f such “ h igh -tech n o lo gy” industries as electronics o r in form ation
tech n ology that are h igh ly research-intensive and (at least since the end o f the
C old W ar) rely h eavily on private-sector sources for R8cD finance. In addition , o f
course, the role o f n o n u n iversity pu blic research institutions differs am ong these
econom ies, and is reflected in the contrasts in universities as perform ers o f pu blicly
funded R8cD. These structu ral contrasts are the result o f a lengthy, path-dependent
process o f h istorical developm ent, in w hich institutional evolution interacts w ith
industrial grow th and change.
inasm uch as they p ro vid e little explanation for trends or cross-national differences.
N onetheless, these data highlight a broad trend o f grow th in such co-authorship,
and this area rem ains a very fruitful one for future research that spans m ore fields,
nations, and types o f pu blications. The results o f the bibliom etric w ork in this area
provide som e su p p o rt for the “ M ode 2” and “ Triple H elix” fram ew orks' argum ents
that research collaboration between universities and indu stry is grow ing throughout
the industrial econom ies, in university system s w ith very different structures
(see C h apter 7 b y E dquist in this volu m e, as well as the studies in Laredo and
M ustar 2001).
know ledge o f the fundam ental physics and ch em istry u n d erlyin g manufacturing
processes and product innovation, an area in w h ich train in g o f scientists and
engineers figured prom inently, and experim ental techniques.
The studies hy Levin et al. (1987) and C o h en et al. (2002) sum m arize industrial
R & D m anagers’ view s on the relevance to indu strial in n o va tio n o f variou s fields of
university research (Table 8.1 sum m arizes the results discu ssed in Levin et al. 1987).
V irtually all o f the fields o f university research that w ere rated as “ im portant” or
“ v ery im portant” for their innovative activities b y su rvey respondents in both
studies were related to engineering o r applied sciences. A s we noted above, these
fields o f U S university research frequently d eveloped in close collaboration with
industry. Interestingly, w ith the exception o f chem istry, v e ry few basic sciences
appear on the list o f university research fields deem ed b y in d u stry respondents to
be relevant to their innovative activities.
The absence o f fields such as physics and m athem atics in Table 8.1, however,
should not interpreted as indicating that academ ic research in these fields does not
contribute directly to technical advance in industry. Instead, these results reflect the
fact that the effects on industrial in n ovation o f basic research findings in such areas
as physics, m athem atics, and the physical sciences are realized o n ly after a consider
able lag. M oreover, application o f academ ic research results m ay require that these
advances be incorporated into the applied sciences, such as chem ical engineering,
electrical engineering and m aterial sciences. The su rvey results summarized in
Cohen et al. (2002) indicate that in m ost industries, u n iversity research results
play little i f any role in triggering new indu strial R & D projects; instead, the stimuli
originate w ith custom ers or from m an u factu rin g operation s. H ere as elsewhere,
pharm aceuticals is an exception, since u n iversity research results in this field often
trigger industrial R & D projects.
Cohen et a l (2002) further report that the results o f “ p u b lic research” performed
in governm ent labs or universities were used m ore frequently b y U S industrial firms
(on average, in 29.3 per cent o f industrial R8cD projects) than prototypes emerging
from these external sources o f research (used in an average o f 8.3 per cent o f indus
trial R & D projects). A sim ilar portrait o f the relative im p o rtan ce o f different outputs
o f university and public laboratory research em erges fro m the responses to ques
tions about the im portance to industrial R & D o f vario u s in fo rm atio n channels
(Table 8.2). A lthough pharm aceuticals once again is u n u su al in its assignment of
considerable im portance to patents and license agreem ents in vo lvin g universities
and public laboratories, respondents from this in d u stry still rated research public
ations and conferences as a m ore im portant source o f in fo rm atio n . For most indus
tries, patents and licenses involving inventions fro m u n iversity or public
laboratories were reported to be o f very little im portan ce, co m pared with publica
tions, conferences, inform al interaction with u n iversity researchers, and consulting.
Data on the use by industrial R&D managers of academic research results are
needed for other industrial economies. Nonetheless, the results of these US studies
U N I V E R S I T I E S IN N A T IO N A L I N N O V A T IO N S Y S T E M S 223
Source: Data from the Yale Survey on Appropriability and Technological Opportunity in Industry. For a
description of the survey, see Levin e t al. (1987).
consisten tly em phasize that the relationship betw een academ ic research and in d u s
trial in n o vatio n in the b io m ed ical field differs fro m that in other know ledge-
intensive sectors. In ad d ition , these studies suggest that academ ic research rarely
224 D A V I D C. M O W E R Y A N D B H A V E N N . S A M P A T
Patents 17.5
Licenses 9.5
F rom “ S c i e n c e P u s h ” to
8.5
“ T echnology C o m m er c ia liza tio n ”
As we suggested in Section 8.1, since 1980 a num ber o f industrialized countries have
im plem ented or considered policies to strengthen ""linkages” between universities
(and public research organizations) and industry, in order to enhance the con
tributions o f university-based research to innovation and econom ic performance.
U N I V E R S I T I E S IN N A T I O N A L I N N O V A T I O N S Y S T E M S 225
These initiatives all share the prem ise that universities su p port innovation in in d u s
try p rim arily through the p ro d u ction by universities o f “ deliverables” for co m m er
cialization (e.g., patented d iscoveries)>despite the m odest su pport for this prem ise
in the research discussed above. We illustrate these points in this section with case
studies o f two types o f policies: (1) policies encouraging the form ation o f regional
econom ic “ clusters” and sp in -offs based on university research, and (2) policies
attem pting to stim ulate u n iversity patenting and licensing activities.
The global d iffu sion o f these “ tech nology com m ercialization” policies illustrates a
phenom enon that has received too little attention in the literature on innovation
policy— the efforts b y p o licy m akers to “ b o rro w ” policy instrum ents from other
econom ies and ap p ly these instrum ents in a very different institutional context. As
Lundvall and B o rras p o in t out in their chapter, history, path dependence, and
institutional “ em beddedn ess” all m ake this type o f “ em ulation” very difficult.
N onetheless, such em u latio n has been especially w idespread in the field o f technol
ogy policy. In ternational p o licy em ulation o f this sort is characterized by tw o key
features: (1) the “ learn in g ” that underpins the em ulation is h igh ly selective; and (2)
the im plem entation o f p ro gram designs based on even this selective learning is
affected b y the different institutional landscape o f the em ulator.
But little evidence su pports the argum ent that the presence o f universities some
h ow “ causes” the developm ent o f regional h ig h -tech n o lo g y agglom erations. And
even less evidence supports the argum ent that the region al o r in n o vatio n policies of
governm ents are effective in creating these agglom eration s. O ne can point to high-
technology clusters w ith highly produ ctive research universities in a num ber o f areas
in the United States and other industrial econom ies; b u t there are also a number of
research universities that have not spaw ned such agglom eratio n s. M oreover, efforts
to replicate the “ Silicon Valley m o d el” in other econ om ies have proven difficult and
the results o f these efforts have been m ixed (a fascinating h istorical account o f the
efforts by Frederick Term an o f Stanford U n iversity to p ro m o te such “ exports” may
be found in Leslie and Kargon 199b).
N ation al and local governm ents in m an y O E C D cou n tries have attempted to
stim ulate the form ation o f these clusters v ia fu n d in g fo r “ science p ark s” (occasion
ally also called incubators, technology centers, o r centers o f excellence). Interest
ingly, there is considerable disagreem ent about exactly w hat a “ science park” is and
what they do; the International A ssociation o f Science Parks characterizes them as
follows:
Despite the w idespread interest in science parks, there is little evidence that they
positively affect universities' contributions to in n o vatio n or sp u r regional economic
developm ent. Using data on U S science parks, Felsenstein (1994) finds no evidence
that firm s located on university-based science parks are m o re in n ovative than other
local firm s, and Wallsten (2001) finds that science parks have little effect on regional
econom ic developm ent and rates o f innovation.
The research on science parks” in other indu strial econ om ies is also limited. One
exam ination ol science parks” in the U K (M assey et al. 1992) is dated, but presents
interesting evidence on the characteristics o f nearly 200 firm s in tw enty U K science
parks. 1 he study found that startup firm s represented 2 5 -3 0 per cent o f the tenants in
the science parks surveyed; in the absence o f som e kin d o f “ con trol population,” it is
difficult to reach conclusions about whether startup firm s are over-represented or
under-represented in these U K science parks. Perhaps m ore su rprising was the
study's finding that
formal research links between academic institutions and establishments on science parks
were no more evident than similar links with firms located off-park... Formal research links
U N I V E R S I T I E S IN N A T I O N A L I N N O V A T I O N S Y S T E M S 227
This and other evidence on the results o f governm ent policies to prom ote university-
based regional agglom erations suggests that such policies have a m ixed record o f
success. A n d even successful regional agglom erations m ay require considerable tim e
to em erge. Recent w o rk by Sturgeon (2000) argues that Silicon Valley's h istory as a
center for n ew -firm fo rm atio n and innovation dates back to the early decades o f the
twentieth century, suggesting that m uch o f the region s innovative “ culture" de
veloped over a m uch longer period o f tim e and predates the ascent to global research
em inence o f Stanford University. Sim ilarly, the N orth C arolin a “ Research Triangle,"
which w as p ro m oted m uch m ore aggressively b y the state governm ent, was estab
lished in the late 1950s and becam e a center for n ew -firm fo rm atio n and innovation
only in the late 1980s.
Still other w o rk on the developm ent o f Silicon Valley b y Leslie (1993, 2000) and
Saxenian (1988) em phasizes the m assive increase in federal defense spending after
1945 as a catalyst for the fo rm atio n o f new h igh -techn ology firm s in the region. In
this view , the presence o f leading research universities m ay have been necessary, but
was b y no m eans sufficient, to create Silicon Valley du ring the 1950s and 1960s.
Saxenian in particu lar em phasizes the v ery different structure o f British defense
procurem ent policies in explain in g the lack o f sim ilar d ynam ism in the C am b ridge
region.
The links between university research and the emergence of regional high-
technology agglomerations thus are more complex than is implied by the correlation
between the presence of high-technology firms and research universities in a
number of locales. The US experience suggests that the emergence of such agglom
erations is a matter of contingency, path-dependence, and (most importantly) the
presence of other supporting policies (intentional or otherwise) that may have
little to do with university research or the encouragement of university-industry
linkages.
The policy initiatives in the United States and other OECD economies that seek to
use university research and “science parks" to stimulate regional economic develop
ment suffer from a deficiency that is common to many of the other recent efforts to
stimulate university-industry linkages in OECD countries, i.e. a lack of attention to
supporting institutions, a focus on “success stories" with little attention to system
atic evidence on the casual effects of the policies, and a narrow focus on commercial
ization of university technologies, rather than other more economically important
outputs o f university research. These characteristics are also seen in recent efforts
elsew here w ith in the O E C D to em ulate the B a y h -D o le Act.
228 D A V I D C, M O W E R Y A N D B H A V E N N . S A M P A T
cham pion o f patentholder rights. But even before the establishm ent o f the C A F C ,
the 1980 U S Suprem e C o u rt decision in D iam on d v. C h akrab arty upheld the valid ity
o f a b road patent in the new indu stry o f biotechnology, facilitating the patenting and
licensing o f inventions in this sector.
Rather than em phasizing pu blic fu nding and relatively liberal disclosure and
dissem ination, the B a y h -D o le Act assum es that restrictions on dissem ination o f
the results o f m any R8cD projects will enhance econom ic efficiency by supporting
their com m ercialization. In m any respects, the B a y h -D o le Act is the ultim ate
expression o f faith in the “ linear m o d el” o f in n ovation — if basic research results
can be purchased b y w ou ld-be developers, com m ercial innovation will be
accelerated.
and licensing b y universities by itself indicate that u n iversity research discoveries are
being transferred to indu stry m ore efficiently o r com m ercialized m o re rapidly as
C olyvas et a l (2002) and M o w ery et al. (2001) p o in t out.
These assessm ents” o f the effects o f the B a y h -D o le A ct also fail to consider any
potentially negative effects o f the Act on U S u n iversity research o r innovation in the
broader econom y. Som e scholars have suggested that the “ com m ercialization mo
tives created b y B a y h -D o le could shift the orien tation o f u n iversity research away
from “ basic” and tow ards “ applied” research (H end erson et a l 1998), but thus far
there is little evidence o f substantial shifts since B a y h -D o le in the content of
academ ic research.
A second potentially negative effect of increased u n iversity patentin g and licens
ing is the potential weakening o f academ ic researchers' co m m itm en ts to “ open
scien ce” leading to publication delays, secrecy and w ith h o ld in g o f data and
U N I V E R S I T I E S IN N A T I O N A L I N N O V A T I O N S Y S T E M S 231
Issu e Y e a r
И Chemicals and Chemical Processes (Excluding Drugs) О Mechanical
■ Drugs and Medical Technology □ Other
■ Electronic, Optical, and Nuclear
m aterials (D asgu p ta and D avid 1994; Liebeskind 2001). In view o f the im portance
assigned b y indu strial researchers to the “ nonpatent/licensing” channels o f inter
action w ith universities in m ost indu strial sectors, it is crucially im portan t that these
channels n ot be constricted or im peded b y the intensive focus on patenting and
licensing in m an y universities. The effects o f any increased assertion by institutional
and in d ivid u al inventors o f p ro p erty rights over inputs to scientific research have
o n ly begun to receive serious scholarly attention. Patenting and restrictive licensing
o f in pu ts into fu tu re research (“ research too ls” ) could hinder dow nstream research
and p ro d u ct d evelopm ent (H eller and Eisenberg 1998; M erges and N elson 1994).
A lth o u gh there is little evidence as yet that the B a y h -D o le A ct has had significant,
negative consequences fo r academ ic research, technology transfer, and industrial
in n o vatio n in the U nited States, the data available to m o n itor any such effects are
v e ry lim ited. M oreover, such data are necessarily retrospective, and in their nature
232 D A V I D C. M O W E R Y A N D R H A V E N N . S A M P A T
8.6 C o n c l u s i o n
U niversities p lay im p o rtan t roles in the “ know ledge-based” econom ies o f m odern
industrial and indu strializing states as sources o f trained “ know ledge w orkers” and
ideas flo w in g fro m both basic and m ore applied research activities. B u t conventional
(and, perhaps, evo lu tio n ary) econom ic approaches to the analysis o f institutions are
very difficult to apply to universities, for several reasons. First, w ith the exceptions o f
the U S and B ritish u n iversity system s, inter-un iversity “ co m petition ” has been
lim ited in m ost national system s o f higher education. Inter-university com petition
was a very im p o rtan t historical influence on the evolution o f U S universities and
their links w ith in d u stry; but this aspect o f the “ selection environ m en t” is lacking in
m ost other national system s o f higher education.
Second, analyzing universities as econom ic institutions requires som e defin ition
o f the objectives pu rsued by in d ivid u al universities. Partly because universities
perform m u ltiple roles in m an y national system s and partly because the internal
structure o f m ost research universities m ore closely resem bles that o f a cooperative
organization rather than the hierarchical structure associated w ith industrial firm s,
characterizing “ the objectives o f the u n iversity” is difficult if not m eaningless. The
m odern u n iversity has its roots in the M id dle Ages, rather than the Industrial
R evolution, and its m edieval origin s continue to influence its organization and
operation. I f universities are to be conceptualized as econom ic institutions for
purposes o f analyzing their evolution, the current analytic fram ew orks available in
neo-classical o r evo lu tio n ary econom ics are insufficient.
The developm ent o f such an analytic fram ew ork is im portant, not least for under
standing the consequences for academ ic research o f governm ent policies that seek to
accelerate the transfer o f research results to industrial firm s. The intensified dem ands
from governm ents to raise the (m easurable) econom ic returns to their substantial
investm ents in academ ic research and education m akes the developm ent o f better
tools for understanding and m easuring the operations and outputs o f universities all
the m ore im portant. A s w e argued above, m any o f the current initiatives in the United
States and other industrial econom ies to enhance the econom ic returns from univer
sity research are based on a p o o r understanding o f the full spectrum o f roles fulfilled by
research universities in industrial econom ies, as well as a tendency to em phasize the
outputs o f university research that can be easily quantified.
234 D A V I D C . M O W E R Y A N D B H A V E N N . S A M P A T _________________
A lth ough the analytic fram ew orks pro vid ed b y the ‘ n atio n al in n ovation systems,”
“ M ode 2,” and “ Triple H elix” m odels o f scientific research an d innovation shed
som e light on the roles o f universities and largely agree in th eir assessm ent o f these
roles, these fram ew orks provide lim ited guidance fo r p o licy o r evaluation. More
over, these fram ew orks tend to d ow n p lay the v ery real ten sions am on g the different
roles o f research universities w ithin know ledge-based econom ies. Such tensions are
likely to intensify in the face o f pressure fro m p o licy m akers and others on
universities to accelerate their pro d u ction and tran sfer to co m m ercial interests of
tangible, m easurable research outputs.
The developm ent o f useful theoretical o r conceptual tools o r m odels for analyzing
universities as econom ic or other institutions w ith in kn ow ledge-b ased economies is
seriously ham pered by the lack o f data on the roles o f universities that enable
com parisons across tim e or across national in n o vatio n system s. Indicators that
enable longitudinal analysis o f the roles o f universities in train in g scientists and
engineers, contributing to “ public know ledge,” o r tran sferrin g inventions to indus
trial firm s are scarce if not entirely lacking fo r m ost n ation al system s o f higher
education. Few o f these indicators in corporate in fo rm a tio n on the geographic
dim ensions o f u n iversity -in d u stry interactions, despite the im p o rtan ce o f agglom
eration econom ies in the current po licy approaches o f m an y governm ents in this
area. M oreover, such indicators as do exist rarely are co m parab le across national
system s o f higher education.
The absence o f broader longitudinal and cross-n atio n ally com parable indicators
o f u n iversity-in d u stry interaction thus im pedes b oth the fo rm u latio n and the
evaluation o f policies. A nd the lack o f better indicators reflects the lack o f a stronger
analytic fram ew ork for understanding the roles o f universities w ithin national
innovation system s. Such a fram ew ork m ust ad o p t a m o re evolu tion ary, historically
grounded approach to the understanding o f the roles o f universities, especially the
influence o f the structure o f national higher edu cation system s o n these roles. As we
have argued in this chapter, m any o f the efforts b y O E C D govern m en ts to encourage
technology transfer and to increase the econom ic payoffs to investm ents in univer
sity research are ham pered b y a lack o f such und erstand in g. M ore comparative
institutional w ork on the evolution and roles o f research universities, including the
contrasting division o f lab or” am ong universities and other p u blicly supported
research institutions in both industrial and indu strializing econom ies, is an indis
pensable starting point for analysis o f the current and likely fu tu re position o f the
research university w ithin national innovation system s.
The developm ent o f better indicators o f the lu ll array o f channels through which
industries and universities interact w ithin know ledge-based econom ies represents
another im portant research opportunity. In add ition , m o re in fo rm atio n is needed
on m easures o f firm -level absorptive cap acity” and investm ents in its creation and
m aintenance how do existing firm s develop these vario u s channels o f interaction?
Flow and why are new firm s form ed to exploit u n iversity research advances, and how
U N I V E R S I T I E S IN N A T I O N A L I N N O V A T I O N S Y S T E M S 235
does this sp in o ff process v ary across tim e, geographic space, and national in n o v
ation systems? The extensive discussion o f all o f these im portant econom ic ph e
nom ena still lacks a stron g eviden tiary basis for m aking com parisons am ong the
higher education system s o f the industrialized econom ies. The current em phasis on
the countable rather than the im portan t aspects o f u n iversity -in d u stry interactions
could have un fortu n ate consequences for innovation p o licy in the industrial and
industrializing w orld.
N otes
1. Godin and Gingras (2000:273) note that “After having been left out of major government
policies centered on industrial innovation, universities seem, over the past 5 years, to
have become the object of a renewed interest among students of the system of knowledge
production”
2. Mowery and Rosenberg (1989: 154) note that the conduct of scientific research and
education within many research universities “exploits a great complementarity between
research and teaching. Under the appropriate set of circumstances, each may be per
formed better when they are done together”
3. This list draws from Rosenberg (1999), Cohen et al. (1998), and other sources.
4. David, Mowery, and Steinmueller (1992) and Nelson (1982) discuss the economic
importance of the “ informational” outputs of university research.
5. See Rosenberg s (1994) discussion of universities as a source of innovation in scientific
instruments.
6. See Rosenberg (1999).
7. Thus, Nelson's concluding chapter in his 1993 collection of studies of national innovation
systems argues that “ One important feature distinguishing countries that were sustain
ing competitive and innovative firms was education and training systems that provide
these firms with a flow of people with the requisite knowledge and skills. For industries in
which university-trained engineers and scientists were needed, this does not simply
mean that the universities provide training in these fields, but also that they consciously
train their students with an eye to industry needs” (1993: 511).
8. According to Ben-David (1971: 48), “ To create order among the turbulent crowds of
scholars and to regulate their relationships with the environing society, corporations
were established. Students and scholars were formed into corporations authorized by the
church and recognized by the secular ruler. The relationships of their corporation with
that of the townspeople, with the local ecclesiastical officials, and with the king were
carefully laid down and safeguarded by solemn oaths.... The important result of this
corporate device— which was not entirely unique to Europe but which attained a much
greater importance there than elsewhere—was that advanced studies ceased to be
conducted in isolated circles of masters and students. Masters and/or students came to
form a collective body The European student of the thirteenth century no longer went to
study with a particular master but at a particular university”
236 D A V I D C, M O W E R Y A N D B H A V E N N . S A M P A T
9** The Japanese higher education system has a large number o f private universities*
although the bulk o f these are devoted primarily to undergraduate education.
10. US “ science and engineering” degreeholders also account for a smaller share of all
advanced degrees awarded in the United States in 1999 than is true o f France, Taiwan*
and the United Kingdom, although this share in the United States exceeded those for
Finland and South Korea (National Science Foundation 2002).
11. These data must be interpreted with caution, since the definitions used by the national
statistical agencies whence they are drawn often differ. For example, in France CNRS is
classified as part o f the Higher Education Sector, whereas in Italy similar organizations
are treated as part o f the government sector. See OECD 2001k Annex 2.
12. By comparison, the share o f US university publications co-authored by industrial and
university researchers grew from 4.9 per cent in 1989 to 7.3 per cent in 1999, while this
share in Canadian university publications grew from 1.4 per cent in 1980 to 3.5 per cent in
1998.
13. A recent OECD report notes that “ Spinning off is the entrepreneurial route to commer
cializing knowledge developed by public research and as such is attracting a great deal of
attention, given the 'start-up' fever in many countries” and that governments “have a
special interest in this specific type o f industry-science linkage because it maybe one of
the factors that explain differences in performance in newfast-growing science based
industries” (OECD 2002: 41).
14. According to Katz and Ordover (1990), at least fourteen Congressional bills passed
during the 1980s focused on strengthening domestic and international protection for
intellectual property rights, and the Court o f Appeals for the Federal Circuit created in
1982 has upheld patent rights in roughly 80 per cent o f the cases argued before it, a
considerable increase from the pre-1982 rate o f 30 per cent for the Federal bench.
15. “ Regulatory reform in the United States in the early 1980s, such as the Bayh-Dole Act,
have [sic] significantly increased the contribution o f scientific institutions to innovation.
There is evidence that this is one o f the factors contributing to the pick-up of US growth
performance” (OECD 2000: 77).
R eferen ces
------ (1982), “ The Role o f Knowledge in R and D Efficiency” Quarterly Journal of Economics
97(3): 453“ 70.
------(i993)> National Innovation Systems; A Comparative Analysis. New York: Oxford Uni
versity Press.
o ecd (2000), A New Economy Paris: OECD.
------ (2001a), “ Education at a Glance. OECD Indicators 2001 Edition,” Paris: OECD.
------ (2001 b), Main Science and Technology Indicators, Paris: OECD.
------ (2001c), Science, Technology, and Industry Scoreboard, Paris: OECD.
------ (20oid), Basic Science and Technology Statistics, Paris: OECD.
------ (2002), Benchmarking Science-Industry Relationships, Paris: OECD.
R o s e n b e r g , N. (1992), “ Scientific Instrumentation and University Research,” Research Policy
21: 381-90.
------ (1999)5 “American Universities as Economic Institutions,” Mimeo.
*------ and N e l s o n , R. R. (1994), “American Universities and Technical Advance in Industry,”
Research Policy 23: 323-48.
S a x e n ia n , A. (1988), “ The Cheshire Cat’s Grin: Innovation and Regional Development in
England,” Technology Review 91: 67-75.
S l a u g h t e r , S., and L e s l i e , L. L. (1997), Academic Capitalism: Politics, Policies, and the
Entrepreneurial University, Baltimore: Johns Hopkins University Press.
S t u r g e o n , T. J. (2000), “ How Silicon Valley Came to Be,” in M. Kenney (ed.), Understanding
Silicon Valley: The Anatomy of an Entrepreneurial Region, Stanford, Calif.: Stanford
University Press, xvi, 285.
T r a jt e n b e r g , M., H e n d e r s o n , R., and J a f f e , A. B. (1997), “ University Versus Corporate
Patents: A Window on the Basicness o f Inventions,” Economics of Innovation and New
Technology 5:19-50,
T ro w , M. (1991), “American Higher Education: ‘Exceptional5 or Just Different,” in В. E.
Shafer (ed.), Is America Different? A New Look at American Exceptionalistn, Oxford: Oxford
University Press.
W a l l s t e n , S, (2001), “ The Role o f Government in Regional Technology Development: The
Effects o f Public Venture Capital and Science Parks,” Stanford University: SIEPR Working
Paper.
CHAPTER 9
FINANCE AND
INNOVATION
MARY O*SULLIVAN
9. 1 I n t r o d u c t i o n 1
em pirical research has not kept pace w ith theoretical developm ents and the evidence
that does exist, even on basic propositions, is often am biguous. Furtherm ore, in
Section 9.4 , 1 p o in t to serious lim itations associated w ith the dom inant analytical
approaches em ployed in m icro- and m acroecon om ic research on finance for an a
lyzing the d ynam ics o f econom ic change.
Intellectual exchange between evolu tion ary econom ist and financial econom ists
seems a better route to an im proved understanding o f the relationship between
finance and innovation. H ow ever, m ethodological differences, especially w ith re
spect to the im portan ce o f h istory in econom ic analysis, are m ajor obstacles to
integration between the fields. Therefore, it m ay be m ore fruitful for econom ists
o f in n ovation to collaborate w ith econom ic and business historians w ho research
the origins and evolu tion o f financial system s. In closing, I draw attention to
som e crucial questions that need to be addressed in new research on finance and
innovation.
9. 2 T h e R o l e o f F i n a n c e i n t h e
E c o n o m i c s of I n n o v a t i o n
Schum peter’s analysis o f the relationship between innovation and resource alloca-
tion, especially the allocation o f financial resources, was central to his study o f the
econom ics o f in n ovation . M an y o f the details o f Schum peter’s econom ics o f in n o v
ation w ere controversial and/or incom plete and con tem p orary research on the
subject has m ade considerable progress in refining, and expanding on, his analysis.
In particular, questions about the appropriate characterization o f the innovation
process have received considerable scrutiny. H owever, w ith a few notable exceptions,
the im plication s o f characteristics o f innovative activity for resource allocation,
especially the allocation o f financial resources, have been largely overlooked.
The Entrepreneur the Financier and Innovation. B egin n in g w ith an analysis o f the
allocation o f econom ic resources in the absence o f in n o vatio n , The Theory of
Economic Development is Schum peter's initial treatise on the econom ics o f innov
ation. Schum peter's central claim was that the com petitive capitalist economy
w ould settle into a routine that m ight be in m o tio n b u t w o u ld tend towards a
stationary or equilibrium state. Resources w o u ld flo w aro u n d the econom y, along
routinized paths, in w hat Schum peter described as the “ circu lar flow o f economic
life." T h ey w ould be fully utilized in that perpetual m o tio n an d as such could not
accum ulate into stocks (Schum peter 1996: 45).
Schum peter then posed the question o f how in n o vatio n co u ld occu r under these
conditions. H e defined innovation as the com m ercial or in d u strial application of
som ething new, such as a new product or process, a new type o f organization, a new
source o f supply or product m arket (Schum peter 1996: 66). H e em phasized three
characteristics o f the innovation process that had crucial im p licatio n s for resource
allocation.
First, innovation depended on the investm ent o f resources: “ m ajo r innovations
and also m any m inor ones entail construction o f N ew Plant (and equipm ent)— or
the rebuilding ot old plant requiring nonnegligible tim e and o u tla y ” (Schumpeter
*939 * 68). Second, innovation, as a general rule, was em b od ied in new firm s that are
founded to undertake the new com bination (Sch u m peter 19 6 4 :6 9 ; see also Schump-
eter 1996. 66). Finally, innovation was usually driven b y entrepreneurs who were
FINANCE AND INNOVATION 243
“ new m en, that is, w ho w ere not already prom inent in business circles (Schum peter
1996: 78; 1964: 70).
T h rou gh the process o f innovation, the econom y's existing productive resources
were to be com bined in new ways but h ow were new m en, operating in new firm s, to
get the control over the resources that they needed in order to innovate (Schum peter
1996; 68)? Since existing resources were already fully utilized in the circular flow,
som ehow they had to be detached from their current uses and m ade available to
entrepreneurs so that they could com bine them in new ways. W hat was the m ech
anism through w hich resources w ou ld be reallocated from existing to new uses in the
economy?
The solution that Schum peter provided to the puzzle put the financial system at
center stage. Specifically, he argued that innovation was financed through the
creation o f credit. The pu rchasing pow er required by entrepreneurs to detach
resources fro m the circu lar flo w to undertake new com binations w as generated ex
nihilo. C redit creation did not need to be backed b y an existing stock o f m on ey or
goods (Sch um peter 19 9 6 :10 6 ).
Schum peter believed that credit could be created in a variety o f ways but he gave
prom inence to the role o f the com m ercial bank in generating new purchasing pow er
and m aking it available to entrepreneurs. In entrusting an entrepreneur w ith the
productive resources o f society, “ [the banker] m akes possible the carrying out o f
new com bin ation s, authorises people, in the nam e o f society as it were, to form
them. H e is the ep h or o f the exchange econ o m y” (Schum peter 1996: 74).
The Large-Scale Enterprise, Innovation , and the Role o f Finance . O ver the course o f
his scholarly life, Schum peter dram atically altered his characterization o f in n ovation
from a process d om in ated b y new m en and new firm s to one driven b y the activities
o f large-scale indu strial enterprises. D u rin g the period o f alm ost fo rty years between
the w riting o f TED and CSD, Schum peter m oved to the U nited States. H e was struck
b y the m ajo r changes u n d erw ay in the U S econ om y in the first decades o f the
twentieth cen tu ry and he believed that they represented a shift from nineteenth-
century “ com petitive capitalism ” to tw entieth-century “ trustified capitalism .”
W hereas p revio u sly he had argued that the life and pu rpose o f new firm s faded
over tim e, in CSD he claim ed that the “ perfectly bureaucratized giant industrial
un it” had succeeded in ration alizing and routinizing the process o f Innovation to
such an extent that the large-scale enterprise had becom e “ the m ost pow erful
engine” o f econ om ic progress (Schum peter 1942: 106), A s a result, technological
progress had in creasin gly becom e the business not o f in d ivid u al entrepreneurs but
o f “ team s o f train ed specialists w h o turn out w hat is required and m ake it w ork in
predictable w ays” (Sch u m p eter 19 4 2 :13 2 ; see also Schum peter 1949* 71 )-
Sch um p eter’s revised characterization o f the innovation process led to im portant
changes in his analysis o f resource allocation in a dynam ic econ om y and, m
particular, in his analysis o f finance. Specifically, it led him to d ow n play the role
244 MARY О SULLIVAN
initiate, direct and sustain organizational learning. H ow ever, it is rare to find system
atic discussions o f the im plications o f o rgan ization al learn in g fo r the allocation of
financial resources to investm ents and returns. O ne b asic qu estio n that needs to be
addressed is whether, and to w hat extent, the process o f o rgan izatio n al learning affects
the scale o f investm ent that m ust be m ade b y an in n o vative firm . Resources have to be
invested in the learning process itself, as w ell as the im p lem en tatio n o f the results of
such learning w ithin the organization. In add ition , fin an cial resources are required to
develop o r acquire the necessary co m plem en tary assets (Teece 1986) to commercialize
innovations based on this organizational learning. H ow ever, v e ry little research
has been conducted on the im plications o f the ch aracteristics o f organizational
learning for the resource requirem ents o f d ifferent types o f enterprises.
I f the organizational character o f the in n o vation process has im plications for the
scale o f enterprise investm ent, it will also affect the o rg an izatio n and governance of
the process o f resource allocation w ithin the firm . W h o are the key decision makers
in allocating resources to innovative investm ents? O n w h at basis do they make
their investm ent decisions? H ow do they coordinate investm ent decisions across
the firm? W hat capabilities do they have for m akin g these decisions? H ow do their
incentives relate to those o f participants in the learn in g process and to organiza
tional objectives?
These questions only hint at the plethora o f issues that are raised b y a consider
ation o f the organizational d im ension o f firm -level in n o vatio n . Econom ists have
also highlighted other im portant characteristics o f the in n o vatio n process, such as
its cum ulative character and inherent uncertainty; their im p licatio n s for resource
allocation also require consideration in any com preh ensive econ om ics o f innovative
enterprise. Scholarly analysis o f these issues is, as yet, u n co m m o n . Indeed, studies of
the internal processes that com panies use to allocate fin an cial resources to any type
o f investm ent, not just innovative ones, are rare (fo r exceptions, see Bower 1970;
Burgelm an and Sayles 1986).
E m pirical studies have confirm ed the im portan ce o f cro ss-in d u stry variation in
certain basic characteristics o f innovative activity. In du stries v a ry not only in the
distribution o f innovative activity between entrants and in cu m b en ts at a point in
tim e but also over tim e as m easured b y the stability o r tu rb u len ce o f these popula
tions shares o f innovative inputs and outputs (for a su m m ary, see M alerba, Ch. 14*
this volum e). Sectoral differences in innovative activity have im p o rtan t implications
for resource allocation.
To the extent that entrants dom inate innovative activity, fo r exam ple, the ques
tion of how they get access to the resources that they require to innovate is crucial.
W hile the financing o f de novo entry is relevant here, em p irical studies have shown
that entrants are not necessarily new ventures; en try also occu rs through spin-offs
F IN A N C E AND I N N O V A T I O N 247
from established firm s as well as diversification by incum bent firm s in other indu s
tries, As a result, understanding how the resources o f existing firm s are reallocated,
by decision m akers in those firm s o r the em ployees that leave them to establish new
ventures, is also im portant.
I f the heterogeneity o f entrants" origins generates a range o f questions about
resource allocation, so too do other types o f observed variation in sectoral patterns
o f innovation. Industries disp lay im portant differences in com petitive interactions
am ong firm s. I f we focus on the relationship between entrants and incum bents, for
exam ple, w e find that som e entrants com pete directly with incum bent firm s for
custom ers w hereas others build an innovative strategy based on licensing arrange
ments or jo in t ventures. C an s, H su, and Stern contrast biotechnology, where join t
ventures and alliances between entrants and incum bents are com m on, w ith hard
disk drives, where entrants confronted incum bents in head-to-head com petition in
product m arkets (C an s, H su, and Stern 2 0 0 0 :3). These differences have im plications
for the type and am o u n t o f investm ent that needs to be m ade by innovative firms.
They affect the extent to w hich firm s are involved in different business activities,
such as p ro d u ctio n o r m arketing, and, m ore generally, the costs for different players
o f organizing them selves to innovate.
Industries differ not o n ly in the com petitive interactions am ong firm s but also in
terms o f the relationships between firm s and other industry actors. Com petitive firm s
m ay have relationships w ith their suppliers and custom ers as well as universities and
governm ents (local, regional, national, and international) that exert an im portant
influence on their allocation o f resources (M alerba, Ch. 14, this volum e). In the early
stages o f the developm ent o f the U S software industry, for exam ple, the federal
governm ent played a crucial role in supporting the education and training o f large
num bers o f software engineers (M ow ery and Langlois 1996). As a result, m any start-up
com panies were able to enter the industry as viable com petitors w ithout having access
to the resources that w ould have been necessary if they had undertaken this effort on
their ow n. Sim ilarly, in the U S biotechnology industry the citation patterns for patents
show a m ajor reliance by private com panies on public science in their innovative
activities w ith sim ilar im plications for their capacity to econom ize on the resources
that they needed to invest to becom e players (M cM illan, N arin, and Deeds 2000),
In add ition to intersectoral variation , the relationship between finance and
innovation m ay v a ry over tim e w ithin a given industry. Industries in w hich entrants
drive in n o vatio n in one era m ay change over tim e to a structure in w hich innovation
becom es m ore increm ental in character and is dom inated b y incum bent firm s, as in
the au tom ob ile industry. Cases o f the opposite trend also exist, as incum bent-
dom inated indu stries face waves o f innovation-led entry— the developm ent o f
bio tech n o logy w ith in pharm aceuticals is a good exam ple. The characteristics o f
the enterprises that require finance, the sources o f finance on w hich they rely, and the
im plications o f these finan cial arrangem ents for innovation, are likely to differ
significantly as a result o f these evo lu tio n ary changes.
248 M A R Y 0 *S U L L I V A N
W c m u st also co n sid er the p o ssib ility th at tb e d ire c tio n o f c a u sa tio n goes in the
o th e r d irectio n , fro m fin an ce to innovation» A s fa r as the d istrib u tio n o f innovative
a c tiv ity b etw een en tran ts an d in cu m b e n ts is c o n c ern e d , fo r e x a m p le , w e could ask
w h eth er in cu m b e n t firm s d o m in a te b ecau se th e y are m o re in n o v a tiv e o r because
en tran ts are to o fin a n cia lly co n strain ed to co m p e te w ith th e m . C o n versely, when
en tran ts d o m in a te is it b ecau se th ey are m o re in n o v a tiv e th a n in cu m b en ts or
b ecau se a ready, an d even excessive, a v a ila b ility o f fin a n c ia l re so u rc es allow s them
to d o so?
T h e latter q u estio n is the su b ject o f tw o related stu d ie s o f th e h a rd d isk drive
in d u s tr y In an article called “ C a p ita l M a rk e t M y o p ia ,” W illia m S ah lm an and
H o w a rd S teven so n (1985) arg u e th at v en tu re ca p italists a n d th e sto ck market
m assiv ely o ver-in vested in the in d u stry d u rin g the p e rio d fr o m 19 7 7 to 1984 with
negative co n seq u en ces fo r fin an cial retu rn s an d fo r in n o v a tio n . H o w ever, another
stu d y o f the d isk d rive in d u stry claim s that th at the d ia g n o sis o f c a p ital market
m y o p ia is su spect in lo n g -te rm p ersp ective. B y g ra ve , L a n g e, R o e d e l, a n d W u accept
that m an y players in the d isk d rive in d u stry failed b u t th e y a rg u e th at the survivors
u ltim ately en jo yed su fficien t co m m ercial success to ju s tify th e fin a n c ia l bets that
w ere m ad e o n the in d u stry (B ygrave, L an ge, R o ed el, a n d W u 2 0 0 0 :17 ) .
T h ese tw o stu d ies h in t at the p o ten tial o f an alyses o f the jo in t in flu e n c e o f finance
an d in n o v a tio n o n in d u stry evo lu tio n b u t th e y are e x c e p tio n s (see also Carpenter,
L azo n ick , an d O 'S u lliv a n 2003). T h e exten t to w h ic h la b o u r is reallo cated from
in cu m b en t firm s to en tran ts th ro u g h s p in -o ff a c tiv ity h as attra cted so m e recent
interest (see e.g. K le p p e r 2 0 0 1). To date, h ow ever, the re la tio n sh ip betw een the
a llo catio n o f fin a n c ia l reso u rces an d in d u stria l p a ttern s o f in n o v a tiv e activ ity has
b een largely ign o red .
when its significance is recognized (N elson 1993: 13), In contrast, the literature on
techno-econom ic paradigm s has long been concerned with the interaction between
innovative activity and the allocation of labou r (Freem an 1977). W ith the recent
publication o f Technological Revolutions and Financial Capital: The Dynamics of
Bubbles and Golden Ages b y C arlota Perez (2002), the relationship between finance
and tech n o-econ om ic paradigm s has also received system atic treatm ent.
Perez follow s Schu m peter in placing technological revolutions, that is, “ clusters o f
radical in n ovation fo rm in g successive and distinct revolutions that m odernize the
whole productive structure,” at the heart o f her theory. H ow ever, her concept o f a
revolution places m uch greater em phasis on the diffusion o f innovation than is
found in Sch um p eter’s w o rk .4 She also em phasizes, to a m uch greater degree than
Schum peter d id in his w o rk on business cycles, that the effects o f technological
revolutions go far beyond their econom ic im pact to include “ a tran sform ation o f the
institutions o f governance, o f society and even o f ideologies and culture” (Perez
2002: 2 4 -5 ).5
Perez’s analysis o f the interaction between the financial and productive system
is also m ore com prehensive than that o f Schum peter. W hile the latter focused
prim arily on the role o f the financial system in fu nding an initial burst o f innovative
investm ent, Perez characterizes the ways in w hich the financial system m ay be
involved in the p rodu ctive system throughout the life cycle o f technological revolu
tions. She argues that the relationship between the financial and productive sectors
alters as the eco n o m y m oves fro m one stage o f the life cycle to another.
P erezs b o o k represents an im portan t contribution to our understanding o f
finance and in n o vation . N ot surprisingly, given its originality, it suffers from certain
lim itations that w ill need to be addressed in future w ork. In particular, system atic
em pirical su p p o rt fo r som e o f the key argum ents that she m akes about the role o f
finance in fu n d in g tech nological revolution s is not pro vid ed .6
H ow ever, a sim ilar Schum peterian analysis o f finance and technology is found in
several recent em p irical articles b y B oyan Jovan o vic and his co-au th ors on the
relationship betw een the developm ent o f the stock m arket and technological revo
lutions in the U S eco n o m y fro m the late nineteenth century until the present
(Jo van o vic and G reen w ood 1999; Jo van o vic and R ousseau 2001; H ob ijn and Jo v a
novic 2001). F o r exam ple, Jo van o vic and R ousseau (2001) com pare and contrast the
im plications o f the IT revolu tion w ith that o f the “ electricity-era” technological
revolution fo r the U S stock m arket. T h ey sh ow that the transition periods from
founding to listin g on the stock m arket, from incorp oration to listing and from first
product o r process in n o vatio n to listing were shorter at the beginning and end o f the
twentieth cen tu ry than in the intervening years, and argue that the explanation for
this pattern can be fo u n d in the characteristics o f technological change (Jo van o vic
and R ousseau 2 0 0 1:3 3 6 ).
250 M A R Y o ’ S U L L I V A N ___________
9 .3 In n o v a t io n a n d t h e
E co n o m ics of F in a n c e
The United States is the country most closely associated with a professional venture
capital industry. Although individuals and families in the United States, as in most
industrial economies, long have used their private fortunes to fund new ventures, the
US venture capital “industry” began with the founding in 1946 o f American Research
and Development (ARD) for the finance o f new ventures. ARD was founded in Boston
by members o f the local investment community, as well as professors and adminis
trators from MIT. With the expansion o f federally funded research at M IT during
World War II, the principal sources o f ARD’s deal were the federally funded laborator
ies at Harvard and MIT. By far its most successful deal was its 1957 investment in a new
firm founded by Kenneth Olsen o f MIT's Lincoln Laboratory, the Digital Equipment
Corporation (DEC) (Rosegrant and Lampe 1992: 72, 110- 14).
The enormous success ofA R D 's investment in DEC— when the computer firm went
public in 1966 ARD's original investment o f $ 70,000 was valued at $37 m illio n -
generated huge interest in venture capital in the financial community. By that time,
a new breed of financiers had emerged out of Silicon Valley's high-technology enter
prises. Successful entrepreneurs like Fairchild's Eugene Kleiner and Don Valentine
reinvested the capital that they accumulated in promising local start-ups and brought
cash and technical skill, operating experience, and networks of industry contacts to the
ventures they funded. Silicon Valley's venture capitalists were heavily involved with
their ventures, advising entrepreneurs on business plans and strategies, helping find
: -investors, recruiting key managers, and serving on boards o f directors (Saxenian
252 MARY O SULLIVAN
Particularly controversial has been the com m on practice o f im pu tin g the im p o rt
ance o f liq u id ity constraints from the sensitivity o f a com pany's investm ent to its
cash flow. Som e scholars claim that “ investm ent-cash flow sensitivities provide no
evidence o f the presence o f financing constraints" thus calling into question the
valid ity o f the em pirical results reported in this literature (K aplan and Zingales 2000;
Fazzari, H ubbard, and Petersen 2000).
Theories that posited that the influence m ight go in the other direction, that the
characteristics o f enterprise investm ent m ight influence enterprise finance, were also
developed. O ne stream o f literature posited a relationship between the types o f
activities in w h ich enterprises invested and their financial behavior. A n oth er sug
gested that characteristics o f the investing enterprise, for exam ple, its stage o f
developm ent, m attered to the w ay it w as financed.
In analyses o f the influence o f enterprises' activities on their financing, R M >
investm ents attracted p articu lar attention since they were deem ed to create acute
in form ation asym m etries between corporate m anagers and financiers (Bah and
D um on tier 2001: 675; H im m elb erg and Petersen 1994; H all 2002). The m ain im p li
cation that has been d raw n fro m such analyses is that the gap between the costs ot
financing R & D investm ent fro m internal and external sources should be greater than
254 MARY O SULLIVAN
for other form s o f investment* Therefore, R & D -in te n siv e firm s sh ou ld be more
inclined than other firm s to rely on internal fu n d s to fin an ce their investments*
M oreover, financing constraints arising fro m im p erfectio n s in capital markets
should have a m uch greater im pact o n R & D , than other, investm ents (H all 2002;
C arpenter and Petersen 2002: F55). Recently, so m e sch olars have suggested that these
argum ents ap p ly not just to R & D investm ents b u t to all investm ents in high-
technology industries (see Bank o f England 2001: A n n e x, 81—5).
E m p irical analyses o f the relationship betw een finan ce an d R & D are primarily
analyses o f the links between cash flo w and R & D exp en d itu res. T h e com m on finding
o f these studies is that R & D investm ent is indeed p o sitiv ely correlated w ith cash flow
(for a su m m ary see H all 2002). H ow ever, these studies are subject to the same
m ethodological criticism s as the em pirical w o rk o n liq u id ity constraints and capital
investm ent to w hich we have already referred, and until these issues are resolved we
cannot be confident o f their findings.
Financial econom ists have also begun to analyse w h eth er the characteristics of
investing firm s m ight m atter to their fin an cin g beh avior. P articu lar attention has
been paid to a com pany's stage o f d evelopm ent an d the co n cep t o f a “ financing
grow th cycle" is n o w w idely used to characterize the challenges fo r firm s as they
evolve from new venture to going concern h ig h lig h tin g once again the extent of
inform ational asym m etry involved (Berger and U dell 1998: 622). Perhaps the most
straightforw ard im plication o f this type o f analysis is that firm s at earlier stages o f the
cycle, such as start-up com panies, are likely to have d ifficu lties raising external
finance. A s a result, they should be m ore h eavily d ep en d en t o n insider finance
than firm s at later stages o f developm ent.
Once again, however, em pirical research has n ot kept pace w ith theoretical
developm ents. The evidence that has been com piled , m oreover, does not provide
clear su p port for som e o f the m ost basic p ro p o sitio n s ad van ced in the theoretical
literature. In this regard, Berger and U dell h igh light tw o fin d in gs fro m their empir
ical analysis w hich seem particu larly surprising. T h e first is that the funds provided
by the principal ow ner are m ore im portan t as the firm gets o ld e r th an at early stages.
Second, their evidence suggests that finance fro m in siders never outw eighs that
provided b y outsiders even for the youngest firm s (B erger and U dell 1998: 625).
T h e tw o stream s o f literature on the influence o f investm ent characteristics on
enterprise finance the one on characteristics o f the investm ents being made and
the other on the characteristics o f investing firm s— have recently been brought
together in research on the financing o f sm all firm s m ak in g R & D , high-technology,
or technology-based investm ents, often referred to as tech n o logy-b ased small firms
(T B SFs). Theoretical m odels from the “ n e w " co rp o rate finan ce predict that these
firm s will be m uch m ore tightly constrained b y th eir o w n internal resources in
financing their investm ent than other firm s. A s yet, how ever, the ju r y is out on
whether even this basic proposition is b orn e out b y em p irical evidence. As a recent
report by the Bank ot England concluded: “ the evidence fro m such studies is
FINANCE AND INNOVATION 255
there is a technological reason why some industries depend more on external finance than
others. To the extent that the initial project scale, the gestation period, the cash harvest
period, and the requirement for continuing investment differ substantially between indus
tries, this is indeed plausible. (Rajan and Zingales 1998: 563)
256 MARY O’ S ULLIV A N
Econ om ists o f innovation, in turn, can learn fro m research iti finance. In part,
their having greater contact w ith financial econ o m ists m ig h t stim ulate a general
interest in finance am on g them but the specific details o f existin g research can also be
a source o f useful insights. For exam ple, the recent atten tion b y fin an cial economists
to the relationship between stages o f an enterprise s d evelo p m en t and its financing
requirem ents could be fru itfu lly in corp orated b y eco n o m ists o f innovation.
I f the o n ly b arrier to intellectual integration w as lack o f fam iliarity with each
other's w ork then it w ould be relatively easy to o vercom e. H ow ever, ignorance of
another's field o f in q u iry typically conceals deeper barriers. In add ition to theore
tical differences described above, m eth o d o logical d ifferences pose a formidable
barrier to the integration o f research o n finance and in n o vatio n .
Perhaps the m ost im portan t m eth odological obstacle to cross-fertilization be
tween the fields o f in n ovation and finance is the different p rio rities that they assign
to h isto ry in econom ic analysis. In recognition o f the im p o rta n ce o f change over
tim e in the process that they study, econom ists o f in n o va tio n o ften describe them
selves as "ev o lu tio n ary econom ists” and they typ ically assign considerable import
ance to historical research in their econ om ic analysis. In th eir em pirical work, they
tend to em phasize the im portan ce o f qualitative variatio n s across the econom y and
over tim e in innovation and technological change. T h e y are often skeptical of
theories o f innovation and econom ic d evelopm en t that are abstracted from the
historical contexts and em phasize the im p o rtan ce o f “ reasoned h isto ry ” and “ his
to ry-frien d ly m odels” (Freem an and L o u $a 2001; L azo n ick 19 9 1; M alerb a et a l 1999).
In contrast, m ost financial econom ists neglect the fact that capitalism is an
evo lu tio n ary process. E q u ilib riu m analysis o verw h elm in g ly d om in ates theoretical
research to the neglect o f the historical origin s and evo lu tio n o f econom ic behavior.
In em pirical research, quantitative analysis is the n o rm an d attention to qualitative
variation s w ithin the econom y and over tim e is m odest. It is h ard ly surprising,
therefore, that lim ited attention is paid in theoretical o r em p irical research to the
historical developm ent o f the financial system , the stru ctu ral evolution o f the
econ o m y o r change over tim e in the relationship betw een the finan cial system and
the real econom y.
It seems unlikely that such m ethodological differences can be overcom e in the
foreseeable future. A m ore p rom isin g avenue to intellectu al progress on the rela
tionship between innovation and finance, therefore, is fo r econ o m ists o f innovation
to collaborate with financial historians. E vo lu tio n ary econ o m ists have already had
direct experience o f the value o f a shared m eth o d o logical co m m i tm ent to historical
analysis in facilitating intellectual exchange th ro u gh their co llab o ration in research
on technological change w ith historians o f technology, labor, and business. Similarly
fruitful exchanges are likely to result fro m interaction w ith eco n o m ic and business
historians who study the h isto ry o f financial system s. In deed, there are already
som e prom isin g signs o f research in this d irection (see e.g. L a m o reau x and Sokoloff
2004).
FINANCE AND INNOVATION l 6l
9.5 C o nclusio n
What are the priorities in new research on finance and innovation? It is w idely
recognized that the p au city of, and flaws in, existing em pirical research on the
financial system are m ajo r barriers to econom ic analyses o f its role. The dearth o f
evidence is a p rob lem even i f w e focus on con tem p orary patterns o f financial
dem and and supply. W hen we lo o k to historical data, the holes in our understanding
o f these patterns are even m ore gaping (Levine 2003; O 'Sullivan 2004a, b; Zingales
2003).
It is true that patterns o f financial dem and and supply by enterprises, firm s, and
econom ies serve o n ly as backgrou n d evidence for m ore specific em pirical questions
about the relationship betw een finance and innovation. In an ideal w orld, scholars
who are p rim arily interested in researching that relationship could com pile these
data from readily available sources. H ow ever, the current, rather dire, state o f
em pirical research on patterns o f financial dem and and supply m ean that this option
is not available.
It is hard to see, therefore, h o w substantial progress in our understanding o f
finance and in n o vation can be m ade unless scholars w ho are interested in the subject
are w illing to contribu te to the com pilation o f these data so that basic em pirical
questions can be addressed. A t a m in im u m , we need evidence that allow s us to
identify variation and change in the historical patterns o f financing b y firm s, in d u s
tries, and nations. We also need to understand long-term trends in the supply o f
resources by different finan cial institutions. For exam ple, how im portant has the
stock m arket been as a source o f finance for corporate investm ent in different
econom ies at vario u s tim es? W hat types o f industries and firm s did it finance during
different periods? A re there m arked differences in the characteristics o f firm s and
industries fu n ded b y b on d m arkets and banks? (For the weaknesses o f existing
em pirical research on these questions, see O 'Su llivan 2004a, b.)
O nly when w e have identified these basic patterns in financial dem and and supply
can we hope to understand h ow they are related to the dynam ics o f econom ic
change. For exam ple, in the case o f a p articu lar in d u stry we m ight ask how patterns
o f financial dem and and su p p ly are related to the balance between incum bents and
entrants in in n o vative activity. F o r an econom y, the relationship between financial
patterns and changes in the structure o f the econom y, such as the shift from
m anufacturing to services, are clearly relevant. For understanding these types o f
interactions betw een finance and the dynam ics o f econom ic change, em pirical
studies that treat the d em an d for, and su p p ly of, finance as characterized prim arily
by quantity and price can o n ly take us so far. D etailed case studies are needed to
generate insights o n the qu alitative dim ensions o f these interactions that could then
be used to illustrate, enrich and perhaps confront som e o f the im plications that can
be draw n from q u an titative studies.
262 M A R Y o ’ s U L L l VAN
N o t e s
1. The author gratefully acknowledges the helpful critiques and suggestions that she received
from the participants in the various workshops organized to produce this book. She
would like to thank Jan Fagerberg» Bronwyn Hall» Bill Lazonick» David Mowery, and
Richard Nelson for the particularly detailed comments that they provided on earlier drafts
o f this chapter as well as Ron Adner and Bruce Kogut, her colleagues at INSEAD» for their
helpful suggestions.
2. The term “ enterprise finance” is used herein to refer to research on corporate and venture
finance.
3. The other major exception is Alfred D. Chandler. From a theoretical perspective» much of
his work can be seen as contributing to a similar concept o f the firm as that o f Penrose.
Indeed» in recent years» in reflecting on the general implications o f his research» Chandler
has tended to use the language o f organizational learning (see e.g. Chandler 1992).
4. This is true of most of the contemporary literature on techno-economic paradigms
(Freeman and Lou^a 2001:149)
5. In CSDy however» Schumpeter was centrally preoccupied with the social and political» as
well as the economic, implications o f changes in the characteristics o f the innovation
process.
6. Indeed, Perez explicitly portrays her book as “ a Think-piece»' the spelling out of an
interpretation, with enough illustrations to strengthen the case and stimulate discussion
(Perez 2002: xix).
/, In economics, the term information asymmetry” is used to refer to a situation in which
one economic agent has more information than another about a task that affects both
agents' welfare.
8. For general discussions o f the difference between structural or radical uncertainty and
param etric uncertainty, see Loasby 1976; CTOnscoH and Rizzo 1985; Langlois 1986; Shackle
1992.
FINANCE AND INNOVATION 263
References
C an s . J., H su , D., and Stern , S. (2000), “ When Does Start-Up Innovation Spur the Gale of
Creative Destruction?” NBER Working Paper 7851. ^
♦ H all, B. (2002), “The Financing of Research and Development,” NBER, Working Paper
8773.
H im m elb er g , C., and P etersen , B. (i994)> R&D Internal Finance. A Panel Study of Small
Firms in High-Tech Industries,” Review of Economics and Statistics 76(1): 38-51.
H o b ijn , B., and J o vano vic , B. (2001), “ The Information-Technology Revolution and the
Stock Market: Evidence ,” American Economic Review 91(5): 1203-20.
H u bba r d , G. (1998), “ Capital-Market Imperfections and Investment” Journal of Economic
Literature 36(1): 193-225.
J o vano vic , B., and G reenw ood , J. (1999)> “ The Information-Technology Revolution and
the Stock Market,” American Economic Review 89(2): 116-22.
* ---- an<3 R o u sseau , P. (2001), “ Why Wait? A Century o f Life before IPO,” American
Economic Review 91(2): 336-41.
K a pla n , S., and Z in g a les , L. (2000), “ Investment-Cash Flow Sensitivities are not Valid
Measures o f Financing Constraints,” Quarterly Journal of Economics 115: 707-12,
K l e p p e r , S. (2001), “ Employee Startups in High-Tech Industries,” Industrial and Corporate
Change 10: 639-74.
K l in e , S., and R o sen berg , N. (1986), “An Overview o f Innovation ” in R. Landau and N.
Rosenberg (eds.), The Positive Sum Strategy: Harnessing Technology for Economic Growth)
Washington, DC: National Academic Press, 275-305.
* Lam o reau x , N., and S okoloff , K. (eds.) (2004), The Financing of Innovation in Historical
Perspective) Cambridge, Mass.: M IT Press.
Lan g lo is , R. (1986), “ Rationality, Institutions and Explanation,” in id., Economics as a
Process: Essays in the New Institutional Economics, Cambridge and New York: Cambridge
University Press, 225-55.
Lazo n ick , W. (1991), Business Organization and the Myth of the Market EconomyyNew York:
Cambridge University Press.
L e v in e , R. (2003), “ More on Finance and Growth: More Finance, More Growth?” Federal
Reserve Bank o f St. Louis, July-August, 31-46.
L o asby , B. (1976), Choice, Complexity; and Ignorance: An Inquiry into Economic Theory;
Cambridge and New York: Cambridge University Press.
M a l e r b a , R, N elso n , R., O rsen igo , L., and W in t e r , S. (1999), “ ‘ History-friendly’ Models
o f Industry Evolution: The Computer Industry,” Industrial and Corporate Change 8:3-40*
M c M illa n , G. S., N a r in , R, and D eed s , D. (2000), “An Analysis o f the Critical Role of
Public Science in Innovation: The Case o f Biotechnology,” Research Policy 29(1): 1-8.
M o d ig lia n i , F. and M il l e r , M. H, (1958) “ The Cost o f Capital, Corporation Finance and the
Theory of Investment,” American Economic Review 48(3): 261-97.
M o w ery , D., and La n g lo is , R. (1996), “ Spinning off and Spinning on (?): The Federal
Government Role in the Development o f the US Computer Software Industry,” Research
Policy!5(6): 947-66.
N elson , R. (1991), “ Why do Firms Differ and how does it Matter?” Strategic Management
Journal 12: 61-74,
(2993)> National Innovation Systems: A Comparative Analysis, Oxford: Oxford Univer
sity Press.
О D risco ll , G., and Rizzo, M. (1985), The Economics of Time and Ignorance, Oxford and
New York: Basil Blackwell.
FIN AN CE AND INNOVATION 265
О S u l l iv a n , M. (2004a), The Financing Role o f the US Stock Market in the 20th Century,”
Working Paper, INSEAD.
------ (2004b), H istorical Patterns o f C orporate Finance at General Electric and Westing-
house Electric” Working Paper, INSEAD.
* P erez , C. (2002), Technological Revolutions and Financial Capital: The Dynamics of Bubbles
and Golden Ages>Cheltenham, UK and Northampton, Mass.: Edward Elgar.
R a ja n , R., and Z in g a l e s , L. (1995), “ W hat do we know about Capital Structure: Som e
Evidence from International Data,” Journal of Finance 50(5): 1421-60.
* -----------(1998b Financial Dependence and Grow th,” American Economic Review 88(3):
559-86.
R o s e g r a n t , S., and L a m p e , D. (1992), Route 128: Lessons from Bostons High-Tech Commu
nity New York: Basic Books.
R osenberg , N. (1994b Exploring the Black Box: Technology Economics, and History Cam
bridge and New York: Cambridge University Press.
Sahlm an , W. (1990b “ The Structure and Governance o f Venture-Capital Organizations,”
Journal of Financial Economics 27(2): 473-521.
* ----and S tev en so n , H. (1985), “ Capital Market M yopia” Journal of Business Venturing
1(1): 7“ 3°*
Sax en ia n , A. (1994 ) уRegional Advantage: Culture and Competition in Silicon Valley and Route
128, Cambridge, Mass, and London: Harvard University Press.
Sch u m peter , J. (1939), Business Cycles, vol. 1, New York: McGraw Hill.
------(1942), Capitalism, Socialism, and Democracyy New York and London: Harper 8c
Brothers.
------(1949b “ Economic Theory and Entrepreneurial History,” in Research Center in Entre
preneurial History, Harvard University, Change and the Entrepreneur, Cambridge, Mass.:
Harvard University Press.
------(1954 b History of Economic Analysis, Oxford: Oxford University Press.
—-— (1964), Business Cycles: A Theoretical, Historical and Statistical Analysis of the Capitalist
Process, abridged, with an introd., by Rendigs Eels, New York: McGraw-Hill.
------(1975), Capitalism, Socialism and Democracy, New York: Harper Torchbooks.
------(1996), The Theory of Economic Development>New Brunswick: Transaction Publishers.
Sh ackle , G. L. S. (1992), Epistemics and Economics: A Critique of Economic Doctrines, New
Brunswick, Transaction Publishers.
T eece , D. (1986), “ Profiting from Technological Innovation,” Research Policy 15(6): 285-305.
------P isano , G „ and S h u e n , A. (1997), “ Dynamic Capabilities and Strategic Management,”
Strategic Management Journal 18(7): 524-6.
W in t er , S. (1984), “ Schumpeterian Competition in Alternative Technological Regimes,”
Journal of Economic Behavior and Organization 5: 287-320.
W rig h t , G. (1997), “ Towards a More Historical Approach to Technological Change” The
Economic Journal 107(444): 1560-6.
* Z in g a les , L. (2000), “ In Search o f New Foundations,” Journal of Finance 55(4): 1623-53*
(2003), “ Commentary,” Federal Reserve Bank o f St. Louis, July-August, 47-52.
C H A P T E R 10
INNOVATION AND
INTELLECTUAL
DDA Dl i D T V D Т Г І І Т Г
F KUFBKl I KlvjrllS
OVE G R A N S T R A N D
l o. i I n t r o d u c t i o n 1
into som ething new, replacing the old.) These changes provided policy makers in both
developed and developing countries with new challenges.
10. 2 H i s t o r y of t h e IPR S y s t e m
The b rie f historical account below w ill focus p rim arily on patents, being in general
the m ost im portan t and representative IP R , and will be divided into eras, su m m ar
ized in Table 10 л .
Characteristics
Note:
Discerning eras, epochs or stages in a historical stream of events may be a useful sorting device but it always
involves some arbitrariness, even if good criteria are used. (Here the degrees of codification and geographical
diffusion of the patent system are used as primary criteria for distinguishing different eras.) Also, beneath the
events that surface in an era is often an undercurrent of events that lead up to a later era.
IN N O V AT IO N AND IN T E L L E C T U A L PROPERTY RIGHTS 269
linked to trade policies, a link that has been im portant as well as controversial
ever since.
An im portant event in the early diffusion o f the patent system was the passage in
1623 o f the Statute o f M o n o po lies b y the English Parliam ent, which gave a clear
recognition o f the underlying ideas and specific form o f a patent system .4 This later
came to serve as a m odel, fo r exam ple, for British colonies in N orth A m erica, which
started to adopt sim ilar patent laws in the seventeenth century. An interesting
feature o f the statute w as that although the patent granted m o n o p o ly privileges to
the true and first inventor, the invention had to be new in England. This provision
was intended to stim ulate dom estic technical progress (e.g. by attracting foreign
engineers and entrepreneurs to England) and reflected concern b y England's p o lit
ical leadership that the nation had fallen back in som e technical areas and needed to
catch up. T h e statute established a 14-year lifetim e for a patent (twice the tim e
needed for a m aster to train a generation o f apprentices). A third interesting feature
o f the statute was its explicit shift o f the granting authority from a royal ruler o r
sovereign to a govern m en t o r its bureaucracy. The governm ent was considered the
source o f patent rights, in contrast to the view s that patent rights derived from
sovereigns o r were n atural rights o f the individual. The latter view underlined the
French patent law at the tim e o f the French Revolution in 1791 and lived on in
nineteenth-century France.
A nother im p o rtan t event was the U S enactm ent o f a federal patent law in 1790.
The im portance attached to patents and in divid u al IP R s in the new ly created U SA is
clear from the fact that the A m erican C on stitu tion stated that Congress had the
power “ to p ro m o te the progress o f science and useful arts, by securing for lim ited
times to authors and inventors the exclusive right to their respective w ritings and
discoveries”.
Thom as Jefferson played a key role in the early days o f the U S patent system. As
Secretary o f State he was responsible for adm inistering the patent laws, and as a head
o f a new ly created “ Patent B o a rd ”, he personally exam ined patent applications. He
was noted fo r his o p p o sitio n to m on opolies but believed in the value o f lim ited
m onopolies fo r au thors and inventors. The new U S patent system had a slow start,
just as it had had in Venice three centuries earlier and w ou ld have in Jap an a century
later. The A ct o f 1793 m ade substantial changes, om itting the requirem ent that a
patentable invention had to be “ sufficiently useful and im p o rtan t”. The exam ination
o f applications fo r n o velty and usefulness w as replaced by m ere registration, m aking
the issuing o f patents m ore o r less a clerical matter, and the Patent B oard w as
abolished. T h e Patent A ct o f 1836 in essence reestablished the exam ination system
that had been in place until 1793 and created an executive Patent Office as a separate
bureau w ith in the D ep artm en t o f State. T h e U S Patent O ffice created by the 1836 Act
was adm inistered b y a C o m m issio n er o f Patents, appointed by the President upon
approval b y the Senate. T h e present U S system for review ing and adm inistering the
patents is largely based on the principles set forth in the Act o f 1836.
270 OVE G R A N S T R A N D
approved such laws in a referendum in 1887, m ainly because its im portant watch
industry was under pressure from foreign im itations. H owever, the 1887 law lim ited
patent protection w ithin Swiss borders to m echanical inventions, since firm s in the
em erging Swiss chem ical indu stry w anted to im itate and catch up with the m ore
advanced G erm an chem ical industry. A fter G erm any threatened Swiss chem ical
firm s with retaliatory tariffs, Sw itzerland extended its patent coverage to include
chem ical process (but not product) inventions in 1907 (see Penrose 1951 and
Kaufer 1989).
The Japanese patent system lim ited both the num ber and scope o f patent claim s.
M any Japanese firm s acquired large portfolios o f relatively narrow dom estic patents
and participated in dense patent netw orks (also in their foreign patenting* see
Granstrand 1999)* If* disputes were avoided and cross-licensing and diffusion o f
technical in fo rm atio n w ere prom oted by special features o f Japanese patent laws and
practices (see O rdover 1991). Use o f patents (both foreign and dom estic) by Japanese
firm s for technological catch -u p” purposes was facilitated by the often lax enforce
ment by W estern firm s o f their IPRs* as w ell as the lim ited attention paid to dynam ic
com petition and IP m atters b y W estern nations until the 1980s.
Nonetheless* Jap an su p po rted international harm onization efforts. In 1978 Japan
acceded to the PCT. The Japanese Patent Office ( JP O )* together w ith a sm all num ber
(around ten) o f other patent offices (in other P C T m em ber states)* was entrusted
with au th o rity to p erfo rm international searches for prior art to assess whether the
novelty criterion for patentability was met. Japan subsequently becam e active in
trilateral patent office cooperation am ong the EPO* JPO , and U SPT O (U nited States
Patent and T radem ark Office)* another vehicle for international coordination and
harm onization am o n g the industrial nations.
By 19 9 9 ,155 nations had adopted the Paris Convention* which in 1883 had been
signed b y ten. In 1994* another large step tow ards international h arm onization was
taken w ith the signing o f the U S-in spired T R IP S agreem ent (see B o x 10.2)* co n
sidered b y m ost experts to be the m ost im portant international IP R agreem ent since
the Paris C on ven tion . The T R IP S agreem ent has been criticized for favoring de
veloped nations and im p ed in g econom ic developm ent in developing ones. The least
developed nations in particular* m an y o f w hich lack the capabilities to enter a
virtuous catch-u p developm ent circle b y themselves* m ay be hindered b y the
T R IPS agreem ent, although m ost such nations have a considerable period o f tim e
to adhere to all o f the p rovision s o f T R IP S (see e.g. the collection o f articles in
M ansfield and M an sfield 2000).
D espite lo n g-stan d in g efforts to coordinate and harm onize the national patent
laws, m an y im p o rtan t differences remain* and a global patent system , w ith inter
national o r global patents* seem s far away.
Four developments in the United States led to the “pro-patent era” (see Table 10.2
for an overview). The first concerned the creation in 1982 of the Court of Appeals for
the Federal Circuit (CAFC) to hear patent appeals in lieu of the other circuit courts
of appeal.5This type of specialized court had been discussed for a long time in patent
circles.6As the complexity of patent disputes grew, the pressures within pro-patent
circles in law and industry for a specialized court of appeals mounted and finally
resulted in the creation of the CAFC. As many of its proponents had hoped, the
CAFC began to act in a pro-patent manner, in contrast to what US courts had done
previously. The validity of patents was upheld far more often (as if they were “bom
valid” ), and patent damages were increased. The effect of the C A F C s creation and its
decisions was to increase the economic value of patentholder rights.
A second factor behind the em ergence o f the p ro -p ate n t era w as linked to a change
o f attitude w ithin the A n titrust D ivisio n o f the U S D ep artm en t o f Justice in the early
1980s under Assistant A ttorn ey G eneral W illiam Baxter. S in ce the late 1930s, the
A ntitrust D ivisio n had been hostile to IP legislation an d IP licensin g, interpreting
patents as m on opolies h arm in g com petition. B axter w as in stru m en tal in shifting the
Justice D epartm ent s enforcem ent p o licy to em ph asize the role o f patents in pro-
m otin g in n ovation , em phasizing the d yn am ic benefits rath er th an the static costs,
T h is change in attitude could be traced b ack to ideas an d perspectives emerging in
the 1960s am ong econom ists, especially w ith in the em erg in g field o f law and
econom ics.7 The shift in antitrust p o licy in the early 1980s in the U SA is a good
(albeit rare) exam ple o f h ow changes in sch o larly th in k in g h ave had a direct impact
on policies.
The third stream of events contributing to the rise of the pro-patent era came from
large U S corp orations that pressed for stron ger IP p ro te ctio n and enforcement
against infringers and counterfeiters d om estically an d ab ro ad . U S industry also
pressed for a “ trade-based ap p ro ach ” to im p ro ve IP p ro tectio n b y including IP
m atters in U S trade negotiations and in the G A T T fra m e w o rk o f international trade
negotiations, resulting in a n u m ber o f “ trade related aspects o f IP R s” (TRIPS)
subjected to negotiations (see B o x 10.2). These initiatives, w h ich w ere spearheaded
by U S pharm aceuticals, entertainm ent, and electronics firm s, w ere part o f a larger
upsurge in political concern over the com petitiveness o f U S In d u stry and a growing
b elief that technology was a key asset that had to be protected . Individual US
corporations such as Texas Instrum ents and M o to ro la b ecam e aggressive litigators
against both dom estic and foreign, especially Jap an ese, in frin g ers in the mid-1980s,
M ost o f the largest awards o f dam ages in patent in frin g em en t cases, however,
occurred in litigation am ong U S firm s, A lan d m ark case in this connection was
Polaroid Corp. v. Eastman Kodak , w hich in 1991 resulted in a dam ages award to
Polaroid o f alm ost $U S9oo m illio n .8 Cases like these an d the fin an cial success of the
litigation strategy o f Texas In strum ents w ere w id ely p u b licized and drew the atten
tion o f top corporate m anagem ent to IP m atters an d the e co n o m ic value o f strong
patent po rtfolio s and w ell-conceived IP strategies ^
IN N O V A T IO N AND IN T E L L E C T U A L PROPERTY RIGHTS 275
The idea of linking IP policies to trade policies can be traced far back in history (for
instance, in the national patent era IPRs were often used in a mercantilistic way). The
acronym TRIPS refers to a US initiative of the 1 9 8 0 s that sought to link more stringent,
internationally harmonized IP policies to international trade policy. The US strategy
was to move IPR issues from the auspices of WIPO (seen by US as too weak and
narrowly focused) into the GATT Uruguay Round of multilateral trade negotiations in
which the US had more influence. The outcome was a success for the US and its allies,
but developing nations were frustrated. When the World Trade Organization (WTO)
came into being in 1 9 9 5 as a successor to GATT, the TRIPS agreement was one of its
founding components. The agreement consisted of seven parts and seventy-three
articles covering all aspects of IPRs, their enforcement and institutional arrangements.
It provided general obligations regarding national nondiscrimination and transpar
ency, it stipulated substantive minimum standards in almost all IPR areas (patents,
copyrights, trademarks etc.) plus standards for effective enforcement of IPRs (also
involving dispute settlement mechanisms at the WTO). It further set up a TRIPS
Council for monitoring the operations of the agreement. Finally, transition periods
were stipulated, giving one year for developed nations from entry into the WTO to
comply with all TRIPS requirements and eleven years for least-developed nations (i.e.
until 1 January 2 0 0 6 ) with an option to request extensions. The TRIPS agreement
implied particularly significant changes in the coverage of patents (forcing many
nations to extend patent protection to chemical, pharmaceutical, and biotechnological
inventions), requirements for protection of plant varieties, protection of computer
software and effective measures to protect trademarks and trade secrets (see e.g.
Maskus 2 0 0 0 for details).
The TRIPS agreement of 1 9 9 4 has been characterized as the most significant
international harmonization effort of IPRs in history, certainly on par with the Paris
Patent Convention of 1 8 8 3 . It also appears to become the most controversial one,
perhaps creating an anti-IP movement of much larger international proportions than
the anti-patent movement in Europe in the 1 8 5 0 s to 1 8 7 0 s. Particularly controversial
issues concern developing nations’ access to new technologies, especially drugs, and the
effects of stronger IPRs on the efforts of these nations to catch-up economically (see
e.g. Scherer 2 0 0 4 , Scherer and Watal 2002, and the chapters by Anawalt, Barton, and
Verspagen in Granstrand 2 0 0 3 ).
A fourth force beh ind the em ergence o f the pro-patent era was the U S G o vern
ment, especially the R eagan adm inistration. This “ political stream was also related
to the grow ing dom estic concern o f the 1980s for U S industrial com petitiveness,
which included the w idespread perception that a num ber o f Asian econom ies were
“ free-riding” on U S tech n o lo gy as they m ade significant inroads into U S m arkets. In
addition, U S in d u strially funded R & D spending was grow ing slow ly during the early
1980s w ith little o r no increase in patenting. M eanw hile foreign corporations.
2j 6 OYE G R A N ST R A N D
especially Japanese firm s, increased their p aten tin g in the U S A .10 O ne com ponent of
a broader po licy response to the perceived decline in U S com petitiveness was
legislative action to strengthen IP R s an d o th er incentives to invest in R & D (such
as R & D tax credits and favorable co n d itio n s fo r the creatio n o f R & D consortia), and
to encourage patenting o f the results o f fed erally fu n d ed R & D to facilitate interin-
stitutional R & D collaboration and tech n o lo gy transfer. T h e B a y h -D o le Act o f 1980
sim plified the procedures under w hich U S u n iversities co u ld patent and license the
results o f federally fu nded R & D (see Ch. 8 b y M o w e ry an d S am p at in this volume).
Year Event
1949 Patents so frequently declared invalid when litigated that Supreme Court Justice Jackson
remarks, “the only patent that is valid is one which this Court has not been able to get its
hands on". (Jungerson v. Ostby Et Barton Co.)
1952 The present (as of 2003) US Patent Law is passed. Revisions have occurred continually.
1979 US Senate and President Carter desire to strengthen domestic patent enforcement
1981 The US Justice Department revises its antitrust enforcement activity to make it easier for
patents not to violate antitrust statutes. US Supreme Court decision in the Diehrcase
leads through its USPTO interpretation to patentability of certain computer software.
1982 CAFC is established. In quick order, the court changes the validity of litigated patents
from 30°/o to 89%, thus initiating an era in which patents are of much greater interest to
industry.
1985 WIPO Harmonization conference. USITC litigations increased. The Young Report
delivered to President Reagan by the Commission on Industrial Competitiveness (headed
by Hewlett-Packard's John Young),
1986 Tl semiconductor patent litigation initiated at USITC. GATT TRIPS negotiations started,
1989 The Structural Impediments Initiative (SII) talks initiated between the USA and Japan
remove structural impediments to trade between the two nations, and include
intellectual property protection. Japan on Watch List of Special 301,
1992 US Patent Law reform report Honeywell won patent litigation against Minolta.
IN N O V A T IO N AND I N T E L L E C T U A L PROPERTY RIGHT S 2/7
1994 W o rld s in d u s trialize d n a tio n s agree to h arm o n ize aspects o f th e ir in te lle c tu a l p roperty
protection under the auspices of GATT, known as the TRIPS agreement
U S -J a p a n Patent C om m issioners’ U n d erstan d in g signed. After years of favorable court
decisions, all software is now clearly patentable.
1995 GATT-related TRIPS agreement causes USA (and other nations) to amend its patent laws
to expand the patent term to 20 years from filing date (from previous 17 years from
issuing date, thus giving mixed effects depending upon the application processing time
at the USPTO), allow inventive activity abroad to be considered by the patent office, and
permit the filing of provisional patent applications.
1998 The CAFC declares inventions of so-called business methods to be patentable (which
include e.g. financial inventions, teaching methods, and e-commercial methods) in Sta te
S tre e t Bank and Trust v. Signature Financial Group by stating that "since the 1952 Patent
A ct, business m eth o d s have been, and should have been, subject to th e sam e legal
re q u ire m e n ts for p a te n ta b ility as ap p lied to a n y o th e r process or m e th o d ". The Digital
M ille n n iu m C o p yrig h t A c t enacted.
The CAFC and the change in antitrust policies paved the way for effective
domestic enforcement of existing US IP laws. The trade-based approach to IP
legislation* however* focused primarily on international standards and enforcement
of intellectual property protection. This effort was largely successful (from the US
point of view)* in part because the US Congress created leverage for US trade
negotiators through a number of changes in US trade laws.11 However* the pro
patent era* set in motion by the actions of US corporations and policy makers*
gained ground internationally for other reasons as well. Technology-based MNCs*
not only in the USA but also in Europe and especially in Japan* shared an interest in
stronger international protection for intellectual property.
There is an ongoing debate, fueled by the bursting of the “ IT b u b b le ” as to
whether in fact a new type of economy has emerged and what characterizes such a
“New Economy”. Although much of the “ New Economy” rhetoric is now dis
credited, many scholars believe a new type of economy has emerged* albeit gradually*
in which intellectual capital has surpassed physical capital in importance. “ Intellec
tual capitalism”, then, refers to a capitalist economic system with a dominance of
intellectual capital (see Granstrand 1999). What role did the IP system and the pro-
patent era play in the emergence of “ Intellectual capitalism” ? A definitive answer to
this question is difficult at this stage, but a few observations may still be in order.
ICTs are generally recognized as a key technological contributor to the em ergence
o f “ intellectual capitalism ” as w ell as the “ N ew Econom y. It is natural to ask,
therefore, h o w im p o rtan t the IP system was for the em ergence o f I d s ; let us
consider som e w ell-k n ow n cases. The transistor w as patented at Bell Labs but
278 OVE G R A N STR A N D
licensed liberally (in p srt because o f antitrust litig atio n <md pressu re from the US
Justice D epartm ent). The subsequent em ergence o f the sem ico n d u cto r industry was
significantly spurred by pu blic procu rem ent and a la x IP regim e (M o w e ry 1996). The
sam e could be said abou t the em ergence o f In ternet u n d er the D efense Advanced
Research Projects A gency (D A R PA ). T h e softw are in d u stry also em erged under alax
IP regim e (Sam uelson 1993)- The telecom in d u stry w as larg ely operated by national
m o n opolies until the 1980s and 1990®» and IP R s played little role in the rapid advance
o f tech n ology in this industry. M obile teleph o n y also em erged u n til the late 1980s
under a lax IP regim e (G ran stran d 1999). T h e co n clu sio n seem s to be that the IP
system has not been o f m ajo r im portan ce for the em ergen ce o f IC T s (at least not in
the early stages). In fact it m ay even be argued that lax IP regim es w ere instrumental
fo r the em ergence o f several IC T industries.
The strengthening o f the IP regim e m a y have rein forced so m e features o f intellec
tual capitalism , but it appears that the p ro -p aten t era w as as m u ch a consequence of
intellectual capitalism as a cause o f it, and that it w as n o t a n ecessary condition for
the em ergence o f the industries and technologies that fostered it.
Surprisingly, fo r m uch o f the twentieth century, econom ists devoted little atten
tion to the patent system , and even less to other IPRs. From the 1960s onw ards,
however, the literature on the econom ics o f IPRs, and patents in particular, has
grown significantly, A sem inal w ork in this category is A rrow (1962) who argued
that, from society s poin t o f view, private firm s will underinvest in R & D because o f
their inability to app ropriate sufficient returns o f their R & D investm ents.12
Follow ing this view , patent protection can be justified as one o f several alternative
means, such as contracts, prizes, subsidies, and research consortia, to deal w ith this
market failure. A lternative m eans to address m arket failures in the innovation
process are also discussed by, am ong others, W right (1983) and D avid (1993)
em phasizing conditions and factors (such as uncertainty and elasticity o f research
supply) determ ining their relative advantages.
The opposite view , that capitalist econom ies m ay overinvest in R & D and in n o v
ation, also has its adherents. A n um ber o f recent theoretical contributions in in d u s
trial econom ics, grow th theory, and behavioral finance exam ine this possibility,
highlighting the ways in w hich com petitive races affect incentives for innovation and
can result in overinvestm ent in R & D . The intuition behind such m odels is that
patent races m ay result in substantial duplication o f R & D investm ents b y the
com peting firm s and lead to in d u stry-w id e levels o f R & D investm ent for which
the social returns m ay be less than the cost o f the overall investm ent.13 Such over
investment is m ost likely in races in w hich rew ards are skewed heavily to the early
finishers. In this case agents face strong incentives to accelerate both the start and the
com pletion o f R & D projects (at least until rents are dissipated).
Thus ap p ro p riab ility problem s m ay lead to underinvestm ent in innovation, e.g.
in “ w aiting gam es”, o n the one hand, or prospects o f a quick success m ay lead to
“ patent races” that result in overinvestm ent. W hich type o f gam e is actually observed
will depend on m an y circum stances. M ost o f this literature is theoretical, and few
em pirical studies have addressed these theories and m odels.
A nother im p o rtan t issue that attracts considerable interest concerns the length o f
the time for w hich an inventor should be awarded a patent. For instance, N ordhaus
(1969) argued that increasing the length o f patent protection increases the incentives
for investm ent in process in n ovation (and hence “ dynam ic efficiency” ) but at the
expense o f “ static efficien cy” (since increased protection m eans less com petition,
higher prices, and slow er d iffu sio n ). A n “ optim al patent length”, N ordh aus pointed
out, involves a tra d e -o ff between these two effects, and w ill depend on the nature ot
com petition, the price elasticity o f dem and and the R & D elasticity o f process cost
reduction.
More recent research focuses on the optimal breadth or scope of a patent (see Jaffe
2000 for an overview) as well as optimal combinations of length and breadth
(Klemperer 1990). The scope of a patent defines the range of its industrial applica
tions by delineating the set of technological designs that the claims in the issued
patent give p rotection to (i.e., exclusion o f im itators). In m ost indu strial-econom y
280 ove g r a n str a n d
patent system s, patent scope is initially determ in ed b y n ego tiatio n s between the
applicant and a patent exam iner; ultim ately, how ever, the sco p e fo r “ im portant”
patents (those deem ed b y patentholder or co m p etito rs as especially valuable) is
likely to be determ ined in the courts as a result o f private law suits initiated by the
patentholder o r others.
The scope o f a patent is far m ore d ifficult to param etrize th an its duration. Thus,
the “ o p tim al” scope for patents is a v ery co m p lex issue, as sh o w n b y Merges and
N elson (1990). The scope o f a patent affects the private as w ell as the social rates of
return fro m patented indu strial in n ovation s (just as the tim e d u ra tio n o f a patent
does), and these returns w ill v a ry am on g indu stries an d tech nologies, Accordingly, it
is difficult for legislators to design an overall “ o p tim a l” paten t system that fits equally
well everyw here. For exam ple, patent offices often issue b ro ad patents to applicants
pioneerin g in a new technological field characterized b y little patent-based "prior
art” and a great deal o f uncertainty. (Such patents are o ften regarded as being too
broad b y ex post observers.)
In o rd er to obtain a patent for an invention, the in ven to r has to disclose infor
m ation. The resulting disclosure o f patent in fo rm a tio n accelerates the diffusion of
patented technical in form ation , and m ay reduce d u p licate R8cD, indu ce substitute
technologies (through “ inventing aro u n d ” an im p o rta n t p aten t), stim ulate new
ideas, direct RScD efforts to o p p o rtu n ity-rich areas o r b ottlen eck problem s, provide
a basis fo r b en ch-m arking and com petitive intelligence, and stim u late technology
exchange and co o p eratio n .14 Thus, the d isclosure req u irem en t is one o f the key
features o f the patent system and a rationale behind it (see O rd o ver 1991). It should
be kept in m ind, however, that there are several oth er ch an nels fo r dissemination
than just disclosure o f patent in fo rm atio n , and that, in general, in fo rm atio n about
new technologies leaks out fairly q u ickly (see M an sfield 1985).
In sum m ary, IP R s, particu larly patents, play several im p o rta n t roles in innovation
system s— to encourage in n ovation and investm ent in in n o va tio n , and to encourage
d issem ination (diffusion) o f in fo rm atio n ab ou t the p rin cip les an d sources o f innov
ation throughout the econom y. H ow ever, as w e shall see, the im portan ce o f these
roles varies across sectors (industries) and cou ntries, and o ver tim e.
The p ro pen sity to utilize the patent system (instead o f alternative m eans) is the
subject o f a large literature: see e.g. Scherer (1983) and A rundel and Kabla (1998). The
Yale study b y Levin, N elson and others (Levin et aL 1987) investigated, through a
survey o f hundreds o f R8cD m anagers in the U S in m ore than a hundred industries,
sector-specific variation s in ap p ro p riab ility conditions and the role o f patents. Their
study also concluded that innovations w ould continue to appear in the absence o f
patent protection, and that in general patents were not sufficient to appropriate or
capture all benefits from innovation (once again, a significant exception to this
general finding was pharm aceuticals, where patent protection was deem ed to be
especially valuable). The Yale study was follow ed up b y an expanded international
study (the C a rn e g ie -M ello n study), which revealed substantial nation-and sector-
specific differences in the use o f patents, secrecy, lead tim es and other m eans for
appropriation o f the returns from innovation (see e.g. C ohen et a l 2003). For
instance, the latter study indicated that lead tim e and patents were the m ost
im portant ap p ro p riation m echanism s for Japanese firm s, w hile lead tim e and
secrecy were m ost im p o rtan t for U S firm s.
A com parative study o f Japanese and Swedish corporations, accounting for m ore
than 50 per cent o f industrial R8cD in Japan and over 90 per cent in Sweden
(Granstrand 1999), also shed light on this issue. Table 10.3 com pares the results from
this study w ith the results in Levin et al. (1987). As in the C arnegie-M ellon study,
Japanese firm s stand out b y assigning greater im portance to patent protection than
Atotes: '■
1 Sample of 24 large corporations. Perceptions for 1992.
2 % reported m Levin et al. (1007). Perceptions for mid-1980s, rescaled to the scale used in Granstrand (1999).
fpiTO; .Qranstrand (1999).
282 OVE G R A N S T R A N D
firms from other countries. A 1988 study of Swiss firms, however, concluded that lead
time was the most important mechanism for appropriating the returns from innov
ation, while patents ranked lowest (Harabi 1995)* One has to interpret these results
with care, though. First, the different mechanisms are to some extent complementary
(both patents and secrecy may be seen as means to create lead time by increasing speed
to market relative to competitors). Second, the attitudes of firms towards IPRs have
changed recently, following the emergence of the pro-patent era. Particularly
European industry in general was slow to adapt to this new environment.
10.4 S u m m a r y a n d C o n c l u s i o n s
Instrum ents, strategies, and policies for p rotection o f IP, ran gin g fro m trade secrets
and tradem arks to copyrigh ts and patents, have a lo n g h isto ry. F ro m this perspec
tive, TRIPS is o n ly the latest expression o f the links betw een trade and intellectual
p ro p erty protection. D espite this lo n g history, su rp risin g ly little sch o larly attention
IN NO V A TIO N AND IN T E L L E C T U A L PROPERTY RIGHTS 285
has been devoted to the study of intellectual property rights and innovation. One
impetus for research on IPRs has been a series of controversies in various national
jurisdictions over their effects. These controversies reveal many basic similarities
across nations and centuries, although they often concern different types of IPRs
(patent, copyrights, trademarks, design rights, etc.).
It seems fair to say that until recently the role o f IP R s in national, sectoral and
corporate in n o vation system s has been relatively m odest (although not w ithout
exceptions). H ow ever, a U S-originated pro-patent or p ro -IP era has em erged since
the 1980s, perhaps as a consequence rather than a cause o f a m uch broader and
gradual transition into a new type o f capitalist econom y that is based m ore heavily
on know ledge (and in fo rm atio n ), innovation and intellectual capital. The conse
quences o f the p ro -IP era have been far-reaching and seem likely to broaden, since
there are no signs o f a reversal o f these global trends. IPR s have been applied to m ore
form s o f intellectual p ro p erty through extensions o f old and creation o f new types o f
rights; they have becom e m ore econom ically valuable and have assum ed a m ore
strategic role in national, sectoral and corporate innovation systems.
M ore research on the com plex relations between IP and innovation is clearly
needed, especially since the em ergence o f the pro-patent era has led to significant
change and created new challenges (e.g. the so-called “ an ticom m on s” problem that
stems from a pro liferatio n o f interdependent IPR s: see H eller and Eisenberg 1998 for
a pioneering w o rk ). Future research in this area needs to focus o n the interplay
between econom ic, technical, and legal dynam ics and, in a better w ay than before,
link theoretical w o rk w ith em pirical longitudinal studies.
N o t e s
1. Helpful comments from the editors and participants in the TEARI-project and the
assistance o f Thomas Ewing are gratefully acknowledged.
2. Breeding rights refer to exclusive time limited rights to commercialize certain cultivated
plant or animal varieties. Database rights refer to exclusive time limited rights to com
mercialize certain collections o f data or content material, including literary, artistic, and
musical works or collections o f material such as texts, sound, images, numbers, and facts,
However, the term “ database” should in this context not be taken to extend to computer
programs used in the making or operation o f a database. Database rights exist in Europe
since the mid-1990s (although with some older precedents) but not in the US, and has so
far created a great deal o f controversies.
3. The history o f the term “ patent” is interesting in this context, in English the term was
short for “ letters patent,” in turn deriving from “ litterae p a te n tesw h ich in medieval
Europe referred to sealed but open royal letters, granting the holders certain rights,
privileges, tides, or offices. The term derives from Latin patere>meaning “ to be open.”
4. In fact, the patent monopoly rights became an exemption in the Statute o f Monopolies,
which generally limited monopoly privileges. The handing out o f such privileges by the
286 OVE G R A N S T R A N D
royalty had degenerated, and the English Parliament wanted to put an end to it,
apparently recognizing the exceptional importance o f technical progress.
5. See further the Federal Courts Im provem ent Act o f 19S2 and Dreyfuss (1989)*
6. A proposal can e.g. be found in recommendations o f the US Senate Committee TNEC
from the 1940s, see Folk (1942: 281-95).
7. Prof. William Baxter, personal communications.
8. The total cost to Kodak was much more, however, since Kodak paid damages to
customers and legal fees, had to shut down a plant and fire about 700 people, and lost
investments and goodwill (see Granstrand 1999 &ftd Rivette and Klein 1999)* The original
Polaroid damages claim was over $US 5 billion, which if awarded could bankrupt even a
large corporation such as Eastman Kodak.
9. A number of law suits against (alleged) Japanese infringers were brought by US firms,
and a number of out-of-court settlements also were reached among these parties.
Royalty rates for licenses were also raised. In general, these events signified the outbreak
of the so-called patent war between USA and Japan. (See Warchofsky (1994) and
Granstrand (1999).) The increased management attention paid to patenting also con
tributed to the surge of patenting in the US (see Kortum and Lerner 1999), as well as in
several other developed nations,
10. In fact, the share o f foreigners’ patenting in the USA rose from 22 per cent in 1967 to 40
per cent in 1980 (Evenson in Griliches 1984, p. 92).
11. Important new trade legislation included the Trade and Tariff Act of 1984, which
included Section 301 (authorizing US government to take retaliatory action against
countries judged to give an inadequate IP protection) and Section 501 (authorizing the
President to judge the adequacy of IP protection in granting tariff preferences to a
country), combining to a stick and carrot approach. The Omnibus Trade and Competi
tiveness Act of 1988 moved further along these lines with a “ Special 301” that required the
US Trade Representative to watch, identify and investigate foreign states denying
adequate IP protection to US firms.
12. Empirical studies have followed this up by collecting data on imitation costs and times in
relation to innovation costs and times for a number o f sectors. Patents do increase
imitation costs, especially in pharmaceuticals, but apart from that industry, patents have
not been essential for the rate o f innovations, at least not before the pro-patent era. (See
in particular Mansfield et al. 1981. See also below.)
13. Central references in this literature include Scherer (1966,1967), Barzel (1968), Dasgupta
and Stiglitz (1980), Fudenberg et al. (1983), and Aghion and Howitt (1992). For over
views, see Baldwin and Scott (1987), Tirole (1988), Romer (1996) and Aghion and Howitt
(1998).
14. There is in fact a whole service industry emerging around the processing of patent
information for various purposes. Japan is a nation that has significantly fostered and
benefitted from patent information analysis through methods subsumed under “patent
mapping , Disclosure of patent information imposes a disadvantage upon patentees but
its perceived advantages to patentees were found to be significantly higher in Japanese
firms (see Granstrand 1999).
15. Cross-national differences m ight be larger than in dustry differences, however, as was the
case between Japan and Sweden in G ranstrand (1999). One has also to keep in mind that
firms are diversified and m ore so regarding their patent p o rtfo lio than their product
portfolio (see Pavitt 1999).
IN N O V A T IO N AND IN T E L L E C T U A L PROPERTY RIGHTS 28 7
R eferen ces
C o h en , W. M., G oto , A., A k iy a , A., N elso n , R. R., and W a l s h , J, R (2003), "RM > Infor
mation Flows and Patenting in Japan and the United States, in Granstrand (200311123—54)•
D asg u pta , P, and S t ig l it z , }. (1980), “ Industrial Structure and the Nature of Innovative
Activity” The Economic Journal 90: 265-93.
D a v id , P. A. (1993), "Intellectual Property Institutions and the Panda’s Thumb: Patents,
Copyrights, and Trade Secrets in Economic Theory and History,” in M. B. Wallerstein, M.
E. Mogee, and R. A. Schoen, (eds.), Global Dimensions of Intellectual Property Rights in
Science and Technology, Washington, DC: National Academy Press, 19-61.
D r e y f u ss , R. C. (1989), "The Federal Circuit: A Case Study in Specialized Courts,” New York
University Law Review 64(1): 1-77.
D u tto n , H. I. (1984), The Patent System and Inventive Activity during the Industrial Revolu
tion 1750-1852, Manchester: Manchester University Press.
E v en so n , R. E. (1984), "International Invention: Implications for Technology Market An
alysis,” in Griliches (1984: 89-126).
F olk , G. E, (1942), Patents and Industrial Progress: A Summary, Analysis and Evaluation of the
Record on Patents of the Temporary National Economic Committee, New York: Harper 8c
Brothers.
F u d e n b e r g , D., G il b e r t , R., S t i g l i t z , J., and T iro le , J. (1983), "Preemption, Leapfrogging,
and Competition in Patent Races,” European Economic Review 22: 3-31.
G ra n str a n d , O. (1982), Technology, M anagem ent an d M arkets: A n Investigation ofR&Dand
Innovation in Industrial Organizations, London: Pinter.
* ---- (1999), The Economics and Management of Intellectual Property, Cheltenham: Edward
Elgar.
* ---- (ed.) (2003), Economics, Law and Intellectual Property, Dordrecht: K luw er Academic
Publishers.
------ (2004), “ The Economics and Management o f Technology Trade— Towards a Pro
licensing Era?” International Journal of Technology Management 27(2/3): 209—40.
------ and S j o l a n d e r , S. (1990), "The Acquisition o f Technology and Small Firms by Large
Firms,” Journal of Economic Behavior and Organization 13: 367-86.
G r il ic h e s , Z. (1984), R&D, Patents, and Productivity, Chicago: University o f Chicago Press.
H a l l , В. H., and Z ied o n is , R. H. (2001), "The Patent Paradox Revisited: An Empirical Study
o f Patenting in the U.S. Semiconductor Industry, 1979-1995” RAND Journal of Economics
32(1): 101-28.
H a r a b i , N. (1995), Appropriability o f Technical Innovations: An Empirical Analysis,”
Research Policy 24: 981-92.
J a ff e , A. (2000), The U.S. Patent System in Transition: Policy Innovation and the Innov
ation Process,” Research Policy 29; 531-57.
K a u f e r , E. (1989), The Economics of the Patent System, New York: Harwood Academic
Publishers.
K lem perer, P. (1990), “ How Broad should the Scope of Patent Protection be?” RAND
Journal of Economics 21(1): 113-30.
K o r t u m , S., and L e r n e r , J. (1999), “ What is behind the Recent Surge in Patenting?” Research
Policy 28: 1-22.
L e r n e r J, (2000), 150 Years o f Patent Protection,” NBER , Working paper No. 7478.
,
Machlup , F. (1958), An Economic Review of the Patent System, Study No 15 of the Subcommit
tee on Patents, Trademarks, and Copyrights of the Committee on the Judiciary, US Senate,
Washington, DC: US Government Printing Office.
M a n s f ie l d , E. (1985)» 'How Rapidly Does New Industrial Technology Leak Out?; The
Journal of Industrial Economics, XXXIV, No. 2. 217-23
—— (1986), Patents and Innovation: An Empirical Study”, Management Science 32(2):
173-81.
------Rapo po rt , J., R omeo , A., W a g n er , S., and B ea r d sley , G. (1977), “ Social and private
rate o f return from industrial innovations,” Quarterly Journal of Economics 71 (May):
221-40.
—— Sch w artz , M., and W a g n e r , S. (1981), “ Imitation Costs and Patents: An Empirical
Study,” Economic Journal 91 (Dec.): 907-18.
♦ M a n s f i e l d , E, D., and M a n sfie ld , E. (guest eds.) (2000), “ Intellectual Property Protection
and Economic Development,” International Journal of Technology Management, Special
issue, 19 (1/2).
♦ M a s k u s , K. E. (2000), Intellectual Property Rights in the Global Economy; Washington, DC:
Institute for International Economics.
*M azzo leni , R., and N elso n , R. R. (1998), “ The Benefits and Costs o f Strong Patent
Protection: A Contribution to the Current Debate,” Research Policy 27(f): 273-84.
M erges , R. R, and N elso n , R. R. (1990), “ On the Complex Economics o f Patent Scope,”
Columbia Law Review 90(4): 839-916.
M oser , R (2003 ), “ How Do Patent Laws Influence Innovation? Evidence from 19th-Century
World Fairs,” National Bureau o f Economic Research, Working Paper No. 9909.
M owery , D. (1996), The International Computer Software Industry New York: Oxford
University Press.
N elson , R. R. (ed.) (1993), National Innovation Systems: A Comparative Analysis, New York
and Oxford: Oxford University Press.
N ordhaus , W. D. (1969), Invention, Growth and Welfare, Cambridge, Mass.: MIT Press.
N orth , D. C. (1981), Structure and Change in Economic History; New York: W.W. Norton.
O rdover , J. A. (1991), “A Patent System for Both Diffusion and Exclusion,” Journal of
Economic Perspectives 5(1): 43-60.
Pa v itt , K. (1999), Technology»Management and Systems of Innovation, Cheltenham: Edward
Elgar.
P enrose , E. T. (1951), The Economics of the International Patent System, Baltimore: Johns
Hopkins University Press.
R eichm an , J. H. (1994), “ Legal Hybrids between the Patent and Copyright Paradigms,”
Columbia Law Review 94(8): 2432-558.
R iv ette , K. G., and K l in e , D. (1999), Rembrandts in the Attic: Unlocking the Hidden Value of
Patents, Boston: Harvard Business School Press.
R o m e r , D. (1996), Advanced Macroeconomics, Berkeley, University o f California: McGraw-
Hill.
Sam uelson , P. (1993), “A Case Study on Computer Programs,” in Wallerstein et ai. (1993:
284-318).
Sch erer , E M. (1966), “ Time-Cost Tradeoffs in Uncertain Empirical Research Projects,”
Naval Research Logistics Quarterly 13: 71-82.
---- -(1967), “ Research and Development Resource Allocation under Rivalry, Quarterly
Journal of Economics 81:359-94*
290 OVE G R A N S T R A N D
THE GEOGRAPHY
OF I N N O V A T I O N
REGIONAL
INNOVATION SYSTEMS
B J 0 R N T. A S H E I M
M E R I C S. G E R T L E R
li.i In tr o d u c tio n
time. Given these rather striking stylized facts, it would appear that the process of
knowledge production exhibits a very distinctive geography.
We argue in this chapter that this geography is fundamental, not incidental, to the
innovation process itself: that one simply cannot understand innovation properly if
one does not appreciate the central role o f spatial proxim ity and concentration in
this process. Our goal is to demonstrate why this is true, and to examine how
innovation systems at the subnational scale play a key part in producing and
reproducing this uneven geography over time.
This chapter addresses four key issues. First, w hy does location “ matter” when it
comes to innovative activity? If one considers the production and circulation o f new
knowledge to be the core o f innovation, then it is important to have a sound
understanding o f the nature o f the different types o f knowledge involved and their
geographical tendencies. Second, what are regional innovation systems, and what
role do they play in generating and circulating new knowledge leading to innov
ation? Third, what is the relationship between regional systems o f innovation and
institutional frameworks at the national level? Finally, what is the relationship
between local and global knowledge flows, and is there any evidence that the global
nature o f today’s economy has weakened or altered the influence o f proximity on the
geography o f innovation?
11.2 T y p e s of K n o w l e d g e a n d
th eir G eographies
A grow in g b o d y o f thought argues that in a com petitive era in w h ich success depends
increasingly upon the ability to produce new o r im p ro ve d p ro d u cts and processes,
tacit know ledge constitutes the m ost im p o rtan t basis fo r in n o vation -b ased value
creation (Pavitt 2002), A s M askell and M alm b erg (1999; 172) have put it, when
everyone has relatively easy access to exp licit/cod ified kn ow ledge, the creation of
unique capabilities and products depends on the p ro d u c tio n and use o f tacit
know ledge:
1 hough often overlooked, a logical and interesting consequence of the present development
towards a global economy is that the more easily codifiable (tradable) knowledge can be
accessed, the more crucial does tacit knowledge become for sustaining or enhancing the
competitive position of the firm.. . . In other words, one effect of the ongoing globalisation is
t at many previously localised capabilities and production factors become ubiquities. What
is not ubiquified, however, is the non-tradable/non-codified result of knowledge creation—
the embedded tacit knowledge that at a given time can only be produced in practice. The
T H E G E O G R A P H Y OF I N N O V A T I O N 293
fundamental exchange inability of this type of knowledge increases its importance as the
internationalisation ot markets proceeds.
Im plicit in the above quote is a fundam entally spatial argum ent: tacit knowledge
is a key determ inant o f the geography o f innovative activity. There are two closely
related elements to this argum ent. First, because it defies easy articulation or
codification (Polanyi i 95$> 1966), tacit know ledge is difficult to exchange over long
distances. It is h eavily im bued w ith m eaning arising from the social and institutional
context in which it is produced, and this context-specific nature makes it spatially
sticky (Gertler 2003). T h e second relates to the changing nature o f the innovation
process itself and, in particular, the grow ing im portance o f socially organized
learning processes. T h e argum ent here is that innovation has com e to be based
increasingly on the interactions and know ledge flows between econom ic entities
such as firm s (custom ers, suppliers, com petitors), research organizations (univer
sities, other pu blic and private research institutions), and public agencies (technol
ogy transfer centers, developm ent agencies). T his is fundam ental to Lundvall and
Johnsons (1994) learning econ om y thesis, and is especially well reflected in their
concept o f “ learning through interacting.” W hen one com bines these two features o f
the innovation process— the centrality o f “ sticky,” context-laden tacit know ledge
and the grow ing im portan ce o f social interaction— it becom es apparent w hy geog
raphy now “ m atters” so m uch.
The recent literature on learning regions further explores the character and
geographical consequences o f tacit know ledge (see Lundvall and Johnson 1994;
Florida 1995; A sheim 1996, 2001; M organ 1997; C ooke and M organ 1998; Lundvall
and M askell 2000). It argues that tacit know ledge does not “ travel” easily because its
transm ission is best shared through face-to-face interaction between partners who
already share som e basic com m onalities: the sam e language; com m on “ codes” o f
com m unication and shared conventions and norm s that have been fostered b y a
shared institutional environ m en t; and personal know ledge o f each other based on a
past h istory o f successful collaboration or inform al interaction. These co m m on al
ities are said to serve the vital p u rp ose o f bu ilding trust between partners, which in
turn facilitates the local flo w o f tacit (and codified) know ledge between partners.
This approach adopts the learning-by-interactin g m odel as the cornerstone o f its
conceptual fram ew ork, and argues that the production o f tacit know ledge occurs
sim ultaneously w ith the act o f tran sm ission— p rim arily through the m echanism o f
user-producer interaction (Lu n dvall 1988; G ertler 1995)* A ccording to this perspec
tive, know ledge does not flo w u n id irection ally from technology producers to users.
Instead, users p ro vid e tacit and proprietary, codifiable know ledge to producers in
order to enable the latter to devise innovative solutions to users" practical problem s.
But at the sam e tim e, b y su p p lyin g users w ith innovative technologies, producers are
also sharing their tacit and other p ro p rietary know ledge w ith their custom ers. The
end product arisin g fro m this close interaction benefits both users and producers,
294 B J 0 R N T. A S H E I M A N D M E R I C S . G E R T L E R
and em bodies w ithin it new know ledge that co u ld not h ave been p ro d u ced b y either
p arty w o rk in g in isolation. This, in effect, describes a so cial process o f jo in t innov
ation and know ledge produ ction .
Lam (1998, 2000) points out that the skills req u ired fo r effective knowledge
transfer w ithin collective learning processes are h ig h ly tim e- and space-specific.
Interactive, collective learning is based on co m p atib le intra- o r interorganizational
routines, tacit norm s and conventions regulatin g collective actio n as well as tacit
m echanism s for the ab sorp tion o f codified know ledge. T h is requires that the actors
in question have a shared understandin g o f alocal codes,” o n w h ich collective tacit as
w ell as d isem bodied codified know ledge is based (A sh eim 1999; Lu ndvall 1996).
T h us, the ab ility to interpret local codes in consisten t w ays w ill be critical for the
integration o f the operations o f a firm w ith in a lo cal in terfirm learn in g network.
Since spatial p ro xim ity is key to the effective p ro d u c tio n an d transmission/
sharing o f tacit know ledge, this reinforces the im p o rtan ce o f in n o vative clusters,
districts, and regions. M oreover, as M askell an d M a lm b e rg (1999) p o in t out, these
regions also benefit from the presence o f localized capab ilities an d intangible assets
that further strengthen their centripetal pu ll (D o si 1988; S to rp e r 1997). M an y o f these
are social assets— i.e. they exist between rather than w ith in firm s. A lth ou gh they are
therefore not fu lly appropriable b y in d ivid u al firm s, o n ly lo cal firm s can enjoy their
benefits. These assets include the region's u n iq u e in stitu tio n al endow m ent, which
can act to su p p o rt and reinforce local advantage. Because such assets evolve slowly
over tim e, exh ibiting strong tendencies o f p ath -d ep en d en t d evelopm ent (David
1994; Z ysm an 1994), they m ay prove to be v ery d ifficu lt to em ulate b y would-be
im itators in other regions, thereby preservin g the in itial ad van tage o f “ first mover”
regions. M askell and M alm b erg argue (19 9 9 :18 1):
It is the region s distinct institutional endowment that embeds knowledge and allows for
knowledge creation which—through interaction with available physical and human re
sources constitutes its capabilities and enhances or abates the competitiveness of the
firms in the region. The path-dependent nature of such localised capabilities makes them
difficult to imitate and they thereby establish the basis of sustainable competitive advantage.
be used by organizations and firm s, and there is m ore interdependence and a finer
division o f lab ou r am on g actors: individuals, com panies, and other organizations
(Cowan et a l 2000). N on aka and Takeuchi (1995) and Lundvall and Borras (1999)
have pointed out that the process o f know ledge generation and exploitation requires
a dynam ic interplay betw een, and transform ation of, tacit and codified form s o f
knowledge as well as a strong interaction o f people w ithin organizations and
between them. Thus, these know ledge processes have becom e increasingly inserted
into various form s o f netw orks and innovation system s— at regional, national and
international levels (see Ch. 3 b y Powell and G rodal in this volum e for a discussion o f
the role o f netw orks in in n ovation ).
Despite the general trend tow ards increased diversity and interdependence in the
knowledge process, Pavitt (1984) and others have argued that the innovation process
o f firms is also strongly shaped by their specific know ledge base, which tends to vary
systematically b y indu strial sector (see also Ch. 1 b y Fagerberg and Ch. 15 by von
Tunzelmann and A cha, in this volu m e). For the purposes o f this chapter, we
distinguish between tw o types o f know ledge base: “ analytical” and “ synthetic”
(Laestadius 1998). These types entail different m ixes o f tacit and codified knowledge,
as well as different codification possibilities and lim its. T h ey also im ply different
qualifications and skills, reliance on different organizations and institutions, as well
as contrasting in n o vation challenges and pressures.1
A synthetic know ledge base prevails in industrial settings where innovation takes
place m ainly throu gh the app lication o r novel com bination o f existing knowledge.
Often this occurs in response to the need to solve specific problem s arising in the
interaction w ith clients and suppliers. In d u stry exam ples include specialized indu s
trial m achinery, plant engineering, and shipbuilding. R & D is in general less im p o rt
ant than in other sectors o f the econom y. W hen it occurs, it tends to take the fo rm o f
applied research, but m ore often it involves increm ental product or process devel
opment related to the solu tion o f specific problem s presented by custom ers (von
Hippel 1988). U n iversity 4 n d u stry links are relevant, but they are clearly m ore
significant in the realm o f app lied research and developm ent than in basic research.
Knowledge is created less in a deductive process or through abstraction than through
an inductive process o f testing, experim entation, com puter-based sim ulation, or
practical work. K n ow ledge em bodied in the respective technical solution or engin
eering w ork is at least p artially codified. H ow ever, tacit know ledge seems to be m ore
im portant than in oth er types o f activity, due to the fact that know ledge often results
from experience gained at the w orkplace, and through learning by doing, using, and
interacting. C o m p ared to the second know ledge type (“ analytical ) described
below, m ore concrete kn ow -h ow , craft and practical skill is required in the k n o w
ledge pro d u ction and circu latio n process. These form s o f know ledge are often
provided b y pro fession al and polytech n ical schools, or b y on-th e-job training.
The in n o vation process fo r industries w ith a synthetic know ledge base tends to be
oriented tow ards the efficien cy and reliability o f new solutions, or the practical
296 B J 0 R N Т. А S Н Е I М AND MERI C S. G E R T L E R
u tility and user-friendliness o f produ cts fro m the perspective o f the customers.
Inn о vat ion “related activities are d om in ated b y the m o d ifica tio n o f existin g products
and processes. Since these types o f in n o vation are less d isru p tive to existing routines
and organizations, m ost o f them take place in existin g firm s, m ak in g spin-offs and
new firm fo rm atio n for the developm ent and exp lo itatio n o f n ew synthetic know
ledge relatively infrequent.
In contrast, an analytical know ledge base d om in ates e co n o m ic activities where
scientific know ledge is h igh ly im portan t, and w h ere k n o w led ge creation is often
based on fo rm al m odels, codified science and ration al processes. P rim e exam ples are
b io tech n o logy and in fo rm atio n technology. B o th basic and ap p lied research, as well
as the system atic developm ent o f produ cts and processes, are central activities in this
form o f know ledge prod u ction . C o m p an ies typ ically have th eir o w n in-house R&D
departm ents but they also rely on the research o u tp u t o f u n iversities and other
research organizations in their in n ovation process. U n iv e rsity -in d u stry links and
netw orks are thus im portan t, and this type o f in teraction is m o re frequent than in
the synthetic type o f know ledge base. K now ledge in pu ts and o u tp u ts in this type o f
know ledge base are m ore often codified (o r read ily codifiable) than in the case o f
synthetic know ledge. T h is does not im p ly that tacit kn ow led ge is irrelevant, since
both kinds o f know ledge are always involved in the process o f know ledge creation
and in n ovation (N onaka et al. 2000, Joh n son , Lorenz, an d L u n d vall 2002).
T h e im portance o f codification in analytic kn ow ledge reflects several factors:
know ledge inputs are often based on review s o f existin g (cod ified ) studies, know
ledge generation is based on the app lication o f w id ely sh ared and understood
scientific principles and m ethods, know ledge processes are m o re fo rm a lly organized
(e.g. in R & D departm ents), and ou tcom es tend to be d o cu m en ted in reports,
electronic files, or patent descriptions. K n ow led ge ap p licatio n takes the form o f
new products or processes, which are m ore likely to constitu te rad ical innovations
than in those industries for w hich synthetic know ledge constitu tes the principal
know ledge base. N ew firm s and sp in -o ff com panies (i.e. n ew m arket entrants rather
than existing firm s) are an im portan t co n d u it fo r the a p p licatio n o f knowledge
em bodied in these radically new inventions or p ro d u cts.2
H o w is the im portan ce o f tacit, as opposed to cod ified , know ledge, as well as the
geography o f in n ovation , affected b y this d ifferential im p o rtan ce o f synthetic and
analytical know ledge bases across industries and technologies? Clearly, the 'learn in g
through interacting scenario at the core o f the learn in g e c o n o m y and learn-
ing regions thesis seems to be based im p licitly o n activities fo r w h ich synthetic
form s o f know ledge are central. For instance, m an y o f Lu ndvalFs (1988) original
exam ples com e from the realm o l m echanical engineerin g and specialized industrial
m achinery, where non-linear, iterative interaction betw een users and producers
represents the p rim ary m ode o f innovation. For such e co n o m ic activities, the spatial
concentration ot interacting firm s sharing a co m m o n social and institutional
context is an o bviou s prerequisite to socially organized , interactive learning
T H E G E O G R A P H Y OF I N N O V A T I O N
processes (Gertler 2004)» But what about those sectors for which analytical know
ledge is pre-em inent? G iven the greater prom inence o f codified and eodifiable
knowledge in the in n ovation process, m ight we not expect innovation processes
within analytically based industries to be m ore w idely distributed spatially?
A pparently not. F o r starters, econom ists have produced m uch striking evidence
about the highly uneven geography o f innovation in analytically based activities.
One im portant approach proceeds by m easuring knowledge spillovers through the
use o f indicators such as patent citations.3 For exam ple, in their classic study, Jaffe et
al. (1993) find evidence that patent applicants in analytically based industries cite
other patents o rigin atin g in the sam e city m ore frequently than they cite patents
originating non-locally, Fu rtherm ore, they find that patent citations are m ore likely
to be localized in the first year following the establishment of the patent, with the
effect fading over time, as the knowledge diffuses more widely.
A related app roach tracks know ledge spillovers in analytically based industries
such as b iotech nology and pharm aceuticals through the analysis o f “ star scientists.57
Zucker, Darby, and colleagues have tracked the location o f these highly productive
scientists and their impact on innovation in the local economy, demonstrating that
the rates o f start-up o f new biotech firms are significantly higher in those regions in
which these key scientists live and work (Zucker and Darby 1996; Zucker, Darby, and
Armstrong 1998; Zucker, Darby, and Brewer 1998). Moreover, firms that have
established working relationships with star scientists outperform firms that do not
enjoy this kind o f access, in terms o f productivity growth, new product develop
ment, and employment growth.
Both o f these sets o f findings strongly suggest that in fact the innovation process in
industries based on analytical forms o f knowledge is no less spatially concentrated
than those forms o f innovative economic activity based on synthetic types of
knowledge. Indeed, if anything, there is compelling evidence to suggest that the
former may exhibit an even higher degree o f geographical concentration than the
latter (Cortright and M ayer 20 0 2).4
How can one explain this counterintuitive finding? What are the processes
underlying innovation in analytically based industries that explain their distinctive
and highly uneven geography? There are three principal forces at work here. First, it
is clear that, despite the importance o f eodifiable knowledge in analytically oriented
sectors, the circulation o f new knowledge remains highly localized, as the economic
literature on knowledge spillovers (reviewed above) attests. This is because these
spillovers occur first, fastest and most readily within established local social net
works o f scientists— often b y word o f mouth, well before formal results are p u b
lished in widely accessible outlets. Some forms o f valuable knowledge are almost
never transmitted non-locally. For example, knowledge concerning failures in
scientific experiments is rarely, if ever, published. Yet, the knowledge that a particular
research strategy failed to yield expected results can save research teams considerable
time an d expense i f it prevents them fro m p u rsu in g unproductive lines o f inquiry
298 B J 0 R N T. A S H E I M A N D M E R I C S. G E R T L E R
Having presented the most important arguments to explain the consistent tendency
towards the geographical concentration o f innovative activities, we turn now to
T H E G E O G R A P H Y OF I N N O V A T I O N 299
consider the role o f in n ovation system s at the subnational level in fostering and
prom oting this process.
The concept o f a regional in n ovation system (R IS) is a relatively new one, having
first appeared in the early 1990s (A sheim 1995, A sheim and Isaksen 1997; C ooke 1992,
1998,2001), follow in g F reem an s use o f the innovation system concept in his analysis
o f Japan s econ om y (Freem an 1987b and at approxim ately the sam e tim e that the
idea o f the national in n ovation system was exam ined in books by Lundvall (1992)
and Nelson (1993)* A s this ch ro n o logy suggests, the regional innovation system
concept was inspired b y the national innovation system concept, and it is based on
a similar rationale that em phasizes territorially based innovation system s.5
One such rationale stem s fro m the existence o f technological trajectories that are
based on “ sticky” know ledge and localized learning w ithin the region. These can
become m ore innovative and com petitive b y prom oting stronger system ic relation
ships between firm s and the re g io n s know ledge infrastructure. A second rationale
stems from the presence o f know ledge creation organizations whose output can be
exploited for econ om ically useful pu rposes b y supporting newly em erging econom ic
activity. The em ergence o f the concept o f a regional innovation system coincides
with the success o f regional clusters and industrial districts in the post-Fordist era
(Asheim 2000; A sh eim and C o ok e 1999; Piore and Sabel 1984; Porter 1990,1998), and
the elaboration o f the concept represents an attem pt b y students o f the geographical
econom y to understand better the central role o f institutions and organizations in
prom oting in n ovation -b ased regional grow th (A sheim et al. 2003; G ertler and W olfe
2004).6
The regional in n o vation system can be thought o f as the institutional infrastruc
ture supporting in n o vation w ith in the produ ction structure o f a region. Taking each
element o f the term in tu rn (A sheim and C o oke 1999), the concept o f region
highlights an im p o rtan t level o f governance o f econom ic processes between the
national level and the level o f the in d ivid u al cluster or firm . Regions are im portant
bases o f econom ic co ord in atio n at the m eso-level: athe region is increasingly the
level at w hich in n o vation is produced through regional netw orks o f innovators, local
clusters and the cross-fertilizin g effects o f research institutions” (Lundvall and
Borras 1999:39). In v ary in g degrees, regional governance is expressed in both private
representative organizations such as branches o f indu stry associations and ch am
bers o f com m erce, and pu blic organizations such as regional agencies w ith pow ers
devolved fro m the national (or, w ithin the E uropean U nion, supranational) level to
prom ote enterprise and in n o vation su p port (A sheim et al. 2003; C o oke et al. 2000).
The system ic d im en sio n o f the R IS derives in part fro m the team -like character
associated w ith in n o vatio n in netw orks. A lth ou gh an innovation system is a set o f
relationships betw een entities o r nodal points involved in innovation (see Lundvall
1992 for m ore d iscu ssion ), it is m uch m ore than this. Such relationships, to be
systemic, m ust involve som e degree o f interdependence, though to varyin g degrees.
Likewise, not all such system ic relations need be regionally contained, but m any are.'
ЗОО В J 0 R N Т. A S H E I M A N D M E R I C S. G E R T L E R
A s the interactive m ode o f in n o vation grow s in im p o rtan ce, these relations are more
likely to becom e regionally contained, especially in the case o f specialized suppliers
w ith a specific tech n ology o r know ledge base. Su ch su p p liers o ften depend on tacit
know ledge, face-to-face interaction and tru st-b ased relation s and, thus, benefit from
co op eration w ith custom ers in region al clusters, w h ile cap a city subcontractors are
increasingly sourced globally.8 Fu rther rein forcin g the system ic character o f the RI$
is the prevalence o f a set o f attitudes, values, n o rm s, rou tin es, and expectations—
described b y som e as a distinctive “ regional cu ltu re” — that influ ences the practices
o f firm s in the region. A s n oted earlier, it is this co m m o n region al culture— itself the
pro d u ct o f co m m o n ly experienced in stitu tion al forces— that shapes the w ay that
firm s interact w ith one anoth er in the region al econom y.
In addition to these regionally based elements, the entire regional innovation system
is embedded within a national regulatory fram ework that reinforces innovative айіѵііу
at the regional scale* The most important features o f this'- system are: :;
* Labor market structures that fester stable employment relations, ѣ а Ш М щ к т Щ
by doing and strengthening employers’ incentives to train. я
* An industrial relations system that formalizes worker participation in day-to-day
and longer-term strategic decision making, enabling employers to harness workers'
tacit knowledge acquired through learning by doing and using*
* Centralized collective bargaining systems that minimize interfirm variation in .щ р ,
and benefit levels, inducing firms to compete on the basis o f quality and innovative-
ness.
* Capital market structures that encourage longer-term time horizons in firm-level
decision making, thus further reinforcing stability in the workplace*
The region ally netw orked in n o vation system is a result o f p o lic y intervention to
increase in n o vation cap acity and collab oration . S M E s, fo r exam p le, m ay have to
supplem ent their in fo rm al know ledge (characterized b y a h igh tacit component)
w ith com petence arising from m ore system atic research an d d evelo p m en t in order
to carry out m ore radical innovations. In the lo n g ru n , m ost firm s cannot rely
exclusively o n in fo rm al localized learning, but m ust also gain access to w id er pools of
both analytical and synthetic know ledge on a n ation al an d glo bal basis. T h e creation
o f region ally netw orked in n o vatio n system s th ro u gh increased cooperation with
local universities and R & D institutes, o r th ro u gh the estab lish m en t o f technology
transfer agencies and service centers, m ay p ro vid e access to in fo rm a tio n and com
petence that supplem ents firm s' locally derived com petence. T h is n o t o n ly increases
their collective innovative capacity, but m ay also serve to co u n teract technological
ulo ck -in " (the in ab ility to deviate fro m an established b u t o u tm o d ed techno
logical trajectory) w ithin regional clusters o f firm s.
The third m ain type o f R IS, the regionalized national innovation system, differs
fro m the tw o preceding types in several ways. First, parts o f in d u stry and the
institutional infrastructure are m ore fu n ctio n ally integrated in to n ation al or inter
national in n ovation system s— Le. in n o vation activity takes place p rim arily in co
operation w ith actors outside the region. T h u s, this represents a developm en t model
in w h ich exogenous actors and relationships play a larger role. C o oke (1998)
describes this type as dirigiste R IS, reflecting a n arro w er d efin itio n o f an innov
ation system in corp oratin g m ain ly the R & D fu n ctio n s o f un iversities, research
institutes, and corp orations. Second, the co llab o ratio n betw een organizations
w ithin this type o f R IS conform s m ore closely to the lin ear m o d el, as the cooperation
p rim arily involves specific projects to develop m ore rad ical in n o vatio n s based on
form al an alytical-scien tific know ledge. W ithin such system s, co o p eratio n is most
T H E G E O G R A P H Y OF I N N O V A T I O N 303
likely to arise between people w ith the sam e occupational or educational back
ground (e.g. am o n g engineers). This functional sim ilarity facilitates the circulation
and sharing o f know ledge through “ com m unities o f practice,” whose m em bership
m ay cross inter-regional and even international boun daries (A m in and Cohendet
2 0 0 4 )*
One special exam ple o f a regionalized national innovation system is the clustering
o f R & D laboratories o f large firm s and/or governm ental research institutes in
planned “ science p a r k s ” These m ay be located in close pro xim ity to universities
and technical colleges, but the evidence suggests that science park tenants typically
have lim ited linkages to local in d u stry (Asheim 1995). Science parks are, thus, a
typical exam ple o f a planned innovative m ilieu com prised o f firm s w ith a high level
o f internal resources and com petence, situated w ithin weak local cooperative envir
onments. These parks have generally failed to develop innovative networks based on
interfirm cooperation and interactive learning w ithin the science parks them selves
(Asheim and C o o k e 1998; H en ry et al. 1995)- Technopoles, as developed in countries
such as France, Jap an , and Taiw an, are also characterized by a lim ited degree o f
innovative interaction betw een firm s w ithin the pole, and by vertical subcontracting
relationships w ith n o n -lo cal external firm s. In those rare cases where local in n o va
tive networks arise, they have n orm ally been orchestrated by deliberate pu blic sector
intervention at the national le v e l These characteristics im ply a lack o f local and
regional em beddedness, and lead us to question the capability o f science parks and
technopoles to p ro m ote innovativeness and com petitiveness m ore w idely w ithin
local industries (especially SM E s) as a prerequisite for endogenous regional devel
opm ent9 (A sheim and C o o k e 1998; and Longhi and Q uere 1993).
Recent w o rk ap p lyin g the R IS concept has begun to explore the linkage between the
larger institutional fram ew o rks o f the national innovation system and national
business system , and the character o f regional in n ovation systems. This question
has recently been addressed b y C o oke (2001) in studies o f biotech nology in the U K ,
the U SA , and G erm any. C o o k e has Introduced a distinction between the traditional
regional in n o vation system (w hich he refers to as the institutional regional in n o v
ation system — IRIS) and the new econ om y in n ovation system (N E IS), which he
304 B J 0 R N Т. AS HEI M AND MERI C S. G E R T L E R
A recent study (Cortright and Mayer 2002) o f biotechnology in the United States
concludes that innovation in this industry is dominated by just a handful o f metropol
itan centers. Boston and San Francisco are the two largest and best-established centers,
followed by recent entrants San Diego, Seattle, and Raleigh-Durham. Philadelphia,
New York, Washington-Baltimore, and Los Angeles also have significant concentra
tions of biotech activity. The authors of this study conclude that the two most
important overarching factors supporting the emergence of strong biotech clusters
are: (i) the presence of first-class pre-commercial medical research in a local university
or government laboratory, and (ii) local systems to support and encourage entrepre
neurial activity leading to successful translation of research into commercially viable
outputs.
In emphasizing these two factors, this study emphasizes the key roles played by both
public and private sectors actors in such innovation systems. Government support,
through key granting councils such as the National Institutes of Health and state-level
programs to invest in university systems and research, is at least a necessary condition
for the emergence of a local biotechnology cluster. It also plays a role in the active-
recruitment of star scientists. But this process also requires the local presence of
T H E G E O G R A P H Y OF I N N O V A T I O N 305
In m aking these argum ents about a general correspondence between the m acro-
institutional characteristics o f the econ om y and the dom inant form and character o f
its regional in n o vation system s, C ooke provides a link to another useful literature on
"Varieties o f capitalism ” and national business system s (Lam 1998, 2000; W hitley
1999; H all and Soskice 2001). Soskice (1999) argues that different national insti
tutional fram ew orks evolve to su p port particular form s o f econom ic activity— i.e.
that coordinated m arket econom ies such as G erm an y and the N ordic countries base
their com petitive advantage in “ diversified quality prod u ction ” (Streeck 1992), while
liberal m arket econom ies such as the U S and U K are m ost com petitive in industries
characterized b y science-based innovative activities. W ithin the coordinated m arket
economies, the d rivin g force is the non-m arket coordination and cooperation that
exists inside the business sphere and between private and public actors, as well as the
degree to w hich lab or is m ean in gfu lly “ in corp orated ” w ithin the production process
and the financial system is able to su p p ly long-term finance (Soskice 1999)* In a
com parison between coordinated m arket econom ies such as Sweden, Germ any, and
Switzerland on the one hand, and liberal m arket econom ies such as the U S and U K
on the other, he fo u n d that the coordinated econom ies perform ed best in the
production o f “ relatively co m p lex products, involvin g com plex production p ro
cesses and after-sales service in w ell-established industries (e.g. synthetically based
sectors such as the m ach in e too l indu stry). B y contrast, the U S perform ed best in
industries p ro d u cin g co m p lex system ic products such as IT and defense technology,
where the im p o rtan ce o f analytical, scientific-based know ledge often w ith the
m ajor su p po rt o f the state— is significant (Soskice i 999 : U 3- 1 4 )*
ЗОб B J 0 R N T. A S H E I M A N D M E R I C S. G E R T L E R
Thus, liberal m arket econom ies as represented by the U S and the U K seem to have
advantages in industries characterized by an analytical knowledge base, as well as in
those sectors that depend on a high degree o f m o b ility in labor markets. Concerning
the form er, the elite universities and education institutions, often privately o rgan
ized, provide strengths in R8cD, the generation o f form alized knowledge, inventions,
and radical innovations. O ther institutional features such as close u n iversity-in d u s
try links, academ ic spin -offs and an active scientific labor m arket all operate to
prom ote the transfer and application o f scientific knowledge.
Placed in this context the classic “ trad ition al” institutional regional innovation
system typified by a region such as G erm any's Baden-W iirttem berg is m ost com pat
ible with the institutional fram ew ork o f a coordinated m arket econom y, while the
new econom y in n o vation system (Lon d on advertising, Silicon Valley, or N ew York
City's new m edia “ Silicon A lley” ) reflects the institutional fram ew ork o f a liberal
market econom y.
This raises an im p o rtan t issue that has been the subject o f som e debate in the
literature, concerning the extent to w hich m arkedly different regional innovation
systems can em erge w ith in the sam e national institutional space. Saxenian s (1994)
landm ark study co m p arin g the electronics and com puting industries in two d o m
inant regions o f the U nited States— C aliforn ia's Silicon Valley and Route 128 in
M assachusetts— has reinforced the view that w idely divergent regional innovation
systems can and do em erge w ithin a single national institutional fram ew ork. She
argues that Silicon Valley o u tp erform ed R oute 128 in term s o f em ploym ent grow th
and new firm fo rm atio n because o f its m ore open, flexible, h igh -m obility system
com pared to R ou te 128's m ore closed, rigid, hierarchical, loyalty-based system.
While both regions w ere h om e to w orld -renow ned institutions o f higher learning
and research, the Silicon Valley system proved m ore effective in generating successful
innovations in response to p ro fo u n d com petitive challenges from abroad.
It is im portan t to note, how ever, that this analysis was based on the evolution o f
these two regional system s d u rin g the 1980s and early 1990s. In fact, the M assachu
setts innovation system experienced a p ro fo u n d transform ation over the 1990s. Best
(2001) docum ents this tran sition , describing the em ergence o f new industries in
biotechnology, m edical devices, nanotechnology, and related fields. He argues that
this tran sform ation w as und erp in ned b y a m ore fundam ental shift in the social
organization o f the lead in g sectors o f the regional econom y tow ards open systems
architecture— in oth er w ord s, th rou gh its evolution tow ards a structure that m uch
more closely resem bles the new econ om y in n ovation system o f Silicon Valley. Best s
analysis suggests that Saxenian's earlier case studies captured two regional in n o v
ation system s at a tim e w hen one o f them (R oute 128) was exhausting the innovative
capabilities o f an older, already ou tm od ed system , but before a coherent, fully-
form ed alternative had em erged (K enney and vo n B u rg 1999; Saxenian 1999)* N ow
that the organizational contours o f this new system are clear, they suggest that the
character o f d ifferent regional innovation systems within the same national
308 B F 0 R N T . A S H E I M A N D M E R I C S. G E R T L E R
institutional space may vary within a considerably narrower range than was previ
ously thought. This variation is likely to depend primarily upon regionally specific
technological trajectories and knowledge bases.
11.5 A l t e r n a t iv e O r g a n iz a t io n a l F o rm s
a n d E m e r g in g R e l a t io n s h ip s B e t w e e n
L o c a l a n d G l o b a l K n o w l e d g e
Q uestions have lately been raised over w hether the sp atial em b ed dedn ess o f learning
and know ledge creation m ight be challenged b y alternative o rg an izatio n al form s—
in particular, tem p o rary organizations— w hich som e see as b eco m in g m ore preva
lent in the global econ o m y (A sheim 2002; G rab h er 2002). F o r exam ple, G ann and
Salter (2000) suggest that firm s in the con stru ction and en gin eerin g sector now rely
on projects to organize the p ro d u ctio n o f kn o w led ge-in ten sive an d com plex prod
ucts and system s. W hat im pact m ight the ad o p tio n o f te m p o ra ry fo rm s o f organiza
tio n have on the spatial em beddedness o f learn in g an d in n o vatio n ? G rabh er's (2002)
w o rk on projects in L o n d o n advertisin g (discussed above) show s h ow co-location
facilitates the co n tin u ou s and rapid reco n figu ratio n o f p ro ject team s as well as the
circulation o f know ledge concerning the com petencies an d exp erien ce o f potential
project partners.
In contrast, A ld erm an (2004) argues that “ there are . . . im p o rta n t a priori or
theoretical reasons w hy a project-based m odel does not fit c o m fo rta b ly with ideas
ab ou t clustering, localized learning and local in n o vatio n netw orks.” H is argument
relies on a recent literature that sees “ co m m u n ities o f p ractice” as key entities driving
the firm 's know ledge-processing activities. T h is literatu re argu es that routines and
established practices shaped b y organizations (o r subset co m m u n ities w ith in organ
izations) p ro m ote the pro d u ction and sh aring o f tacit and cod ifiable knowledge
(B row n and D u gu id 19 9 6 ,20 0 0 ; W enger 1998). C o m m u n ities o f practice are defined
as gro ups o f w orkers in fo rm ally b ou n d together b y shared experien ce, expertise and
com m itm en t to a jo in t enterprise. These co m m u n ities n o rm a lly self-organize for
the pu rp ose o f solvin g practical problem s facing the larger o rgan izatio n , and in the
process they produce innovations (both pro d u ct and process). T h e com m onalities
shared by m em bers o f the co m m u n ity facilitate the id en tification , jo in t production
and sharing o f tacit know ledge throu gh collab orative p ro b le m -so lv in g assisted by
story-telling and other narrative devices for circu latin g tacit know ledge.
A ccord in g to this view, organizational o r relational p ro x im ity an d occupational
sim ilarity are m ore im portan t than geograph ical p ro x im ity in su p p o rtin g the
T H E G E O G R A P H Y OF I N N O V A T I O N 309
production, identification, ap p ro priation and flow o f tacit know ledge (Allen 2000;
Am in 2000; A m in and C ohendet 2004). The resulting geography o f innovation
differs from that envisioned by adherents to the learning region approach. In this
view, the jo in t pro d u ction and diffusion/transm ission o f tacit and codifiable kn ow
ledge across intra-organizationai boundaries is possible, so long as it is m ediated
within these com m unities. M oreover, because com m unities o f practice m ay extend
outside the single firm to include custom ers or suppliers, know ledge can also flow
across the boundaries o f in divid u al organizations.
Furtherm ore, the com m unities o f practice literature asserts that tacit know ledge
m ay also flo w across regional and national boundaries i f organizational or “ virtual
com m unity” p ro xim ity is strong enough. In other w ords, learning (and the sharing
o f tacit know ledge) need not be spatially constrained if relational pro xim ity is
present. For large, m u ltin ation al firm s with “ distributed” know ledge bases and
multiple sites o f in n ovation , the use o f com m unities o f practice, aided and su p
ported b y ever-cheaper and m ore pow erful ICTs and air travel, is seen as an effective
strategy for overcom ing geographical separation.
These argum ents are useful rem inders o f the im portance o f relationships and the
strength o f und erlying sim ilarities rather than geographical p ro xim ity per $e in
determ ining the effectiveness o f know ledge-sharing between econom ic actors.
However, they fail to answ er a key question: w hat forces shape or defines this
“ relational proxim ity,” enabling it to transcend physical, cultural, and institutional
divides? H o w are shared understandings produced? M uch o f the com m unities o f
practice literature is largely silent on this question. A notable exception is the w ork o f
Brow n and D u gu id (2 0 0 0 :14 3) w ho stake out a very different position on the spatial
reach o f co m m u n ities o f practice:
They are usually face-to-face communities that continually negotiate with, communicate
with, and coordinate with each other directly in the course o f work. And this negotiation,
communication, and coordination is highly implicit, part o f work practice . . . In these
groups, the demands o f direct coordination inevitably limit reach. You can only work closely
with so many people.
organizations arou n d them that such locations foster (see also Asheim and Hers tad
2003; C ooke et aL 2000; Freem an 2002).
This w ork is a w elcom e conceptual contribution, but the em pirical basis for its
fram ework rem ains underdeveloped. Clearly, there is a need for future research—
both case studies and aggregate statistical analyses— to investigate the prevalence o f
this “ dual geograp h y" o f innovation m ore systematically. At the sam e tim e, this
approach raises a fu rth er question that rem ains unanswered: is this m etaphor o f
local buzz and global pipelines appropriate only for those science-intensive in d u s
tries whose in n ovation rests on an analytical know ledge base? To what extent are
non-local know ledge flow s and learning relationships extending between localized
centers o f in n ovation b ecom in g im portan t for those industries that rely m ore
heavily on a synthetic know ledge base?
A problem atic aspect o f the learning econom y and learning regions literature has
been its focus on learning by d oing and using based largely on local synthetic
knowledge w ith a high tacit content and increm ental innovations. We continue to
agree with F reem an s useful insight concerning “ the trem endous im portance o f
incremental in n o vation , learning b y doing, by using and by interacting in the
process o f technical change and diffusion o f in n o vation s" (Freem an 1993: 9 -10 ).
Yet in a highly com petitive, globalizing econom y, it m ay be increasingly difficult for
the reproduction and grow th o f a learning econom y to rely prim arily on increm ental
im provem ents to prod u cts and processes, and not on new products (i.e. radical
innovations). C revo isier (1994:259) argues that the exclusive reliance on increm ental
innovations “ w ou ld m ean that these areas w ill very quickly exhaust the technical
paradigm on w hich they are fou nd ed ."
In future studies it w ill be im p o rtan t to follow these tendencies, which u n d ou b t
edly will be reinforced b y globalization processes (see C h. 12 by N arula and Zanfei in
this volum e). The basic rationale o f regional innovation system s is that the system ic
prom otion o f localized learn in g processes can im prove the innovativeness and
com petitive advantage o f regional econom ies. W hat rem ains to be seen is how the
capacity o f regional in n o vatio n system s to upgrade the know ledge bases o f firm s in
regional clusters w ill develop over tim e.
N otes
synthetic knowledge bases correspond closely to those sectors encompassed by the first
two o f Pavitts categories (supplier-dominated and production-intensive). Similarly, the
analytical category corresponds directly to Pavitt s science-based industries.
2. We should acknowledge that many industries draw significantly upon both synthetic and
analytical forms o f knowledge. A clear example would be medical devices and technolo
gies, whose development rests upon knowledge drawn from fields as diverse as bioscience,
ICT, software, advanced materials, nanotechnology, and mechanical engineering. For this
reason, it makes sense to conceive o f individual industrial sectors being arrayed along a
continuum between purely analytical and synthetic industries, with many— such as the
automotive industry— occupying an intermediate position along this spectrum.
3. See Feldman (2000) for a useful overview o f this literature.
4. Cortright and Mayer produce evidence to show that the degree o f geographical concen
tration in the US life sciences industries is considerably higher than the population as a
whole. They also demonstrate— using indicators such as venture capital, funded research
conducted through inter-firm alliances, and new firm formation rates— that this concen
tration has increased dramatically during the past two decades.
5. This conceptualization o f regional innovation systems corresponds with the one found in
Cooke et al. (2000). In their words any functioning regional innovation system consists of
two subsystems: (i) the knowledge application and exploitation subsystem, principally
occupied by firms within vertical supply-chain networks; and (ii) the knowledge gener
ation and diffusion subsystem, consisting mainly o f public organizations.
6. There is a strong historical correspondence between these concepts and approaches and
agglomeration theories within regional science and economic geography, such as Per-
roux s (1970) growth pole theory.
7. In a recent study Carlsson (2003) shows that the majority o f theoretical as well as empirical
analyses o f innovation systems have a regional focus (see also Bathelt 2003 for a critical
discussion o f RIS).
8. A recent comparative study o f European clusters shows that firms increasingly find
relevant research activities and other supporting services inside the cluster boundaries
(Isaksen 2004).
9. See Ch. 8 in this volume by Mowery and Sampat for a similarly critical assessment of
science parks.
R efer en c es
P a v it t , K. (1984),
Sectoral Patterns of Technical Change: Towards a Taxonomy and a
Theory,” Research Policy 13: 343-73.
------(2002), “ Knowledge about Knowledge since Nelson and Winter: A Mbced Record,”
Electronic Working Paper Series Paper No. 83, SPRU, Uni versity of Sussex, June.
P e r r o u x , R (1970), "Note on the Concept of Growth Poles” in D. McKee et al. (eds.),
Regional Economics: Theory and Practice, New York: The Free Press, 93-103.
P io r e , M., and S a b e l , C. (1984), The Second Industrial Divide: Possibilities for Prosperity, New
York: Basic Books.
P o l a n y i , M. (1958), Personal Knowledge: Towards a Post-Critical Philosophy, London: Rou-
tledge & Kegan Paul.
------(1966), The Tacit Dimension, New York: Doubleday.
P o r t e r , M, (1990), The Competitive Advantage of Nations, New York: Basic Books.
P o r t e r , M. E. (1998), “ Clusters and the New Economics o f Competition ” Harvard Business
Review (Nov.-Dee.): 77-90.
*S a x e n ia n , A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and
Route 128, Cambridge, Mass.: Harvard University Press.
------(i 999)» “ Comment on Kenny and von Burg, 'Technology, Entrepreneurship and Path
Dependence: Industrial Clustering in Silicon Valley and Route 128’,” Industrial and
Corporate Change 8(1): 105-10.
S c h o e n b e r g e r , E. (1997), The Cultural Crisis of the Firm, Oxford: Blackwell.
S o s k ic e , D. (1999), “ Divergent Production Regimes: Coordinated and Uncoordinated
Market Economies in the 1980s and 1990s,” in H. Kitschelt et al. (eds.), Continuity and
Change in Contemporary Capitalism, Cambridge: Cambridge University Press, 101-34.
S t o r p e r , M. (1997), The Regional World: Territorial Development in a Global Economy>New
York: Guilford Press.
------and S c o t t , A. (1995), “ The Wealth of Regions,” Futures, 27(5): 505-26.
*------and V e n a b l e s , A. J. (2003), “ Buzz: The Economic Force o f the City,” Paper presented
at the DRUID Summer Conference 2003 on “ Creating, sharing and transferring know-
ledge: the role o f geography, institutions and organizations,” Elsinore, Denmark.
S t r e e c k , W. (1992), Social Institutions and Economic Performance—Studies of Industrial
Relations in Advanced Capitalist Economies, New York: Sage.
von H i p p e l , E. (1988), The Sources of Innovation, Oxford: Oxford University Press.
W e n g e r , E. (1998), Communities of Practice: Learning, Meaning and Identity, Cambridge:
Cambridge University Press.
W h it l e y , R. (1999), Divergent Capitalism: The Social Structuring and Change of Business
Systems, Oxford: Oxford University Press.
Z u c k e r , L. G., and D a r b y , M. R. (1996), “ Star Scientists and Institutional Transformation:
Patterns o f Invention and Innovation in the Formation of the Biotechnology Industry,”
Proceedings of the National Academy of Science 93:12709-16.
-------------- and A r m s t r o n g , J. (199В), “ Geographically Localized Knowledge: Spillovers or
Markets?” Economic Inquiry 36: 65-86.
-------------- an(i B r e w e r , M. B. (1998), “ Intellectual Human Capital and the Birth o f U.S.
Biotechnology Enterprises,” American Economic Review 88: 290-306.
Z ysm an , J. (1994), “ How Institutions Create Historically Rooted Trajectories o f Growth,”
Industrial and Corporate Changes: 243-83.
CHAPTER 12
GLOBALIZATION
OF INNOVATION
TJl H
jl.JL F
.н У -ROT
Ж.м
ч M
l. .F
п»'кі/ O F
Ш Jrl
M ULTINATIONAL
ENTERPRISES
RAJNEESH NARULA
ANTONELLO ZANFEI
12.1 In t r o d u c t i o n
1 2 . 2 T r e n d s i n THE
I n t e r n a t i o n a l i z a t i o n of
In n o v a tiv e A c tivities
A useful taxon o m y pro p o sed by A rch ibugi and M ichie (1995) identifies three m ain
categories o f the globalization o f innovation (Table 12.1). A lthough a variety o f
econom ic actors undertake innovation and are engaged in its internationalization,
the M N E is the o n ly institu tion w hich b y definition can carry out and control the
global generation o f in n o vation w ithin its boundaries. We briefly discuss each o f the
three categories below,
and early 1990s, and external patenting (i.e. patent ap p licatio n s o f n ation al inventors
abroad) has also rapidly increased.
Industrial RftD (1) Resident patents (2) Non-resident patents (3) External patents (4)
G L O B A L I Z A T I O N OF I N N O V A T I O N
Spain 12.7 1.8е -4 .5 2,0 0.2 19.2 1.3 16.0
Ireland 5.2C 15.4 6.8 2.3 4.9 31.1 6.7 24.3
Portugal 4.6d 2.2h —6.4 0.9 -0.5 37.2 -24.2 52.4
Greece n.a. —1.4* -0.8 —13.4m 2.4 37.0 n.a. 21.5
Sweden 5.9C 0.29 -0 .5 0.0 2.5 7.1 3.0 14.2
Austria 9.8a 5.19 0.3 -1 .6 3.4 9.0 1.4 10.1
Finland 6,8C 5.1 4.7 2.7 0.7 13.4 5.7 23.1
Switzerland 0.8a -0.5' -3,1 -1.5 2.2 7.8 -1.3 5.5
Norway 7.3 1.39 -2.7 0.9 -0.1 11.1 0.8 21.1
Australia n.a. 8.9е 5.2 1.5 -2.0 7.5 6.7 21.7
Canada 5.5 4.9 -1.1 2.2 -2.1 4.5 -0.5 21.5
OECD weighted average n.a. n.a. 1.3 2.7 0.9 9.3 0.9 13.3
1982 2001
The figures for R & D activity reflect sim ilar patterns, since m any o f the largest
firm s engaged in F D I are key actors in the generation and diffu sion o f innovation.
M ore than o n e-th ird o f the top too M N E s are active in the m ost R & D -in ten sive
industries, such as electronics and electrical equipm ent, pharm aceuticals, chem icals
(U N C TA D 2002). Furthermore, large M N Es play a dominant role in the innovative
activities o f their h om e countries. For instance, Siem ens, Bayer and H oechst per
formed 18 per cent o f the total manufacturing R & D expenditures in Germany in 1994
(Kumar, 1998). In 1997 three M N E s accounted for more than the 30 per cent o f the
overall U K R & D investment in manufacturing. These same M N Es also undertake a
growing share o f their total R& D activities outside their home countries.
Significant cross-national differences are also apparent in indicators o f inter
national R&D. The share o f national R& D expenditures accounted for by non
domestic sources varies substantially within the industrialized and developing areas
(see Table 12.4 for some details). The origins o f international R&D investment flows
also differ considerably among industrialized economies (Table 12.5). Cantwell
(1995) suggests that countries such as Switzerland, UK, and the Netherlands,
which have historically been home to large M N Es and that were long-time inter
national investors in R&D, have greatly expanded their offshore R&D investments
since World War II. Another group o f countries (such as France and Germany ) has
relatively few large M N Es, and their outward R& D investments have grown more
324 R A JN E E SH NARULA AND A N TO N ELLO ZANFEI
Country Percentage of R a B
Canada 34.2
Finland (1999) 14.9
France 16.4
Japan 1.7
Netherlands 21.8
Spain (1999) 32.8
UK (1999) 31.2
US 14.9
Czech Republic (1999) 8.4
Hungary 78.5
India (1994) 1.6
Turkey 10.1
rad ually d u rin g the last eigh ty years. A third gro u p inclu des co u n tries that were
major investors in offshore R & D d u rin g the first fourteen years o f the twentieth
„entury. O ffsh ore investm ent by these econom ies actu ally declined after 19 14 and
eturned to pre-W orld War I levels o n ly recently. T h is gro u p inclu des the United
G L O B A L I Z A T I O N OF I N N O V A T I O N 325
w ith in the T riad .7 W hile there are significant differences in the in tern atio n al disper
sion o f in n ovative activity across indu stries, firm s have gen erally n o t international
ized their in n ovative activity at the sam e rate as th eir p ro d u c tio n activities.
Excep tion s to this rule are M N E s o rigin atin g fro m sm all econ o m ies, such as Bel
giu m , the N etherlands, and Sw itzerland. A large p ro p o rtio n o f even the most
internationalized M N E s concentrate at h o m e their m o re strategic activities,
such as R & D and h eadquarters fu n ction s (B en ito et al. 2003).
T h is relatively low — but increasing— degree o f in tern atio n alizatio n is associated
inter alia w ith the co m p lex nature o f system s o f in n o vatio n , an d the em beddedness
o f the M N E ’s activities in the h om e en viron m en t (see e.g. N a ru la 20 0 2a), the need
fo r internal cohesion w ithin the M N E (Blanc and Sierra 1999, Z a n fe i 2000), and the
high q u ality o f local infrastructures and a p p ro p ria b ility regim es that R & D activities
tend to require. These factors, together w ith the d ifficu lties o f m an ag in g complex
technological po rtfolio s, im p ly that the in tern atio n alization o f in n o va tio n occurs at
a slow er pace than the in tern ation alization o f p ro d u ctio n .
12.3 O v e r se a s In n o v a t iv e A c t iv it ie s
o f MNEs: T h e o r e t ic a l a n d
E m p ir ic a l Iss u e s
activities abroad, with strategic decisions— including R&D and innovation— being
rigidly centralized in the home country. Vernon emphasized that coordinating
international innovative activities was too costly, due to the difficulties of collecting
and controlling relevant information across national borders. The R&D activities of
foreign subsidiaries were limited largely to the adoption and diffusion of centrally
created technology.
The second b road m otive for offshore R & D investment is strategic asset-
augmenting activity (D u n n in g and N arula 1995), also know n as hom e-base aug
menting (H BA ) activity (K uem m erle 1996). Firm s use these types o f R & D invest
ments to im prove existing assets o r to acquire (and internalize) or create com pletely
new technological assets through foreign-located R & D . The assum ption in such
cases is that the foreign lo cation provides access to complementary location-specific
advantages that are less available in its p rim a ry o r “ h om e” base (letto-G illies 2001).
In many cases, the strategic assets sough t b y the investing firm are associated w ith the
presence o f other firm s. A location w hich is hom e to a m ajor com petitor m ay attract
asset-augm enting investm ents b y other firm s in the sam e or in other related in d u s
tries (see C antw ell in this v o lu m e on the im plications o f these patterns o f FD I for the
competitiveness o f host countries). A sset-augm enting m otives and technology
sourcing have been p artially incorporated in form al m odels o f the FD I decision.9
The asset-augm enting perspective, w hich considers local contexts m ore as
sources o f com petencies and o f technological opportu nities, and less as constraints
to the action o f M N E s, m arks a fundam ental departure from the conventional
wisdom. In a sem inal co n trib u tio n , H edlund (1986: 2 0 -1) caught the essence o f
this new w ay o f con cep tu alizin g the role o f local contexts: “ The m ain idea is that the
foundations o f com petitive advantage no longer reside in any one country, but in
many. N ew ideas an d pro d u cts m ay com e up in m an y different countries and later be
exploited on a global scale.” (See K ogut 1989 for a sim ilar view .)
There are several reasons w h y such asset-augm enting R & D activities are hard to
achieve through m eans oth er than FD I. Som e o f these reasons are associated w ith the
nature o f technology. W h en the know ledge relevant for innovative activities is
clustered in a certain geograph ical area and is “ sticky,” foreign affiliates engage in
asset-augm enting activities in these areas in order to benefit from the external
economies and kn ow ledge spillovers generated b y the concentration o f production
and innovation activities in the relevant clusters. The tacit nature o f technology
implies that even w here know ledge is available through m arkets, it m ay still require
m odification to be efficiently integrated w ithin the acquiring firm s portfolio o f
technologies. T h e tacit nature o f know ledge associated w ith pro d u ction and in n o v
ation activity in these sectors also m eans that “ physical or geographical p ro xim ity
m ay be im p o rtan t fo r accessing and absorbing it (Blanc and Sierra 1999)- The
m arginal cost o f tran sm ittin g codified know ledge across geographic space does
not depend on distance, b u t the m argin al cost o f transm itting, accessing, and
328 r a jn e e s h n a r u l a AND ANTONELLO ZANFEi
ab sorb in g tacit know ledge increases w ith distance. T h is leads to the clustering of
in n o vatio n activities, especially in the early stages o f an in d u stry life cycle where tacit
know ledge plays an im po rtan t role (A udretsch and F eld m an 199b).
In general, asset-exploiting activities are p rim a rily associated w ith demand-
d riven in n ovative activities (e.g, lo calization o f the p a ren t-firm products for
a specific offshore m arket). A sset-au gm en tin g activities, o n the other hand,
are p rim a rily undertaken w ith the intention to acq u ire an d internalize technological
spillovers that are host lo cation -sp ecific. A sset-ex p lo itin g activity, broadly
speaking, represents an extension o f R & D w o rk u n d ertak en at h om e, while asset-
augm en tin g activity represents a d iversificatio n into n ew scien tific problem s, issues
o r areas.
A n extensive literature has suggested that asset-au gm en tin g internationalization
o f R & D has becom e m o re significant d u rin g the past tw o decades as a result of
several factors that include: ( a) the increasing costs and c o m p le x ity o f technological
d evelopm ent, leading to a grow in g need to exp an d tech n o lo g y so u rcin g and inter
action w ith different and geograph ically dispersed actors en d o w ed w ith comple
m en tary bits o f know ledge; ( b) the faster pace o f in n o vative activities in a num ber of
industries, sp u rrin g firm s to search fo r ap p licatio n o p p o rtu n ities w h ich are mainly
location -specific; (c) grow in g pressures fro m h o st g o vern m en ts, w hich have led
M N E s to increase the interaction w ith local partn ers as k ey co n d itio n s to gain access
to foreign m arkets.
A lth ough the conceptual differences betw een these tw o m otives fo r offshore R&D
investm ent are clear, indicators o f the im p o rtan ce o f these tw o m o tives are scarce.
U ntil recently, m ost em pirical studies o f in tern atio n al R & D investm ent (Mansfield
et al. 1979; Tail 1979; W arrant 1991) reflected the v iew that the role played b y foreign
R & D units was determ ined b y m arket or d em an d -sid e facto rs, i.e., asset-exploiting
m otives were assum ed. M ore recent em pirical w ork, how ever, has focused on asset-
augm en tin g m otives fo r R & D investm ents. D etailed analyses carried o u t by Miller
(1994), O dagiri and Yasuda (1996), and F lo rid a (1997) argu e that technology
sourcing strategies play an im p o rtan t role in a n u m b er o f m a n u fa ctu rin g industries
in N orth A m erica, E urope and A sia .10 Som e studies fin d that “ m arket-oriented”
R & D units established for asset-exploiting m otives have evolved into asset-aug
m enting ones (R ondstad t (1978 ). B u t oth er foreign R & D units experien ce no major
shift in their characters (K uem m erle 1999).
Several studies have used m ultivariate techniques to id en tify the relative import-
ance o f asset-augm entin g vs. asset-exploiting m otives fo r o ffsh o re R & D investment.
U sing patent citations A lm eida (1996) fou nd that foreign firm s in the sem iconductor
in d u stry not only learnt m ore from local sources, but they did so to a greater extent
than their dom estic counterparts. T h is stu d y also fo u n d that, w ith the significant
exception o f subsidiaries o f Japanese M N E s, foreign firm s locate their technological
activities overseas in areas where these firm s exhibited a h om e c o u n try disadvantage
G L O B A L I Z A T I O N OF I N N O V A T I O N 329
In one of the most extensive empirical exercises to date, Le Bas and Sierra (2002)
studied the R&D investment strategies o f the 345 MNEs with the greatest patenting
aflhdty in Europe between 1988 and 1996. These companies, which accounted for
about one half o f total patenting through the European Patent Office (EPO) over
this period, were predominantly of US, European, or Japanese origin.
To measure the technological strength o f companies and locations, the authors used
a patent-based indicator ("Relative Technolological Advantage" RTA). For a company,
HomeRTA is defined as the firm's share o f total European patents in a particular
technological field relative to its overall share o f all European patents. Patents from
"'".'foreign affiliates o f the firm (filed from outside the country in question) were excluded
v fro ip the calculation. For a location (country) in which a given firm has invested,
/.^HostRTAIs defined as the host country's share o f all European patenting in that field,
divided by its share of all European patents in all fields. In all cases an RTA> 1 signals a
>;y $efeive advantage o f the country (firm). Based on these definitions, four different R&D
identified:
330 RAJNEESH NARULA AND A N T O N E LIO ZANFEI
Source: adapted from Patel and Vega (1999, p. 152) and from Le Bas and Sierra (2002 p. 606).
The numbers in brackets indicate the frequency of the strategy in question for the
sample of firms considered. As is evident from the table, Le Bas and Sierra found that
the great majority of MNEs located their activities abroad in technological areas or
fields for which they were strong at home (strategies 3 and 4). However, the most
frequent strategy is clearly number 4, in which case not only the firm but also the host
country has a relative technological advantage (HostRTA > 1 ) . This may indicate the
formation of “centers of excellence” in which strong domestic research environments
function as global attractors.
12.4 F o r c e s S u p p o r t in g C o n c e n t r a t io n
a n d D is p e r s io n o f R&D
The literature on the location o f R& D activities views the location of MNEs
innovative activities as affected by centrifugal and centripetal forces that determine
whether the MNE centralizes (in the home location) or internationalizes to create
additional centers abroad. But all too often, this dichotom y— while substantially
correct presumes that the M N E has a single center in the first place. In order to
allow for the possibility that the M N E may have multiple home bases or several
locations of R&D concentration rather than a single “ hub", this section uses the
terms “ concentration" and “ dispersion,"
G L O B A L I Z A T I O N OF I N N O V A T I O N 331
We can single out at least fo u r b ro ad sets o f factors affecting the concentration and
dispersion o f innovative activities. These forces are active at both the m acro-level o f
countries, regions and system s o f firm s involved in the globalization o f innovation;
and the m icro-level o f in d ivid u al firm s and o f their internal networks o f innovati ve
activities across n ation al borders.
Innovation systems are built upon a relationship of trust* iteration* and імешгійэд
between firms and the knowledge infrastructure* within the framework of institution^
based on experience of and familiarity with each other over relatively long periods of
time. In engaging in foreign operations in new locations* firms which already face
opportunities and constraints created by their home innovations systems gradually
become embedded in the host environment The self-reinforcing interaction between
firms and infrastructure perpetuates the use o f a specific technology or technologies* or
production of specific products* and/or through specific processes. Increased ffed#*
ization often results in a systemic lock-in. Institutions develop that support шй
reinforce the interwoven relationship between firms and the knowledge infrastrucfore
through positive feedback* resulting in positive lock-in. When innovation systems
cannot respond to a technological discontinuity, or a radical innovation that has
occurred elsewhere, there is a mismatch between what home locations can provide
and what firms require, this is known as sub-optimal lock-in (Narnia 2002$).
In general, national innovation systems and industrial and technological specializa
tion of countries change only very gradually, and— especially in newer, rapidly evolv
ing sectors— much more slowly than the technological needs of firms. In other words,
there may be systemic inertia. Firms have three options open to them (Narnia 2002a).
Firms may seek either to import and acquire the technology they need from abroad, or
venture abroad and seek to internalize aspects of other countries" innovation systems,
thereby utilizing an “exit” strategy. Of course, firms rarely exit completely, preferring
often to maintain both domestic and foreign presence simultaneously. There are costs
associated with an exit strategy. On the one hand, they would weaken their contact
with their home market and by so doing they might reduce their ability to absorb
external knowledge. On the other hand, it must suffer the costs of entry in another
location (in terms of effort, capital, and time), and firms may minimize this through a
cooperative strategy with a local firm. Developing alternative linkages and becoming
embedded in a non-domestic innovation system takes considerable time and effort.
They can also use a “voice” strategy which is to seek to modify the home-country
innovation system. For instance, establishing a collective R&D facility, or by political
lobbying. Firms are inclined towards voice strategies, because it may have lower costs,
especially where demand forces are not powerful, or where the weakness of the
innovation system is only a small part of their overall portfolio. But voice strategies
have costs, and may not be realistic for SMEs, which have limited resources and
political clout. Such firms usually cannot afford an “ exit” strategy either, and end up
utilizing a loyalty strategy, relying instead on institutions to evolve, or seeking to
free-ride on the voice strategy of industry collectives, or larger firms.
testing T h is factor m ay account for the frecjuent estab lish m en t o f b oth manufactur
ing and R & D plants close to app lications abilities in fo reig n telecommunications
m arkets (Ernst 1997)- lit other industries, how ever, a large v a rie ty o f international
linkages are required for R & D and in n o vatio n , as ap p ears to be particu larly the case
o f biotechnology (Arora and GambardeUa 1990).
Organizational Issues
Another micro-level determinant is associated with the difficulties of managing
cross-border R&D activities. It is not sufficient for the foreign affiliate to internalize
spillovers if it cannot make these available to the rest o f the MNE (Blanc and Sierra
1999). A dispersion o f R&D activities across the globe requires extensive coordin
ation between them, and particularly with headquarters, if they are to function
efficiently. This acts as a centripetal force on R&D, and accounts for a tendency of
firm s to locate R & D (or at least the m ost strategically sig n ifican t elem ents) closer to
headquarters.
Com plex linkages, both within the firm, and between external networks and
internal netw orks, require co m p lex co o rd in a tio n i f th ey are to provide optimal
benefits (Zan fei 2000). Such co ord in atio n m a y req u ire exp ertise, m anagerial and
financial resources that are m ost likely available to larger firm s w ith m ore experience
in transnational activity (Castellani and Z a n fe i 2004), Large firm s tend to engage in
both asset-augm enting and asset-exploitin g activities. In d eed , large M N E s may have
several sem i-au ton om o u s sister affiliates in the sam e lo catio n that operate in similar
technological areas. Lastly, M N E s tend to engage in p ro d u c tio n activities (whether
in the sam e or another physical facility) in the h ost lo catio n , an d this prompts a
certain level o f asset-exploiting activity. T h u s, an M N E in a given location may seek
to internalize spillovers from n on-related firm s and to e xp lo it in trafirm knowledge
transfers w ith in the sam e m u ltin atio n al gro u p (C risc u o lo et al. 2005).
12.5 In n o v a t io n t h r o u g h In t e r n a t io n a l
S t r a t e g ic T e c h n o l o g y P a r t n e r in g
The previous sections have discussed the g ro w in g in tern atio n al d im en sio n o f R&D,
concentrating on the in tra -M N E aspect o f this d evelo p m en t. H ow ever, not all
in n o vato ry activity is undertaken w ithin hierarchies; d u rin g the last two decades,
non-internal R & D activities that rely on interfirm cooperative agreem ents have
grown rapidly in num ber.
Fully exam in in g the role o f (international) netw orks m the generation and
diffusion o f in n o vatio n is beyon d the scope o f this chapter (see Powell and G rodal,
Ch. 3 in this volu m e, fo r a m o re com prehensive discussion). A key issue for this
discussion is w h eth er and to w hat extent there is substitution or com plem entarity
between internal in n ovative activities and technological collaborations on a global
scale,
In som e circum stan ces, international S T P m ay substitute for internal innovative
activities. O ne such circu m stan ce is that o f R & D alliances aim ing to enter foreign
markets protected b y n o n -ta riff barriers, as is the case for environm ental regulations
in the chem ical industry. N onetheless, there are lim its to how m uch a firm can
substitute S T P fo r in -h o u se R & D , and by extension, international STP for overseas
R& D facilities.14 S T P tends to develop in areas in w hich partner com panies share
com plem entary capabilities, and these alliances create a greater degree o f interaction
between the p artn ers' respective paths o f learning and innovation (M ow ery et al.
1998; C antw ell and C o lo m b o 2000; Santangelo 2000).
One w ay to lo o k at this issue is to tackle the problem o f firm size, technological
capabilities, and co llab o ration s. P articipation in STPs tends to be correlated w ith
firm size in tech n o logy-in ten sive sectors. In these sectors, cooperation is a w ay to
keep up with the tech n o logical frontier: by associating com plem entary resources and
com petencies, it m akes it possible to explore and exploit new technological o p p o r
tunities. But sm aller tech n o logy-b ased M N E s also are involved in such agreem ents,
and their grow in g significance raises n u m erou s conundrum s (N arula 2002b; see also
Ch. 5 b y Lam , an d C h . 3 b y Pow ell and G rod al, this volu m e). Firm s— regardless o f
size— m ust m ain tain a g ro w in g breadth o f technological com petences, and this m ay
require p articip atio n in in tern ation al internal and external networks. SM Es need to
rely on n o n -in tern al sources, as they often experience w ider gaps in term s o f
com petencies and d evelo pm en t abilities than their larger counterparts (Zanfei
1994) but m ust be m o re skilfu l at m anaging their po rtfolio o f technological assets,
because they have lim ited resources (N aru la 2002b). Indeed, the costs o f m anaging a
web o f cross-b ord er agreem ents h igh ligh t the im portance o f transaction-type o w n
ership advantages fo r the M N E . T h is com plem en tarity between firm size, tech no
logical capabilities and the d evelopm ent o f innovation netw orks is consistent w ith
some o f the trends h igh ligh ted in Section 12,2.2. In particular, the geographical
concentration o f S T P activ ity w ith in the Triad reflects inter alia the fact that firm s
from these areas ten d to be larger and account for a m ajor share o f R & D activity.
The issue o f complementarity or substitution between the internal and non
internal innovative activities of MNEs can also be examined by looking at the
interdependencies between multinational expansion and international STP. Draw
ing on the transaction-cost literature, several works on international market entry
strategies argue that multinational experience may lower the risks faced by an MNE
336 RAJNEESH NARULA AND ANTONELLO 2 A N FE I
presence in som e em erging econom ies such as K orea, T aiw an, And Singapore
(H o b d a y 2000; L im 1999)? and som e EU m em b er states (B a rry and Strobl 2002;
C astellani and Z an fei 2003). H ow ever the evidence in the case o f m o st developing
countries does not p o in t to significant spillovers (see H arriso n 1999)* Indeed,
accord in g to a recent su rvey on econ om etric studies o f p ro d u c tiv ity spillovers
fro m F D I, the n u m ber o f studies in w h ich negative o r n o n -sig n ifican t results are
obtained is ap p ro xim ately as h igh as cases w here p o sitive sp illo vers w ere observed
(G o rg and Strobl 2001). T h is suggests a cau tiou s ap p ro ach to this issue, and calls for a
refinem ent o f analytical tools (see B o x 12.3). T h ere is a need to develop more
app ro p riate m easures o f technological spillovers, w h ich are n o t p ro p e rly captured
b y p erfo rm an ce in dicators like pro d u ctivity. T h e channels th ro u g h w h ich spillovers
occu r also need to be exam ined m o re carefully, i f F D I-related sp illovers are to be
exp licitly used as m eans for tech nological u p grad in g.
A n o th er p o sitio n in this p o licy debate argues that the in tern atio n alizatio n o f R& D
m ay lead to a “ h o llo w in g o u t” o f the h om e co u n try's in n o v a to ry cap acity when the
dom estic in n o vation system does not m eet the needs o f firm s in certain industries.
One of the strongest and most popular arguments in favor of inward investment as a
vehicle for local technological upgrading is that foreign firms usually outperform
domestic ones (see Beliak 2 0 0 2 for a review on empirical evidence on this aspect).
The underlying policy issue is whether or not foreign presence can generate techno
logical opportunities for the local economy. There is a dear connection here to the
literature on technology gaps and catching up (Godinho and Fagerberg, Ch. 19 in this
volume). On the one hand, some works suggest that the larger the productivity gap
between host country firms and foreign-owned firms, the larger the potential for
technology transfer and for productivity spillovers to the former. This assumption,
can be derived from the original idea put forward by Findlay (1978), who formalized
technological progress in relatively “backward” regions as an increasing function of the
distance between their own level of technology and that of the “advanced regions,” and
of the degree to which they are open to direct foreign investment.
On the other hand, scholars have argued that the lower the technological gap
between domestic and foreign firms, the higher the absorptive capacity of the former,
and thus the higher the expected benefits in terms of technology transfer to domestic
firms. It is worth noting that the role of absorptive capacity is also implicitly recognized
in the catching up tradition, when it is acknowledged that a sort of lower bound of local
technological capabilities exists, below which foreign investment cannot be expected to
have any positive effects on host economies.16 The “ technological accumulation
hypothesis goes beyond this simplistic view of absorptive capacity and places a new
emphasis on the ability to absorb and utilize foreign technology as a necessary condi
tion for spillovers to take place.
G L O B A L I Z A T I O N OF I N N O V A T I O N 339
Although there is currently little evidence to support or refute the hollow ing out
hypothesis, this has been raised by po licy m akers in several countries, and represents
an im portant area for future research. The consequences o f a potential hollow ing out
m ay be especially significant in sm all open econom ies that are specialized around a
few products, and/or concentrated around a few large firm s. A noth er related and
potentially im p o rtan t area for future research is the need to distinguish between
hollow ing-out as a sym p tom o f su b -op tim al lock-in and the internationalization o f
innovation to supplem ent dom estic supply lim itations (N arula 2003). A fter all, no
country can p ro vid e w orld-class com petences in all technological fields. Even the
largest, m ost tech nologically advanced countries cannot provide strong innovation
systems to all their industries, and w orld-class com petences in all technological
fields. Som e coun tries regard im ported technologies as a sign o f national weakness,
and have sought to m ain tain and develop in -co u n try com petences, often regardless
o f the cost (N arula 2002a). R elyin g largely on in -co u n try com petences m ay how ever
lead to a sub -op tim al strategy, especially in this age o f m u lti-tech n ology products. In
fact, the cross-b ord er flo w o f ideas is fundam ental to firm s, and this im perative has
increased w ith grow in g cross-b ord er com petition, and international production.
N otes
1. Both changes in the composition o f world trade, and sectoral correlations between
R8cD intensity and internationalization should be considered with caution since
definitions o f industries change over time (see Von Tunzelman and Acha, Ch. 15 in this
volume).
2. For instance, press releases are often used to construct data-sets, and these are not always
factual, sometimes reflecting the public relations objectives o f the firms; the coverage of
large firms is higher than for smaller firms; STP failures are not reported as accurately (or
as often) as STP formation; large databases are hard to update and are frequently subject
to changes in the methodology o f data collection over time.
3. STP refers to interfirm cooperative agreements where R8cD is at least part o f the collabora
tive effort, and which are intended to affect the long-term product-market positioning of
at least one partner.
4. Strictly speaking, the two numbers are not comparable, because GDP is a flow figure.
Nonetheless, it is generally accepted that FDI stock is a monotonic function o f value
added, so the change in this ratio gives a general idea of how the significance o f FDI
activities has changed.
5. Paradoxically, perhaps, this group also includes Swedish MNEs, whose much higher
shares o f offshore R&D and patenting throughout the twentieth century, nevertheless,
display a sharp drop after 1940 and a recovery by 1969—90 to a share that is lower than that
o f 1920-39.
340 RAJ N E E S H N A R U L A AND A N T O N E L L O Z A N F E I
their view w ould reduce uncertainty about the foreign market. Controlling for a num ber
o f sources o f heterogeneity, they show that this factor is positively correlated with the
creation o f new subsidiaries and o f equity agreements. By contrast, what they call
“variety experience," reflecting the heterogeneity and geographical dispersion o f markets
where a M N E is active, should increase the firm's exploratory capacity. They find that, in
the exam ined industry, variety experience has a positive and significant im pact on non
equity technical alliances.
16. As Findlay (1978:2-3) notes: “ Stone age communities suddenly confronted with modern
industrial civilisation can only disintegrate or produce irrational responses. . . Where the
difference is less than some critical minimum, admittedly difficult to define operation-
ally, the hypothesis does seem attractive and worth consideration." Findlay also observes
that the educational level o f the domestic labour force, which is a good proxy for what is
currently named country s “ absorptive capacity," might also affect, inter alia, the rate at
which the backward region improves its technological efficiency (Findlay 1978: 5-6).
R eferen ces*
* P e a r c e , R. (1990)? The In terna ti0 na1isa tio n of Resditch and Development, London: Mac-
millan.
------ (1999), “ D ecentralised R & D and Strategic Com petitiveness: Globalized Approaches to
Generation and use o f Technology in M ultinational Enterprises (M N Es),” Research Policy
28(2-3): 157-78,
P isa n o , G, (1990)>The R & D Boundaries o f the Firm : An Em pirical Analysis,” Adminis
trative Science Quarterly 3 5 :15 3 -7 6 .
* R o n s t a d t , R. C. (1978), “ International R & D : The Establishm ent and Evolution o f Re
search and D evelopm ent A broad by Seven U S M ultinationals,” Journal of International
Business Studies 9(1): 7-24.
S a n t a n g e l o , G . (2000), “ C orporate Strategic Technological Partnerships in the European
Inform ation and C om m un ication s Technology Industry,” Research Policy 2 9 :10 15 -3 1.
S e r a p io , M ., and D a l t o n , D. (1999), “ Globalisation and Industrial R & D : A n Exam ination
o f Foreign Direct Investm ent in R & D in the United States,” Research Policy 28: 303-16.
S io tis , G. (i999)> “ Foreign D irect Investm ent Strategies and Firm s' Capabilities,” Journal of
Economics and Management Strategy 8(2): 251-70,
S joholm , F, (199b), “ International Transfer o f Knowledge: The Role o f International Trade
and G eographic Proxim ity,” Weltwirthschaftliches Archiv 132: 97—115.
St o p f o r d , JL M ., and W e l l s jr., L. T (1972), Managing the Multinational Enterprise: Organ
isation of the Firm and Ownership of SubsidiariesyN ew York: Basic Books.
* T e e c e , D. J. (1986), “ Profiting from Technological Innovation: Im plications for Integra
tion, C ollaboration, Licensing and Public Policy,” Research Policy 15: 285-305.
------ (1992), “ C om petition , C ooperation and Innovation. O rganizational Arrangem ents for
Regimes o f R apid Technological Progress,” Journal of Economic Behavior and Organization
18 :1-2 6 .
------ (1996), “ Firm O rganisation, Industrial Structure and Technological Innovation,” Jour
nal of Economic Behavior and Organization 3 1:19 3 -2 2 4 .
------ R u m e l t , R ,, D o s i , G ,, and W in t e r , S. G. (1994), “ U nderstanding Corporate C oher
ence: T h eo ry and Evidence,” Journal of Economic Behavior and Organization 2 3 :1- 3 0 .
U N C T A D (2001), World Investment Report: Promoting linkages, N ew York: United Nations.
— — (2002), World Investment Report: TNCs and Export Competitiveness, N ew York: United
Nations.
V e r n o n , R. (1966), “ International Investm ent and International Trade in Product C y c le ”
Quarterly Journal of Economics 8 0 :19 0 -2 0 7 .
V e u g e l e r s , R ., and C a s s im a n , B. (1999). “ M ake and Buy in Innovation Strategies: Evidence
from Belgian M an ufacturin g Firm s,” Research Policy 28: 63-80.
W a r r a n t , E (1991), he Deploiement mondial de la R&D industrielle. Brussels, C om m ission
des C om m unautes Europeennes— Fast, December.
Z a n d e r , I. (1995), The Tortoise Evolution of the Multinational Corporation—Foreign Techno
logical Activity in Swedish Multinational Firms 1890-1990, Stockholm : IIB.
* ---- (1999), “ H o w do you M ean ‘Global*? An Em pirical Investigation o f Innovation
N etw orks in the M ultinational C orporation,” Research Policy 2 8 :19 5 -2 13 .
Z a n f e i , A. (1994), “ Technological Alliances Between W eak and Strong Firm s: Cooperative
Ventures w ith A sym m etric Com petences,” Revue dyEconomic Industrielle 7: 255-79.
* ---- (2000), “ Transnational Firm s and the Changing O rganisation o f Innovative A ctiv
ities,” Cambridge Journal of Economics 2 4 :5 15 -4 2 .
In t r o d u c t i o n to P a r t III
HOW
INNOVATION
DIFFERS
In tr o d u ctio n to P a r t III
# # ♦ # * ♦ ♦ * * ♦ * * * # # » #**********»**•******•*******
INNOVATION
T H R O U G H TI ME
K R IST IN E B R U LA N D
D AVID C. M O W E R Y
13.1 In t r o d u c t i o n
form o f this is the w ave th eo ry p ro p o sed b y Sch u m p eter in Business Cycles, in which
steam po w er drove the First In du strial R evo lu tio n , electricity the Secon d Industrial
R evo lu tio n , an d so on. O ther w o rk that does n o t rely on w ave th eo ries also stresses
the role o f a sm all n u m ber o f technologies in d riv in g b ro ad er processes o f econom ic
grow th. A lth o u gh valuable, m an y o f these fram ew o rk s o verem p h asize the im port
ance o f the allegedly critical technologies w h ile sligh tin g o th er areas o f innovation
and econ om ic activity that are no less im p o rtan t. In w h at fo llo w s w e challenge some
o f the h istorical discussions that stress the tran sfo rm ative effects o f “ critical innov
a tio n s ” Instead, w e em phasize the co m p lex m u ltisecto ral ch aracter o f innovation,
and hence the need to take seriou sly the coexistence o f a ran ge o f in n o vatio n modes,
in stitu tion al processes, and organizational form s.
O u r discussion o f in n o vatio n th rou gh tim e h igh ligh ts chan ges in the structure
o f the in n o vatio n process in successive perio d s, and is In fo rm ed b y the innovation
system concept (discussed in Ch. 7 b y E dqu ist, C h . 11 b y A sh e im an d Gertler, and
Ch. 14 b y M alerb a). In ad optin g this fram ew o rk, we fo cu s o n the ch an gin g structure
o f econ om ic activity, changes in relevant in stitu tion s, an d ch an ges in patterns o f
know ledge generation and flow s w ith in em ergent in d u strial e c o n o m ie s.1
We begin the discussion below b y review ing recent historical interpretations o f
the im pulse tow ard industrialization in the w orld econom y. W e then discuss the
changing structure o f the innovation process in different phases o f industrialization,
focusing on the First Industrial Revolution in B ritain from rou gh ly 1760 to 1850, the
so-called Second Industrial Revolution during the late nineteenth and early twentieth
centuries, and w hat m ight be called the T hird Industrial R evolu tion after W orld War
II. O ur discussion thus includes the w idespread appearance of shop-
floor-driven technological innovation in eighteenth-century G reat Britain and
m oves forw ard to consider the invention o f the art o f invention, to use the philosopher
A. N . W hitehead s phrase (W hitehead 1925), in the late nineteenth an d early twentieth
centuries, w ith the em ergence o f organized industrial R8cD w ith in the firm . The Third
Industrial Revolution, m ost clearly illustrated by the p o st-w ar U nited States, is one in
w hich private and public institutions com pete and collaborate In new fields o f
innovation, a m ode o f innovation that is not yet exhausted.
13.2 T he F i r s t In d u s t r i a l R e v o l u t i o n
sustained p ro d u ctivity grow th, in the late eighteenth century* W hy were som e
human societies able to break out o f a M althusian trap, shifting from 'ex ten sive ”
economic grow th that relied on increased labor input and a w ider division o f labor
to innovation-based intensive grow th w ith sustained rises in real output per head?
An im portant con trib u tion to this historical debate is Pom eranz (2000), who argues
that prior to the m id-eighteenth century, Europe, Japan, China, and India were at a
broadly sim ilar level o f econom ic developm ent— this was "a w orld o f surprising
resemblances*” W hy did o n ly N orthw estern Europe m ake the transition to in n o v
ation-based growth? Pom eranz suggests that tw o factors were crucial: the acquisi
tion by the m ajor E u rop ean pow ers o f colonies as m arkets for m anufactures and
sources o f fo o d and raw m aterial, and the developm ent w ithin Europe o f coal as a
new energy sou rce.2
An alternative exp lan ation for the industrialization o f N orthw est Europe stresses
institutional changes (see Braudel 1984; W allerstein 1974; Landes 1998), focusing on
the em ergence o f p ro p erty rights as im pulses to innovation. A varian t o f this insti
tutional analysis is pro vid ed b y Jones (2003) who argues that technology-based
growth has occu rred at several points in w orld h istory; the challenge is less to
understand grow th than to understand the forces that prevent grow th. He stresses
the inh ibitory role o f political institutions that are based on surplus extraction by
political and m ilita ry elites. O nly w hen such rulers are weakened b y crisis do
opportunities arise for gain from innovation. Since the political pow er o f established
political elites in N orthw est E urope eroded during the fourteenth to seventeenth
centuries, the em ergence o f sustained, innovation-based econom ic grow th first
occurred in this region o f the w o rld econom y.
There is disagreem ent w ith in this literature over the tim in g o f the divergence, as
well as the relative im po rtan ce o f different factors in su p portin g the grow th o f such
institutions as private p ro p erty rights and the weakening o f rent-seeking political
and m ilitary elites. B u t all o f the scholars adopting this approach em phasize insti
tutional change as an indispensable precondition for sustained innovation-led
growth.
o f in n o vation studies have follow ed the *key in n o va tio n s in terp retatio n o f the First
In dustrial R evo lu tio n (e.g. Freem an an d Lou^a 2002; fo r an e co n o m ic h istory o f
in d u strializatio n in this fram ew ork, see L lo y d -Jo n e s an d L ew is 1998).
Fo u r in n o vation s the sp in n in g jenny, the w ater fram e, the sp in n in g m ule, and the
autom atic mule- were associated w ith d ram atic grow th in the B ritish textiles
INNOVATION THROUGH TIME 353
industry o f the First Industrial R evolution. Between the late eighteenth and the
middle o f the nineteenth century, the cotton textile indu stry grew spectacularly in
the size o f its output, in lab or productivity, in the scale o f enterprises, in capital
employed, and in its share o f national incom e. Value added in cotton rose from less
than £500,000 in 1760 to about £25,000,000 by the m id-i82o$. In spinning, the
num ber o f direct lab or h ou rs required to process 100 pounds o f cotton declined
from 300 in 1790 to 135 in 1820 (M o kyr 2002: 5 0 -1), and the average annual input o f
raw cotton per facto ry rose by over 1,000 per cent during 1797-1850 (C hapm an, 1972:
70). D ram atic as these changes were, they should be kept in proportion: textiles
made up about 25 per cent o f m anu factu ring output at their peak. In novation and
productivity w ere grow in g elsewhere as well.
Another critical in n ovation o f this period was the steam engine o f Jam es Watt,
first introduced in 1775. W att s innovation is com m on ly described as the em blem atic
technology o f the In dustrial R evolution (see Toynbee 1908; Deane 1965). Yet von
Tunzelm anffs study o f steam pow er (1978) show ed that the m achine diffused
relatively slowly, that it had only m odest econom ic advantages over existing pow er
technologies (and hence could not significantly affect econom ic grow th), and had
limited backw ard and fo rw ard linkages w ith the rest o f the British econom y, further
reducing its “ catalytic” effects (see B o x 13.1). As we noted earlier, the innovations that
co rp o rate firm s; and the rise o f m an agerial con trol o f p ro d u c tio n , w h ich trans
fo rm ed w orkplace o rgan ization and scale. These in stitu tio n al in n o v a tio n s together
m ad e possible the subsequent grow th in facto ry p ro d u ctio n .
M o st in du strial enterprises o p eratin g d u rin g the eigh teenth cen tu ry were
extrem ely sm all. Large-scale factories w ere u n co m m o n b efo re the early nineteenth
century, and the sm all-scale w o rk sh o p o r p ro d u ctio n u n it w as the p rim a ry organiza
tional fo rm for m ost o f the p erio d o f early in d u strializatio n . T h ese sm all firm s were
in d ivid u ally ow n ed o r were partnerships, lo cally fin an ced , in w h ich liab ility for
debts was the personal resp on sib ility o f ow n ers w h o u su ally acted as m anagers. Two
in stitution al fo rm s m ade possible an exp an sio n in the scale o f enterprises: joint-
stock (i.e. lim ited liability) organ izatio n an d the grow th o f fin an cial netw orks.
Jo in t-sto ck association s em erged in the m edieval p e rio d in B ritain , but were
perm itted o n ly via the explicit au th orization o f the state. A series o f piecemeal
reform s after 1825 w ere follow ed b y legislation p erm ittin g the creatio n o f com panies
w ith separate legal identity, lim ited liab ility an d tradeable shares. G en eral legislation
fo r the jo in t stock fo rm w as passed in 1844 and co n so lid ated in the statutes o f 1856
and 1862 (M ath ias 1983:325; see H arris 2000 for a co m p reh en sive acco u n t). Although
m uch in du strial fin an cin g rem ained local and sm all in scale (see H u d so n 1986 for an
account o f local n etw orks5 role in fin an cin g the w o o llen in d u stry ), these legal
reform s enabled substantial grow th in the fin an cin g an d scale o f in d u strial enter
prises. But jo in t stock o rgan ization and access to finance w ere n ecessary rather than
sufficient co n d itio n s for enterprise grow th. Even m o re sign ifican t w as the develop
m ent o f m anagem en t system s and m an agerial control.
M anagers o f these early indu strial enterprises co n fro n ted serio u s challenges in the
assem bly and m aintenance o f a suitable w o rk fo rce, the co n tro l o f w o rk , and the
ad o p tion o f new techniques and organ ization al stru ctu res fo r p ro d u c tio n activities
by a restructured w orkforce. Pollard h igh lights “ tw o distinct, th o u g h clearly overlap
p in g difficulties; the aversion o f w ork ers to enterin g the n ew large enterprises with
their unaccustom ed rules and discipline and the sh ortage o f skilled and reliable
lab o u r (Pollard 19 6 5 :16 0 ). The em ergence o f rule-b ased d isc ip lin a ry m ethods, the
lab orio u s constru ction o f su p erv iso ry system s, and the h ab itu atio n o f w orkers to an
organized and controlled w o rk in g day em erged slo w ly b u t w ere central develop
m ents ot early industrialization. N ew m anagem en t tech n iq u es that app eared during
the In dustrial R evo lu tio n perm itted the d evelopm ent o f larger, centralized pro d u c
tion sites and o f the m echanized factory. In turn, such sites p erm itted the application
o f pow er, the ad o p tion o f new indu strial techniques, an d closer m an agerial control
over the organization and pace o f w ork.
These organizational and m anagerial in n o vatio n s w ere d efin in g characteristics o f
the First In dustrial R evolution. In p o ttery fo r exam ple, the m o st im p o rta n t m an
agerial in n o vato r was Josiah W edgw ood, w h o d eveloped a n u m b er o f product
in n o vation s— new designs, new glazes and finishes, and n ew basic m aterials— and
pioneered new m arketing m ethods. But his m ost im p o rta n t in n o vatio n s were
INNOVATION THROUGH TIME 357
In the second half o f the eighteenth century, rising incomes and increased coffee and
tea consumption accelerated growth in the market for china and other types o f glazed,
fired clay plates, cups, and related items. This was part o f a wider growth In demand for
"lu xu ry5 consumer goods (Berg and Eger 2003). The production o f pottery was
concentrated in Staffordshire in central England, and was dominated by small enter
prises operated by craftsmen, often producing on a piecework basis. Production was
controlled by individual craftsmen, and production rhythms and volumes were hap
hazard. Josiah Wedgwood transformed the industry by developing factory-based
production techniques that supported the creation o f an enterprise o f unprecedented
scale. Wedgwood's success rested on two achievements. First, he successfully lobbied
the British government to improve regional transportation infrastructure (a publicly
financed turnpike was built in 1763 and a canal, on which Wedgwood sited his factory,
was completed in 1771), thereby enabling his factory to serve the British market while
reducing formerly exorbitant breakage rates. Second, he introduced radical organiza
tional innovations, developing new techniques for organizing production and man
aging the workforce (Bruland 1989).
Wedgwood, an acquaintance of Matthew Boulton, the entrepreneur who formed the
successful steam-engine firm of Boulton and Watt, modeled his new production
organization on Boulton’s factory, emphasizing a physical layout that separated and
sequentially organized the various operations that went into production of his china
(Langton 1984). Consistent with this organization, Wedgwood assigned workers to
specific tasks, relying on specialization to enhance skill and consistency in the per
formance of these tasks. Workmen "were not allowed to wander at will from one task to
another as the workmen did in the pre-Wedgwood potteries. They were trained to one
task and they had to stick to it” (McKendrick 1961: 32).
Having reorganized the structure of production and jobs within his organization,
Wedgwood had to develop techniques to encourage and/or force workers to adapt to
this new system. He invested heavily in the retraining o f experienced workers (with
mixed results) and in the training of new employees, many of whom were young
women (women accounted for 25 per cent of his employees as of 1790). Even more
important, however, was Wedgwood’s emphasis on codification o f technical guidelines
for the performance o f the various tasks in his factory and development of extremely
Леія\\рЛ written rules for worker behavior, Wedgwood also introduced sanctions and
rewards for punctuality and absenteeism on the part o f workers, going so far as to
develop an early prototype o f a timeclock for monitoring workers attendance.
358 K R I S T I N E R R U L A N D A N D D A V I D C. M O W E R Y
13.3 T h e S e c o n d I n d u s t r i a l R e v o l u t i o n
conducted such form al training and research increased; and (3) bodies o f em pirically
grounded, codified scientific and technological know ledge internal to the firm
became pow erful engines for expansion and diversification.
The technological shifts o f the late nineteenth century were accom panied by
changes in firm structure. Large-scale, vertically integrated enterprises em erged in
Germ any and the U nited States that incorporated specialized research and develop
ment departm ents or laboratories. W ithin such firm s scientific w ork was carried out
by teams o f researchers and depended on netw orks o f scientific contacts in the
education (particu larly university) system s. These professionally m anaged firm s o f
unprecedented size becam e the agents o f Schu m peter’s “ creative destruction” by the
m id-twentieth century, as industrial innovation becam e a core com ponent o f
corporate strategy.
tech nological leadership. T h e d evelopm ent o f this n ew m ach in e tech n o logy rested
o n m echanical skills o f a high order, as w ell as co n sid erab le in g e n u ity in conception
and design. It required little o r no recourse to the scien tific k n o w led ge o f the time,
an d U S success in this and oth er indu stries, such as ch em icals, m eatpacking, and
co n su m er go od s, relied on access to a large d o m estic m arket.
T h e creation o f a tru ly “ n atio n al" m arket in the U n ited States in turn was
facilitated b y the con stru ction o f a reliable n atio n al in fra stru ctu re fo r com m uni
cations and tran sportation . T h e enterprises m o st h ea v ily in v o lv ed in the creation of
this in frastru ctu re w ere them selves am o n g the largest in d u strial firm s organized in
the U nited States, and the organizational and fin an cial in n o va tio n s developed by
firm s such as W estern U n io n and the P en n sylvan ia R a ilro a d w ere w id e ly emulated in
other industries as new firm s o f unprecedented scale w ere created (C h an d ler 1977).
But few i f any o f these econ o m ically crucial o rg an izatio n al in n o va tio n s relied on
science.
The large, profitab le firm s that em erged in these scien ce-b ased in d u stries actively
lobbied the G erm an govern m en t fo r increased su p p o rt o f h ig h e r edu cation (The
“ C lu b o f G erm an C h e m ists;5 draw n largely fro m sen io r m an ag em en t o f German
chem icals firm s, lobbied fo r ad d ition al facu lty a p p o in tm en ts in ch em istry) and for
other fo rm s o f su p p o rt o f research related to their enterprises. W ern er v o n Siemens
d onated the land for the Im perial Institute o f Physics and T ech nology, located near
his firm 's h ead qu arters in B erlin and the city's tech n ical u n iversity, and the Institute
w as fo rm a lly established w ith pu blic fu nds in 1887. S im ila r lo b b y in g b y the chemicals
in d u stry led to the an n ou n cem en t in 19 10 b y the G e rm an e m p e ro r o f the foundation
o f the K aiser W ilhelm Institute for C h em istry, staffed larg ely b y acad em ic chemists
an d fu n ded b y industry. B o th the Institute o f Physics an d T ech n o lo gy and the Kaiser
W ilhelm Institute were dedicated to “ m issio n -o rie n ted ” fu n d am en tal research,
m uch o f w hich was lon ger-term in nature than the R & D p e rfo rm e d in industry
b u t nonetheless m ore applied than the w o rk o f u n iversity faculties (Beyerchen 1988).
The creation b y G erm an chem icals firm s o f in -h o u se in d u strial research labora
tories also w as associated w ith change in the m an agem en t an d stru ctu re o f these
firm s (see B o x 13.3). F am ily m anagers w ere replaced b y p ro fe ssio n al m anagers and,
eventually, b y p rofession al chem ists. T h eir in -h o u se R & D activities produced new
products in fields other than dyestuffs, e.g. B ayer s asp irin . A n d the im portance o f
The US chemicals firm Du Pont established its first industrial research facility, the
Eastern Laboratory, in 1902, and founded the Experimental Station in 1903. Creation
of the Eastern Laboratory followed the acquisition of control of the Du Pont Company,
founded in the early nineteenth century, by T. Coleman Du Pont, Pierre S. Du Pont,
and Alfred I. Du Pont from other family members. The Company’s transformation
from a loose holding company to a multifunction, diversified industrial corporation
began with this 1902 change in control.
Du Pont s Eastern Laboratory was the first laboratory to be physically and organiza
tionally separated from the manufacturing operations of the firm. Its R&D activities
were devoted almost entirely to improvements in manufacturing processes for Du
Pont’s existing product line of dynamite and high explosives. By contrast, the Experi
mental Station, founded one year later, focused on the development of new products
and improved applications of Du Pont’s smokeless-gunpowder products. The Experi
mental Station also monitored and evaluated inventions from sources outside of the
Du Pont Company.
AUS government antitrust suit against Du Pont forced the divestiture of a portion of
its black powder and dynamite businesses in 1913, and the firm used its R&D
laboratories to diversify its product lines through R&D and the acquisition of tech
nologies from external sources during and after World War I.
dose links between research personnel and the users o f these new products, as well as
the proliferation o f new products, triggered expansion in these firm s’ internal
distribution and m arketing capabilities. Just as U S firm s were d oing by the end o f
the nineteenth century, the G erm an chem icals firm s expanded their boundaries to
incorporate new fu n ction s and a m uch broader and m ore diversified product line.
A sim ilar sequence occu rred at rou gh ly the same tim e in the G erm an electrical
equipm ent industry, as Siem ens and A E G , am ong other leading firm s, established
in-house research laboratories d u rin g the 1870s and 1880s.
The developm ent o f indu strial research w ithin U S m anu factu ring firm s follow ed
these developm ents in the G erm an chem icals and electrical m ach in ery industries.
M any o f the earliest U S corporate investors in industrial R & D , such as General
Electric and A lcoa, w ere fo u n d ed on product or process innovations that drew on
recent advances in physics and chem istry. The corporate R & D lab orato ry brought
m ore o f the process o f developing and im p ro vin g industrial technology into the
boundaries o f U S m an u factu rin g firm s, reducing the im portance o f the independent
inventor as a source o f patents (Sch m ookler 1957).
But the in -h ou se research facilities o f large U S firm s were not concerned exclu
sively w ith the creation o f new technology. Just as the G erm an dyestuff firm s
laboratories had, these U S indu strial laboratories also m on itored technological
developm ents outside o f the firm and advised corporate m anagers on the acq uisi
tion o f extern ally developed technologies.
K R I S T I N E BRULAND AND DAVI D C. M O W E R Y
13.4 A “ T hird In d u s t r i a l R e v o l u t i o n ”?
R&D a n d I n n o v a t i o n d u r i n g
the posT-1945 Period
post-1945 period in the U nited States were transform ed by the dram atic increase in
central governm ent spen ding on R & D , m uch o f which was allocated to indust ry and
academic research. A s Lu ndvall and B orras point out in C hapter 22 o f this volum e,
this expansion in pu blic R & D fu nding was m otivated prim arily by national security
and public-health concerns and secondarily by political support for basic research.
The post-w ar U S R & D system differed from those o f other industrial econom ies in at
least 3 aspects: (1) sm all, new firm s were im portant entities in the com m ercialization
o f new technologies; (2) defense-related R & D fu nding and procurem ent exercised a
pervasive influence in the h igh -tech n ology sectors o f the U S econom y; and (3) US
antitrust p olicy d u rin g the p o st-w ar period was unusually stringent.
The prom inence o f new firm s in com m ercializing new technologies in the post
war United States contrasts w ith their m ore m odest role during the interw ar period.
In industries that effectively did not exist before 1940, such as com puters, sem icon
ductors, and biotechnology, new firm s were im portan t actors in R & D and com m er
cialization, in contrast to po st-w ar Jap an and m ost Western European econom ies.
Moreover, in both sem icond u ctors and com puters, new firm s grew to positions o f
considerable size and m arket share.
Several factors contribu ted to this prom inen t role o f new firm s in the post-w ar U S
innovation system . The large basic research establishm ents in universities, govern
ment, and a n u m b er o f private firm s served as im portant “ incubators” for the
developm ent o f in n ovation s that “ w alked out the d o o r” w ith individuals w ho
established firm s to com m ercialize them . Relatively w eak form al protection for
intellectual p ro p erty d u rin g the 1945-80 period paradoxically also aided the early
growth o f new firm s. C o m m ercialization o f m icroelectronics and com puter h ard
ware and softw are in n o vation s b y new firm s was aided b y a perm issive intellectual
property regim e that facilitated tech nology diffusion and reduced the burden on
young firm s o f litigation over inventions that originated in part w ithin established
firms. In m icroelectronics and com puters, liberal licensing and cross-licensing
policies were b yp ro d u cts o f antitrust litigation, illustrating the tight links between
these strands o f U S governm ent policy. U S m ilitary procurem ent policies also
contributed to the grow th o f new firm s in m icroelectronics and contributed to
high levels o f tech nological spillovers am ong these firm s.
The progress o f co m p u ter technology since the 1950s has been driven by the
interaction o f several trends: (1) dram atic declines in the price—perform ance ratios
o f com ponents, inclu ding central processing units and such essential peripherals as
data storage devices; (2) resulting in part from (1), the rapid extension o f com puting
technology into new applications; and (3) the increasing relative costs o f software.
Throughout the b rie f h isto ry o f the electronic com puter industry, these trends have
created bottlenecks that have influenced the path o f technological change. The I BM
360 m ainfram e com puter, for exam ple, w hich cem ented IB M 's dom inance o f the US
com puter in d u stry d u rin g the 1960s and 1970s, created a 'p ro d u c t fam ily” o f
com puters in different perform an ce and price classes that all utilized a com m on
operating system and other software.
The Intel C o rp o ra tio n s com m ercialization o f the integrated circuit m icroproces
sor in 1971 tran sform ed the structure o f the U S com puter indu stry during the next
twenty-five years. D evelopm ent o f the m icroprocessor at Intel resulted from a search
for an integrated circuit that could be used in a w ide range o f applications. Rather
than designing a cu stom "ch ip set” for each application, the m icroprocessor m ade it
possible for Intel to p rodu ce a pow erful, general-purpose solution to m any diverse
applications, breakin g a bottleneck that lim ited technological progress and
diffusion.
The m icrop rocessor was the fou n d ation for a new generation o f com puting
technology that tran sfo rm ed the IC T industry. D esktop com puters diffused rapidly
throughout m ost indu strial econom ies, and the sheer size o f the resulting "installed
base” radically changed the econom ics o f the com puter hardw are and software
industries. C o m p u te r softw are, fo rm erly dom inated by relatively sm all independent
vendors o r subsidiaries o f established com puter system s producers, becam e a
m ass-m arket in d u stry that supported entry b y a flood o f new firm s, som e o f
which proved to be en o rm o u sly profitable. Established producers o f com puter
hardware w ere severely affected b y the rise o f desktop system s, which encroached
rapidly on m arkets fo r m ain fram e and m inicom puters. B y the end o f the twentieth
century, o n ly a few o f the d om in an t com pu ter system s firm s o f the 1970s rem ained
active.
The final dramatic transformation in computing technology, the Internet,
depended on the appearance of the desktop computer (see Box 13.4). The develop
ment of computer networking and the Internet began during the 1960s and was
sponsored by governments in the United States, France, and the United Kingdom.
The large-scale and early commitment to deployment of computer networks in the
United States, which relied entirely on public funds, as well as the rapid diffusion of
desktop computers in the United States during the 1980s, created a huge domestic
network by the late 1980s. Although the HTTP and HTML "hypertext” software
protocols were invented at the European research center, CERN, in 199L A e A st
commercial application of these protocols as part of a "Web browser” emerged from
a U S u n iversity in 1993.
370 K R I S T I N E BRULAND AND DAVID C. MOWERY
Com m ercial exp loitation o f the Internet proceeded m ost rapidly In the United
States during the rem aind er o f the 1990s. Like previous U S innovations in ICT, the
Internet's developm ent d rew on a m ix o f public and private R & D funding, as well as
entrepreneurial new firm s seeking to com m ercialize new applications and an active
(and entrepreneurial) c o m m u n ity o f university researchers. M oreover, the Internet's
commercial developm ent, as well as the developm ent o f desktop systems and
software in the U nited States, benefited from the enorm ous size o f the U S m arket
and the large dom estic installed base o f desktop com puters that was rapidly co n
nected to the Internet. In n o vatio n in the Internet benefited from the participation
o f millions o f users in d eveloping o r refining new applications, just as had been the
case with desktop co m p u ter softw are d u rin g the 1980s. In other w ords, the develop
ment o f one o f the m a jo r industries o f the late twentieth and early twenty-
first centuries benefited fro m the large size o f the U S dom estic m arket, a central
feature o f U S econom ic developm ent that has been prom inen t since the nineteenth
century.
industries in electronics, no less than in chem icals, pharm aceuticals, or autom obiles,
has been transform ed.
Natural resources p er se now play a less central role, particularly b y com parison
with dom estic stocks o f h u m an capital that can be expanded through pu blic invest
ments in education and train in g (see Ch. 7 b y Edquist in this volum e). In addition,
many o f the historic scale-based advantages enjoyed b y U S firm s in m anufacturing
industries were eroded d u rin g the post-1945 period by the revival o f international
flows o f trade and capital that m ade it possible for sm aller nations to achieve scale
economies through the exp ort o f their products. Expanded trade and capital flows
have also spurred grow th in cross-b ord er flow s o f technology and know h ow (see
Ch. 12 b y N aru la an d Z an fei in this volu m e). This transform ation m ade it possible
for nations such as Taiw an and South Korea, w ith low pre-1945 levels o f industrial
development (in contrast to Jap an , which had created a relatively sophisticated,
albeit m ilitarized, in d u strial base b y the 1930s) and relatively m odest endow m ents o f
natural resources, to “ catch u p ” w ith the industrial econom ies (see Ch. 19 by
Fagerberg and G o d in h o in this volu m e).
13.5 C o n c l u sio n
H istory rarely presents neat lessons for generalization, and the historical study o f
innovation is no exception. T h e p rim a ry lessons from o u r historical discussion
concern the heterogen eity o f the in n ovation process across tim e, across sectors,
and across countries. M u ch o f the su rvivin g historical evidence that guides the
historical study o f in n o vatio n tends to highlight the form al and obscures the
inform al processes o f know ledge accum ulation, learning, and dissem ination that
underpin technological change and that contribute to its econom ic benefits. A n
im portant area fo r fu rth er research is the enrichm ent o f o u r historical understand
ing o f the in fo rm al processes for know ledge accum ulation and diffu sion that have
been neglected in h istorical research on the First Industrial R evolution in p articu
lar. In addition , as we p o in ted out in o u r in tro d u cto ry discussion, a key h isto rio
graphical m ystery rem ains— w hy w as N orthw est Europe the locus o f the first
transition to sustained, in n ovation -led grow th, rather than A sia o r som e other
region o f the global econom y? M uch o f the discussion o f this age-old question relies
heavily on sw eeping generalizations, and m ore research on the (asserted) failure of
non-European econ om ies to m ake the transition to sustained econom ic grow th
during this early p erio d is needed.
374 K R I S T I N E B R U L A N D A N D D A V I D C. M O W E R Y
Second Industrial R evolution outside o f the electrical equipm ent and chem icals
industries has received far too little attention.
Institutional change in areas ranging from m anagerial hierarchies and enterprise
control to the training o f scientists and engineers is o f central im portance to
innovative perform an ce and to change in the structure o f innovation system s over
time. In m any cases, these institutional changes have occurred in response to
pressure from innovators and entrepreneurs, and are best characterized as “ coevol
ving” with change in indu stries and technologies (see Engerm an and S o k o lo ff 2003
for a useful discussion o f this po in t). The transform ation o f the G erm an and US
university system s, as w ell as the m uch m ore lim ited restructuring o f British
universities, is but one exam p le o f this coevolution; the developm ent o f intellectual
property rights system s is another. But the conditions under which such political
pressure arises, as w ell as the factors contributing to the success o f failure o f such
pressure, rem ain p o o rly u nd erstood .
Although it n o w is w id ely celebrated as a hallm ark o f tw enty-first-century
“knowledge-based econom ies,” science-based innovation is in fact a relatively recent
development. Indeed, it appears w ell after the institutionalization o f R8cD within
industry in the early tw entieth cen tu ry in the U nited States and Germ any. M oreover,
even “ h igh -techn ology” sectors such as biotech nology and sem iconductors co n
tinue to rely o n experim ental m ethods that at their heart are “ trial and error”
approaches (see Pisano 1997; H atch and M o w ery 199S).
Our approach has been lim ited both in tim e and geographical coverage. Perhaps
the two m ost im p o rtan t o m ission s fro m this account are the spread o f industrializa
tion beyond G erm an y w ith in nineteenth-century Europe, and a full account o f the
rise o f the A sian econom ies after 1945 (see C h. 19 b y Fagerberg and G odinh o in this
volume for a fuller accoun t o f the process o f econom ic “ catch-up” in A sia). In the
first o f these cases it is im p o rtan t to rem em ber that the sm aller European econom ies,
now am ong the w ealthiest in the w orld, have benefited from the im port and
adaptation o f tech n ology d u rin g the nineteenth-century, and that Ireland since
1970 has enjoyed rap id grow th fro m sim ilar sources. The post-1945 grow th o f
Japan and Korea was o rigin ally based on altogether different scales and types o f
innovative in du strialization — adaptation o f foreign technologies, largely in such
mature industries as au tom obiles, steel, and sh ipbuilding (see also C h. 19 in this
volume). But both o f these episodes highlight the im portance o f broad Institutional
change, rather than the “ strategic im p o rtan ce” o f any single indu stry or technology,
in m uch the sam e w ay as do the three “ Industrial R evolutions described in this
chapter.
K R I S T I N E B R U L A N D A N D D A V I D C. M O W E R Y
N o tes _ _ _ ________________________
i. Chapter 2 by Lazo nick in this volume also discusses the role o f the institutional environ
ment in contributing to firm-level innovative capabilities,
2 See Peer Vries (2002) for an overview and criticjue o f the literature in this field,
3. “ [Scientific explanations proved to be reliable guides to the commercial development of
new processes and products. Unlike the unrestrained inventions o f myth and fable, they
could not be ignored by industrial firms except at the risk o f being displaced by rival firms.
But to understand and apply scientific explanation required years o f training in the
theology o f an invisible pantheon o f scientific entities. That requirement professionalized
industrial science and diminished the role o f artisan invention” (Rosenberg and Birdzell
1986: 253).
4. This discussion draws on Beer (1959), Murmann (1998), and Murmann and Landau
(1998).
5. See Murmann (1998, 2003) for a more detailed discussion.
R e fer en c es
A br a m o v itz , M. (1994), “ The Origins o f the Postwar Catch-Up and Convergence Boom,” in
J. Fagerberg, B. Verspagen, and N. von Tunzelmann (eds.), The Dynamics of Technology,
Trade and Growth, Aldershot: Edward Elgar, 21-52.
B e e r , J. J. (1959), The Emergence of the German Dye Industry, Urbana, 111.: University of
Illinois Press,
B e r g , M. (1998), “ Product Innovation in Core Consumer Industries in Eighteenth Century
Britain,” in M. Berg and K. Bruland (eds.), Technological Revolutions in Europe: Historical
Perspectives, Cheltenham: Edward Elgar, 138-58.
------ and E g er , E. (2003), Luxury in the Eighteenth Century, N ew York: Palgrave.
B e y e r c h e n , A. (1988), “ On the Stimulation o f Excellence in Wilhelmian Science” in J. R.
Dukes and J. Remak, Another Germany: A Reconsideration of the Imperial Era, Boulder,
Colo,: Westview Press, 139-68.
B ra u d el , F. (1984), The Perspective of the World, Berkeley: University o f California Press.
B ru la n d , K. (1989), The Transformation o f Work in European Industrialization,” in P.
Mathias and J. A. Davis (eds.), The Nature of Industrialization: The First Industrial
RevolutionsуOxford: Basil Blackwell, 154—70.
(2004), Industrialisation and Technological C hange,” in R. Flou d and P. Johnson
(eds), The Cambridge Economic History of Modern Britain, vol. 1, Industrialisation, Cam
bridge: C am bridge U niversity Press, 117-46.
C h a n d ler , A. D., Jr. (1977), The Visible Hand, C am b rid ge, M ass.: H arvard University Press.
C h a pm a n , S. D. (1972), The Cotton Industry in the Industrial Revolution, Lon don : Macmillan.
D e a n e , R (1965), The First Industrial Revolution, C am b rid ge: C am b rid ge University Press.
Encyclopaedia Britanmca (1995), 15th ed n „ v o l 8 , C h icago: E n cyclo p aed ia Britannica.
M ow ery, D. C> (1983)5 “ Industrial Research, Firm Size, Growth, and Survival, 1921—46,
Journal of Economic History 43: 953-80.
*____ (1995), “ The Boundaries o f the U.S. Firm in R&D,” in N. R. Lamoreaux and D. M. G.
Raff (eds.), Coordination and Information: Historical Perspectives on the Organization of
EnterpriseyChicago: University o f Chicago Press for NBER.
------ (1996), The International Computer Software Industry: A Comparative Study of Industry
Evolution and Structure, Oxford: Oxford University Press.
------ (1999), “ The Computer Software Industry” in D. C. M owery and R. R. Nelson (eds.),
Sources of Industrial Leadershipy New York: Cambridge University Press, 133—68.
------ and R o se n b e r g , N. (1998), Paths of Innovation: Technological Change in 20th Century
Americay New York: Cambridge University Press.
------ and S im co e , T. (2002), “ The History and Evolution o f the Internet,” in B. Steil,
R. Nelson, and D. Victor (eds.), Technological Innovation and Economic Performancey
Princeton: Princeton University Press, 229-64.
M u rm a n n , J. R (1998), “ Knowledge and Competitive Advantage in the Synthetic Dye
Industry, 1850-1914,” unpublished Ph.D. dissertation, Columbia University,
------ (2003), “ The Coevolution o f Industries and Comparative Advantage: Theory and
Evidence,” unpublished MS.
------ and L a n d a u , R. (1998), “ On the Making o f Competitive Advantage: The Development
o f the Chemical Industries o f Britain and Germany Since 1850,” in A. Arora, R. Landau, and
N. Rosenberg (eds.), Chemicals and Long-Term Economic Growthy New York: Wiley, 27-70.
N elso n , R. R., and W r ig h t , G. (1994), “ The Erosion o f U.S. Technological Leadership as a
Factor in Postwar Economic Convergence,” in W. J. Baumol, R. R, Nelson, and E, N. Wolff
(eds,), Convergence of Productive New York: Oxford University Press.
O rsen ig o , L, (1988), The Emergence of Biotechnology, London: Pinter.
P h a r m a c eu tic a l M a n u fa c tu r er s A sso ciatio n (2003), Industry Profile 2003 http://
www.phrma.org/ publications/publications/profileo2
P isa n o , G. (1997), The Development Factoryy Boston: Harvard Business School Press,
------ S h a n , W , and T e e c e , D. J. (1988), “ Joint Ventures and Collaboration in the Biotech
nology Industry,” in D. C. Mowery (ed.), International Collaborative Ventures in L/.S.
Manufacturingy Cambridge, Mass.: Ballinger.
P o llard , $. (1965), The Genesis of Modern Management A Study of the Industrial Revolution
in Great Britiany London: Arnold,
^P o m eran z , K. R. (2000), The Great Divergence: China, Europe and the Making of the Modern
World Economy, Princeton: Princeton University Press.
R e ic h , L. S. (1985), The Making of American Industrial Research, New York: Cambridge
University Press,
R e id , T, R. (1984), The Chip, New York: Simon and Schuster.
R o sen ber g , N. (1985)» The Commercial Exploitation o f Science by American Industry,” in
K. B. Clark, R. H. Hayes, and C. Lorenz (eds.), The Uneasy Alliance, Boston: Harvard
Business School Press, 19-51.
and B ir d z e l l , L. E. (1986), How the West Grew Rich, New York: Basic Books.
S a m u el , R. (1977), The Work-shop ot the World: Steam Power and Hand Technology in
mid-Victorian Britain,” History Workshop 3: 6-72.
Sch m o o kler , J. (1957), Inventors Past and Present,” Review of Economics and Statistics 57:
INNOVATION THROUGH TIME 379
Clarendon Press.
*V r ie $, P, (2002), “Are Coal and Colonies Really Crucial? Kenneth Pomeranz and the Great
Divergence,” Journal of World History 12(2): 407-45.
W a d e , R. (1999), Governing the Market: Economic Theory and the Role of Government in East
Asian IndustriaizationyPrinceton: Princeton University Press.
W a l l e r s t e in , I. (1974), Capitalist Agriculture and the Origins of the Modern World Economy;
New York: Academic Press.
W h it e h e a d , A. N. (1925), Science and the Modern World, New York: Macmillan.
CHAPTER 14
SECTORAL SYSTEM S
HOW A N D W H Y
INNOVATION DIFFERS
ACROSS SECTORS
FRANCO MALERBA
14.1 I n t r o d u c t i o n
differences. The role o f in n o vation in the dynam ics and transform ation o f these
sectors is highly diverse.
How is it possible to analyze consistently these differences and their effects on
sectoral growth and perform ance? The industrial econom ics approach pays a lot o f
attention to differences across sectors in R8cD intensity, m arket structure, the range
of viable R & D strategies and R & D alliances, the intensity o f the patent race, the
effectiveness o f patent protection , the role o f com petition policy and the extent o f
R&D support. But, w hile these are very im p o rtan t factors, they are not the only ones
nor are they the m ost relevant for a full understanding o f the differences in in n o v
ation across sectors.
A rich and heterogeneous trad ition o f sectoral studies has clearly shown both that
sectors differ in term s o f the know ledge base, the actors involved in innovation, the
links and relationships am o n g actors, and the relevant institutions, and that these
dimensions clearly m atter fo r u nderstandin g and explaining innovation and its
differences across sectors. H ow ever, these case studies are quite different in terms
of methodology, variables, and countries exam ined.
This chapter w ill b riefly discuss the previous literature on differences across
sectors in innovation (Section 14.2) and then propose the concept o f sectoral systems
of innovation (14.3). In the next sections, the basic building blocks o f sectoral
systems will be discussed: know ledge, technological dom ains, and sectoral b o u n d
aries (14.4); actors, relationships and netw orks (14.5); and institutions (14.6). Then
the dynam ics and tran sfo rm atio n o f sectoral system s (14.7) is exam ined. Finally,
some policy im plication s (14.8) and the challenges ahead (14.9) are discussed.
The chapter w ill discuss a large n u m ber o f sectors that are highly innovative and
technologically advanced and have strong links w ith science, w hich nevertheless
organize innovation v e ry differently: com puters, sem iconductors, telecom m unica
tion equipm ent and services, softw are, chem icals, pharm aceuticals and biotechnol
ogy, and m achine tools. M ost o f the sectoral exam ples in this paper are draw n from
M owery and N elson (1999) and M alerb a (2004).
14.2 P r e v i o u s L i t e r a t u r e on S e c t o r a l
D i f f e r e n c e s in In n o v a t io n
The literature has advanced some distinctions among sectors in innovation and
diffusion based on different dimensions. The simplest one, widely used in inter
national studies by the OECD, EU, and international organizations, refers to sectors
ЗВі FRANCO MALERBA _____________ ___________________________________ —
that are high R&D-intensive (such as electronics or drugs) and low RfkD-intemive
(such as textiles or shoes).
A n oth er d istin ction , co m in g fro m the Sch u m p eterian legacy, focuses on differ
ences in m arket structure and in d u strial d yn am ics a m o n g sectors. Sch u m p eter Mark
I sectors are characterized b y "creative destru ctio n , w ith tech n o lo g ical ease o f entry
and a m ajo r role played b y entrepreneurs and n ew firm s in in n o vative activities.
Sch u m p eter M a rk II sectors are characterized b y 'c re a tiv e a c c u m u la tio n ” (in Keith
Pavitt s w ords) w ith the prevalence o f large establish ed firm s an d the presence o f
relevant barriers to en try for new in n ovators. T h is regim e Is characterized by the
d om in an ce o f a stable core o f a few large firm s, w ith lim ited entry. T h e distinction
refers to the early Schu m p eter o f Theory o f Economic Development (19 11, "Sch u m p
eter M a rk I” ) and to the later one o f Capitalism > Socialism and Democracy (1942,
"S ch u m p e ter M a rk II” ). M ach in ery o r b io tech n o lo g y are exam p les o f Schumpeter
M a rk I sectors, w hile the sem icon d u ctor in d u stry o f the 1990s (th in k o f m icro
processors and d yn am ic m em ories) o r m a in fra m e c o m p u te rs in the period
19505-19905 are exam ples o f Schu m peter M a rk II sectors.
O ther differences across sectors have been related to tech n o lo g ical regimes, a
n o tio n introduced b y N elson and W inter (1982), referrin g to the learning and
know ledge en viron m en t in w h ich firm s operate. A sp ecific tech n o logical regime
defines the nature o f the prob lem firm s have to so lve in th eir in n o vative activities,
affects the m odel fo rm o f tech nological learn in g, shapes the incen tives and con
straints to particu lar b eh avio r and o rgan ization , an d in flu en ces the basic processes
o f variety generation and selection (and therefore the d y n a m ics and evolution o f
firm s). M ore generally, M alerb a and O rsenigo (1996 an d 1997) have p ro p o sed that a
technological regim e is com posed b y o p p o rtu n ity and a p p ro p ria b ility conditions,
degrees o f cum ulativeness o f tech nological kn ow ledge, an d ch aracteristics o f the
relevant know ledge base. M o re specifically, tech n o logical o p p o rtu n itie s reflect the
likelih ood o f in n o vatin g for any given am o u n t o f m o n e y invested in search. High
o p p o rtu n ities p ro vid e p ow erfu l incentives to the u n d erta k in g o f in n ovative activ
ities and denote an econom ic en viron m en t that is n o t fu n c tio n a lly constrained by
scarcity. In this case, potential innovators m ay com e up w ith freq u en t and im portant
tech nological innovations. A p p ro p ria b ility o f in n o vatio n s su m m arizes the possibil
ities o f protectin g in n ovation s fro m im itation and o f reap in g p ro fits fro m innovative
activities. H igh ap p ro p riab ility m eans the existence o f w ays o f su ccessfu lly protect
ing in n o vation from im itation. L o w ap p ro p riab ility co n d itio n s d en o te an economic
en viron m en t characterized b y the w idespread existence o f extern alities (Levin et al.
1987)* C um ulativeness con dition s capture the p ro p erties that to d a y ’s innovations
and innovative activities form the starting p o in t fo r to m o rro w in n o vation s, More
broadly, one m ay say that high cum ulativeness m eans that to d a y ’s in n o vative firms
are m o re likely to innovate in the future in specific tech n o logies an d along specific
trajectories than n o n -in n ovative firm s. C u m u lativen ess m a y be d u e to knowledge/
cognitive factors, organizational factors, or m arket factors o f the "su ccess breeds
S E C T O R A L S Y S T E M S OF I N N O V A T I O N 383
success” type. The properties o f the know ledge base relate to the nature o f knowledge
underpinning firm s innovative activities. Technological know ledge involves various
degrees o f specificity, tacitness, com plem entarity, and independence and m ay
greatly differ across sectors and technologies (W inter 1987). Differences in techno
logical regimes affect the organization o f innovative activities at the sectoral level and
may lead to a fundam ental d istin ction between Schum peter M ark I and Schum peter
Mark II m odels. H igh technological opportu nities, low appropriability, and low
cumulativeness (at the firm level) conditions lead to a Schum peter M ark I pattern.
By contrast, high a p p ro p riab ility and high cum ulativeness (at the firm level) co n d i
tions lead to a Schum peter M ark II pattern: think again o f the sem iconductor
industry o f the 1990s (i.e. m icroprocessors and dynam ic m em ories) and m ainfram e
computers in the p erio d 19505-19905.
Technological regim es and Schum peterian patterns o f innovation change over
time (Klepper 1996). A cco rd in g to an in d u stry life-cycle view, a Schum peter M ark I
pattern o f innovative activities m ay turn into a Schum peter M ark II. E arly in the
history o f an in d u stry— w hen know ledge is changing very rapidly, uncertainty is
very high, and barriers to en try very low — new firm s are the m ajor innovators and
are the key elem ents in in du strial dynam ics. W hen the indu stry develops and
eventually m atures and technological change follow s w ell-defined trajectories,
economies o f scale, learning curves, barriers to entry, and financial resources becom e
important in the com petitive process. Thus, large firm s w ith m onopolistic pow er
come to the forefron t o f the in n o vation process (U tterback 1994; G o rt and K lepper
1982; Klepper 1996). In the presence o f m ajor know ledge, technological, and m arket
discontinuities, a Schu m peter M ark II pattern o f innovative activities m ay be
replaced by a Schum peter M a rk L In this case, a rather stable organization charac
terized by incum bents w ith m o n o p o listic pow er is displaced b y a m ore turbulent
one with new firm s using the new technology or focusing on the new dem and
(Henderson and C la rk 1990; C hristensen and R osen bloom 1995)* A lthough rather
archetypical, these analyses p o in t to the direction o f placing a lot o f attention to
differences across sectors in som e key factors related to knowledge and learning
regimes. A s the exam ples discussed above suggest, change over tim e also reflects
institutional change and the co evolu tion o f industries and institutions.
Other distinctions refer to sectors that are net suppliers o f technology and sectors
that are users o f technology. O n the bases o f the R & D done by 400 A m erican firm s
and o f intersectoral flow s in the A m erican econom y, Scherer (1982) identifies
sectors that are net sources o f R & D for other sectors (such as com puters and
instrum ents), and sectors that are net users o f technology (such as textiles
and m etallurgy). A sim ilar analysis is done b y R obson et al. (1988) who, on the
basis o f 4,378 in n o vation s in the U K between 1945 and 1983, identify (a) “ core
sectors” (such as electronics, machinery, instruments, and chemicals) which gener
ate m ost o f in n o vation s in the econ o m y and are net sources of technology, (b)
secondary sectors (such as auto and metallurgy) which play a secondary role in
384 FR A N CO MALERBA
term s o f sources o f in n ovation for the econ om y, an d (c) u ser sectors su ch as services
w h ich m a in ly absorb technology,
A key difference am o n g sectors refers to the so u rces o f in n o vatio n and the
ap p ro p ria b ility m echanism s. Pavitt (1984) p ro p o ses fo u r typ es o f sectoral pattern
fo r in n o vative activities. In su p p lier-d o m in ated (e.g. textile, services) sectors, new
technologies are em bod ied in n ew co m p o n en ts an d eq u ip m en t, an d the diffusion o f
n ew technologies and learning takes place th ro u g h le a rn in g -b y -d o in g and by-using.
In scale-intensive sectors (e.g. autos, steel), process in n o v a tio n is relevant and the
sources o f in n o vation are both internal (R & D an d le a rn in g -b y -d o in g ) and external
(eq u ipm en t p ro d u cers), w hile a p p ro p ria b ility is o b tain ed th ro u g h secrecy and
patents. In specialized suppliers (e.g. eq u ip m en t p ro d u ce rs), in n o va tio n is focused
on perfo rm an ce im provem en t, reliability, an d cu sto m izatio n , w ith the sources o f
in n o vatio n being b oth internal (tacit know ledge an d exp erien ce o f skilled techni
cians) an d external (u se r-p ro d u c e r in teractio n ); a p p ro p ria b ility com es m ain ly from
the localized and interactive nature o f know ledge. Finally, scien ce-b ased sectors (e.g.
ph arm aceuticals, electronics) are characterized b y a h igh rate o f p ro d u ct and process
in n o vation s, by internal R & D , and b y scien tific research d o n e at universities and
pu b lic research laboratories; science is a sou rce o f in n o va tio n , an d appropriability
m eans are o f variou s types, ran gin g fro m patents, to lead -tim es an d learning curves,
and to secrecy. The Pavitt taxo n o m y has been trem en d o u sly successful in empirical
research and has guided the identification o f firm s an d c o u n try advantages. Refine
m ents and enrichm ents o f the ta x o n o m y h ave been p ro p o sed in the following
decades. A v e r y interesting and relevant w o rk in this d irectio n is the on e by Marsili
(2001).
D ifferences across sectors in a p p ro p ria b ility co n d itio n s h ave been exam ined by
Levin et al. (1987), PA C E (1996) and C o h en et al. (2002) u sin g su rve y questionnaires
fo r R & D m anagers in the United States, E u rop e, an d Ja p an , fo llo w in g the pioneering
Yale survey. H ere, m ajo r differences across sectors h ave been id en tified in terms o f
a p p ro p ria b ility m eans— patents, secrecy, lead -tim es, learn in g cu rves, and com ple
m en tary assets. A ll these surveys have fo u n d m a jo r d ifferen ces acro ss sectors in the
use o f patents.
14*3 S e c t o r a l S y s t e m s o f I n n o v a t i o n
The co n trib u tio n s exam ined above focus on a specific d ifference a m o n g sectors. In
this and the fo llo w in g sections, a m u ltid im en sio n al, integrated , an d d yn am ic view o f
in n o vation in sectors is proposed , related to the fram ew o rk o f sectoral systems o f
S E C T O R A L S Y S T E M S OF I N N O V A T I O N 385
innovation, which pro vid es a m eth o d o lo gy for the analysis and com parison o f
sectors.
A sector is a set o f activities that are unified by som e linked product groups for a
given or em erging dem and and w hich share som e com m on knowledge. Firm s in a
sector have som e co m m on alities and at the sam e tim e are heterogeneous. A sectoral
system fram ew ork focuses on three m ain dim ensions o f sectors:
(a) Knowledge and tech nological d om ain
( b) Actors and netw orks
(c) Institutions
(b) Actors and networks. A sector is com posed o f heterogeneous agents that are
organizations or in d ivid u als (e.g. consum ers, entrepreneurs, scientists). O rganiza
tions m ay be firm s (e.g. users, producers, and input suppliers) or n on-firm s (e.g.
universities, financial institutions, governm ent agencies, trade-unions, or technical
associations), and include subunits o f larger organizations (e.g. R M ) or production
departments) and grou p s o f organizations (e.g. indu stry associations). Agents are
characterized b y specific learning processes, com petencies, beliefs, objectives, o r
ganizational structures, and beh aviors, w hich interact through processes o f co m
munication, exchange, co op eration , com petition, and com m and.
Thus, in a sectoral system fram ew ork, innovation is considered to be a process
that involves system atic interactions am on g a w ide variety o f actors for the gener
ation and exchange o f know ledge relevant to innovation and its com m ercialization.
Interactions include m arket and non -m arket relations that are broader than the
market for technological licensing and know ledge, interfirm alliances, and form al
networks o f firm s, and often their ou tcom e is not adequately captured b y our
existing system s o f m easu rin g econ om ic output.
(c) Institutions. A gen ts’ cognition , actions, and interactions are shaped by insti
tutions, which include n o rm s, routines, co m m on habits, established practices, rules,
laws, standards, and so on. In stitu tions m ay range from ones that bind or im pose
enforcements on agents to ones that are created b y the interaction am ong agents
(such as contracts); fro m m ore b in d in g to less binding; from form al to inform al
(such as patent laws o r specific regulations vs. traditions and conventions). A lot o f
institutions are national (such as the patent system ), w hile others are specific to
sectors (such as sectoral lab o r m arkets o r sector specific financial institutions).
Over tim e, a sectoral system undergoes processes o f change and transform ation
through the co evolu tion o f its variou s elem ents.
386 FR A N C O MALERBA
well as for the institutional context. Thus the learning, behavior, and capabilities of
agents are constrained and “ bounded” by the technology, knowledge base, and
institutional context. Heterogeneous firms facing similar technologies, searching
around similar knowledge bases, undertaking similar production activities, and
“embedded” in the same institutional setting, share some common behavioral and
organizational traits and develop a similar range of learning patterns, behavior, and
organizational forms.
One last remark regards the aggregation issue regarding products, agents or
functions. For example, sectoral systems may be examined broadly or narrowly (in
terms of a small set of product groups) . 2 A broad definition allows us to capture all
the interdependencies and linkages in the transformation of sectors, while a narrow
definition identifies more clearly specific relationships. Of course, within broad
sectoral systems, different innovation systems related to different product groups
may exist. The choice of the level of aggregation depends on the goal of the analysis.
In the following pages we will concentrate on each block of a sectoral system of
innovation and production:
• Knowledge, technological domain, and boundaries
• Agents, interaction and networks
• Institutions
What d o we know about the main dimensions of knowledge? First, knowledge may
have different degrees of accessibility (Malerba and Orsenigo 2000), i.e. opportunities
of gaining knowledge external to firms, which in turn may be internal or external to
the sector. In both cases, greater accessibility of knowledge m a y decrease industrial
concentration. Greater accessibility internal to the sector implies lo w er appropriabil
ity: competitors may gain knowledge about new products and processes and, if
competent, imitate those new products and processes. Accessibility of knowledge
that is external to the sector may be related to the levels and sources of scientific and
technological opportunities. Here, the external environment may affect firms through
human capital with a certain level and type of knowledge or through scientific and
technological know ledge developed in firm s o r n o n -firm organizations, such as
universities o r research laboratories (M alerba and O rsenigo 2000).
The sources of technological opportunities differ markedly among sectors. As
Freem an (1982) and R osen berg (1982), am o n g oth ers, h ave sh o w n , in som e sectors
opportunity conditions are related to m ajo r scientific breakthroughs in universities;
in others, opportunities to innovate may often come fro m advancements in R&D,
equipment, and instrumentation; while in still other sectors, external sources of
knowledge in terms of suppliers or users may play a crucial role. Not all external
knowledge may be easily used and transformed into new artifacts. If external
know ledge is easily accessible, tran sform able into n ew artifacts an d exposed to a
lot o f actors (such as custom ers o r su p pliers), then in n o vative e n try m a y take place
(W in ter 1984). I f advanced integration capabilities are n ecessary (C ohen and
Levin th al 1989), the in d u stry m ay be concen trated and fo rm e d b y large, established
firm s.
Second, know ledge m ay be m ore o r less cu m u lative, i.e. the degree b y which the
generation o f new know ledge bu ilds u p o n cu rren t k n o w led ge. O n e can identify
three different sources o f cum ulativeness.
localized knowledge spillovers. Finally, cum ulativeness at the technological and firm
levels creates first m over advantages and generates high concentration. Firm s that have
a head start develop a new know ledge based on the current one and introduce
continuous innovations o f the increm ental type.
Accessibility, o p p ortu n ity, and cum ulativeness are key dim ensions o f knowledge
related to the notion o f tech nological and learning regim es (N elson and W inter 1982;
Malerba and O rsenigo 1997), w hich, as seen above, m ay differ across sectors. Other
dimensions o f know ledge could be related to its tacitness, codificability, com plexity,
systemic features, scientific base, and so on (W inter 1987; C ow an, D avid, and Foray
2000).
The boundaries o f sectoral system s are affected by the knowledge base and
technologies. H ow ever, the type and dynam ics o f dem and represent a m ajor factor
in the processes o f tran sfo rm ation o f sectoral system s. The sam e holds for links and
complementarities am o n g artifacts and activities. These links and com plem entar
ities are, first o f all, o f the static type, as are in p u t-o u tp u t links. Then there are
dynamic com plem entarities, w h ich take into account interdependencies and feed
backs, both at the dem and and at the p ro d u ction levels. D ynam ic com plem entarities
among artifacts and activities are m ajo r sources o f transform ation and growth o f
sectoral systems, and m ay set in m o tio n virtu ou s cycles o f innovation and change.
This could be related to the concept o f filiere and the notion o f developm ent blocks
(Dahmen 1989). Links and com plem entarities change over tim e and greatly affect a
wide variety o f variables o f a sectoral system : firm s’ strategies, organization, and
performance; the rate and d irection o f technological change; the type o f com peti
tion; and the netw orks am o n g agents. T h u s the boundaries o f sectoral systems m ay
change more or less rap id ly over tim e, as a consequence o f dynam ic processes related
to the transform ation o f know ledge, the evolution and convergence in dem and, and
changes in com petition and learning b y firm s.
In general, the features and sources o f know ledge affect the rate and direction o f
technological change, the o rgan ization o f innovative and production activities, and
the factors at the base o f firm s’ successful perform ance.
Great differences am o n g sectors in the dim ensions discussed above exist. Let us
compare, fo r exam ple, ph arm aceu ticals and m achine tools. In the pharm aceutical
industry, the know ledge base and the learning processes have greatly affected
innovation and the organ ization o f innovative activities. In the early stages (18 50 -
i945)> the in d u stry w as close to chem icals, w ith little form al research until the 193°$
and a m ajor use o f licenses. T h e fo llo w in g perio d (19 45-early 1980s) was character
ized by the in tro d u ction o f ran d o m screening o f natural and chem ically derived
com pounds. T h is led to an exp losio n o f RScD and, although few blockbusters were
discovered in each p erio d , nevertheless, each period enjoyed high grow th. The
advent o f m olecular b io lo g y since the 1980s led to a new learning regim e based on
molecular genetics and rD N A technology, w ith tw o search regim es: one regarding
specialized technologies, the other generic technologies. N ow adays, no individual
390 FR A N CO MALERBA
firm can gain control o n m ore than a subset o f the search space* In n o v a tio n increas
in gly depends on stron g scientific capabilities and o n the a b ility to interact with
science and scientific institution s in o rd er to exp lo re the search space (McKelvey,
O rsenigo, and P am m o lli 2004; H en derson , O rsen igo, an d P isan o 1 999 )*
In m achine tools, innovation has been m a in ly in crem en tal an d n o w is increas
in gly systemic* K n ow led ge ab ou t ap p lication s is v e ry im p o rta n t, an d therefore user-
p ro d u cer relationships as well as partn ersh ip s w ith cu sto m ers are com m on . The
know ledge base has been em bod ied in skilled person n el o n the sh op flo o r level (with
applied technical q ualification) and in design engineers (n ot n ecessarily with a
u n iversity degree but w ith lo n g -term em p lo ym en t in the co m p an y ). Internal
train in g (p articu larly apprenticeships) is qu ite relevant. In sm all firm s, R8cD is not
done extensively and R8cD co o p eratio n is n ot co m m o n . R ecently, the knowledge
base has shifted fro m pu rely m echan ical to m ech an ic, m icro e lectro n ic and infor
m atio n intensive, w ith an increasing co d ificatio n and an in creasin g use o f formal
R8cD. Prod ucts have increasingly being m o d u larized an d stan d ard ized . A key role is
also played b y in fo rm atio n flow s ab ou t co m p o n en ts c o m in g fro m producers o f
different technologies, such as lasers, m aterials, m easu rem en t, an d con trol devices.
N ow ad ays, m an y large m ach in e tool com pan ies o p erate alre a d y o n an international
basis m akin g use o f specific know ledge sources at th eir d ifferen t firm sites (Wengel
and Shap ira 2004; M azzoleni 1999).
14.5 A c t o r s , R e l a t i o n s h i p s ,
a n d N etworks
Sectoral system s are co m p o sed o f heterogen eous actors. In general, a rich, m ulti
d isciplinary, and m u ltisou rce know ledge base an d rap id tech n o lo g ical change im
plies a great heterogen eity o f actors in m ost sectors.
Firm s are the key actors in the generation, ad o p tio n , and use o f n ew technologies,
are characterized b y specific beliefs, expectations, goals, co m peten ces, an d organiza
tion, and are co n tin u o u sly engaged in processes o f learn in g an d know ledge accu
m ulation (N elson and W inter 1982; M alerb a 1992, Teece an d P isan o 1994, Dosi,
M aren go, and Fagiolo 1998, M etcalfe 1998). T h e extent o f firm heterogen eity is the
result o f the o p p o sin g forces o f variety creation, rep lication , an d selection (Nelson
1995) M etcalfe 1998). Selection increases h om ogen eity, w h ile e n try an d technological
and o rgan ization al in n ovation s are fu n d am en tal sources o f heterogeneity. Firm
heterogeneity is also affected by the characteristics o f the k n o w led ge base, specific
experience and learning processes, and the w o rk in g o f d y n a m ic com plem entarities.
S E C T O R A L S Y S T E M S OF I N N O V A T I O N
391
Actors also include users and suppliers w ho have different types o f relationships
with the innovating, p ro d u cin g, or selling firm s, Users and suppliers are character
ized by specific attributes, know ledge, and com petencies, with m ore or less close
relationships w ith p rodu cers (V onH ippel 1988, Lundvall 1993). As previously m en
tioned, in a dynam ic and innovative setting, suppliers and users greatly affect and
continuously redefine the bou n daries o f a sectoral system.
Other types o f agents in a sectoral system are n on -firm organizations such as
universities, financial organizations, governm ent agencies, local authorities, and so
on. In various ways, they su p p o rt innovation, technological diffusion, and pro d u c
tion by firm s, but again their role greatly differs am ong sectoral systems. In several
high technology sectors, universities play a key role in basic research and hum an
capital form ation, and in som e sectors (such as biotechnology and software) they are
also a source o f start-u ps and even innovation. In sectoral system s such as software
or biotech nology-pharm aceu ticals, new actors such as venture capital com panies
have emerged over tim e. These financial organizations have played a different role
according to the stage o f the in d u stry life-cycle. W hen indu stry m atures or large
firms are relevant, capital constraints becom e lighter and m uch investm ent is self-
financed. B y contrast, fo r start-ups in em erging or new high-tech sectors, capital
constraints are v ery high and specific financial interm ediaries such as venture capital
firms are im portan t (R ivau d -D an set 2001; D ubocage 2002).
Often the m ost app ro priate units o f analysis in specific sectoral systems are not
necessarily firm s b u t in d ivid u als (such as the scientist w ho opens up a new biotech
nology firm ), firm s’ subunits (such as the R & D or the production departm ent), and
groups o f firm s (such as in d u stry consortia).
The focus on users, govern m en t agencies, and consum ers puts a different em
phasis on the role o f dem and. In a sectoral system , dem and is not seen as an
aggregate set o f sim ilar buyers o r atom istic undifferentiated custom ers, but as
com posed o f heterogeneous agents w ho interact in various ways w ith producers.
Demand then becom es com posed b y in divid u al consum ers, firm s, and public
agencies, w hich are in tu rn characterized b y know ledge, learning processes, and
competences, and w hich are affected b y social factors and institutions. The em er
gence and tran sfo rm ation o f dem and becom e then a very im portant part in the
dynamics and evolu tion o f sectoral system s. In addition, dem and has often proven
to be a m ajor factor in the redefin ition o f the boundaries o f a sectoral system, a
stimulus for in n o vation , and a key factor shaping the organization o f innovative and
production activities.
W ithin sectoral system s, heterogeneous agents are connected in various ways
through m arket an d n o n -m arket relationships. It is possible to identify different
types o f relations, linked to different analytical cuts. First, traditional analyses o f
industrial organizations have exam ined agents as involved in processes o f exchange,
com petition, and co m m an d (such as vertical integration). Second, in m ore recent
analyses, processes o f fo rm al co op eration or in form al interaction am ong firm s or
392 FRANCO MALERBA
a m o n g firm s an d n o n -firm organ ization s h ave been exa m in e d in d ep th (as one may
see fro m the literature o n tacit o r exp licit co llu sio n , o r h y b rid go vern an ce form s, or
fo rm a l R & D co o p eratio n ). T h is literature has an alyzed firm s w ith certain market
pow er, su p pliers o r users facing o p p o rtu n istic b e h a v io r o r asset specificities in
tran saction , an d firm s w ith sim ilar know ledge h av in g a p p ro p ria b ility and indivis
ib ility problem s in R & D . Finally, the e vo lu tio n a ry a p p ro a c h an d the innovation
system s literature have also paid a lot o f atten tion to the w id e ran ge o f form al and
in fo rm al co o p eratio n an d interaction am o n g firm s. H ow ever, acco rd in g to this
perspective, in uncertain and ch an gin g en viron m en ts n etw o rk s em erge n o t because
agents are sim ilar, b u t because they are different. T h u s, n etw o rk s integrate com ple
m entarities in know ledge, capabilities, and sp ecializatio n (see Lu n dvall 1993;
E dq u ist 1997; N elson 1995; Teubal et al. 19 9 1). R elatio n sh ip s b etw een firm s and
n o n -firm o rgan ization s (such as universities and p u b lic research centers) have
been a source o f in n o vatio n and change in several sectoral system s: pharm aceuticals
and biotech nology, in fo rm atio n technology, an d tele co m m u n ica tio n s have been
relevant (N elson an d R osenberg 1993).
O ne final o b servation needs to be m ade: the k ey role p layed b y netw orks in a
sectoral system leads to a m ean in g o f the term “ sectoral stru ctu re ” d ifferent from the
o n e used in in d u strial econom ics. In in d u strial eco n o m ics, stru ctu re is related
m ain ly to the concept o f m arket structure and o f vertical in tegratio n an d diversifica
tion. In a sectoral system perspective, o n the contrary, stru ctu re refers to links am ong
artifacts and to relationships am o n g agents: it is th erefo re far b ro a d e r than the one
based on e xch a n g e -co m p e titio n -co m m a n d . T h u s w e can say th at a sectoral system
is com posed o f w ebs o f relationships a m o n g h etero gen eo u s agents w ith different
beliefs, goals, com petencies, and behavior, and that these relatio n sh ip s affect agents'
actions. T h ey are rather stable over tim e.
In su m m ary, the types and structures o f relation sh ips an d n etw o rk s d iffer greatly
fro m sectoral system to sectoral system , as a co n sequ en ce o f the features o f the
know ledge base, the relevant learning processes, the b asic tech n o logies, the charac
teristics o f dem and , the key links, an d the d yn am ic co m p lem en tarities. A gain, let's
p ro vid e som e exam ples.
A gain , the co m p ariso n o f fo u r quite d ifferent sectoral system s, such as chemicals,
com pu ters, sem icond u ctors, and softw are, illustrates this p o in t. In chemicals,
the structure o f the sectoral system has been centered a ro u n d large firm s, which
have been the m ajo r source o f in n o vatio n o ver a lo n g p erio d o f tim e. Large
R & D expenditures, econom ies o f scale and scope (C h an d ler 19 9 0 ), cum ulativeness
o f technical advance, and co m m ercialization cap ab ilities have given these firms
m ajo r in n ovative and com m ercial advantages (A rora, et al. 1998). W ith the diffusion
o f the synthetic d yestu ff m odel, firm s scaled up their R & D d ep artm en ts and the
role o f universities increased. T h e in tro d u ction o f p o ly m e r ch em istry (1920s)
affected the structure o f the in d u stry because kn ow led ge ab o u t the characteristics
of different m arket segm ents becam e im p o rtan t, so that firm s had to
S E C T O R A L S Y S T E M S OF I N N O V A T I O N
393
develop extensive linkages w ith dow nstream markets. The other m ajor change
related to the developm ent o f chem ical engineering and the concept o f unit o f
operation led to an increasing d ivisio n o f labor between chem ical com panies and
technology suppliers, w ith the rise o f the specialized engineering firm s (SEFs), which
developed vertical links w ith chem ical com panies. In this period, university research
continued to be im p o rtan t fo r the developm ent o f innovations, and links between
universities and in d u stry increased. In addition, advances in chem ical disciplines
and the separability o f know ledge increased the transferability o f chem ical technolo
gies. Thus, there has been a greater role o f licensing also by large firm s, which in turn
increased know ledge diffusion.
In computers, the different stages o f the evolution o f the industry (related to
different products) have been characterized by different actors and networks. H aving
been a typical Schum peter M ark II sector for m ost o f its history (until very recently),
mainframe com puters have always been dom inated by large firm s, with high cum ula
tiveness o f technical advance. In particular, during the 1960s and 1970s, m ainfram es
were produced and integrated b y vertically integrated firm s, and IB M was the typical
example. IB M was produ cing both com ponents and systems and was active in the
development, m anufacturing, m arketing, and distribution o f large systems and o f
the key com ponents. W hen m inicom puters were introduced, the com puters sector
experienced the entry and grow th o f firm s specialized in com ponents or in systems
(with the early years characterized b y a Schum peter M ark I pattern). The sam e holds
for the early years o f m icrocom puters. Later on, however, com petition became
characterized b y groups o f specialized firm s related to different platform s. Each
platform was characterized b y divid ed technical leadership o f several disintegrated
firms. Innovation becam e decentralized, and the control over the direction by a single
firm became very difficult. Recently, in com puter networks, m odularity and connect
edness increased the role o f netw orks o f firm s with local developm ent and local
feedbacks (Bresnahan and Greenstein 1999; Bresnahan and M alerba, 1999).
In sem iconductors, the in d u stry has been characterized b y a quite different set o f
actors, ranging fro m m erch ant sem iconductor m anufacturers to vertically inte
grated producers. T h e types o f actors have been quite different from period to
period and fro m co u n try to co u n try d u rin g the evolution o f the industry. N ew
entrants and specialized p ro d u cers w ere quite relevant in the U nited States, w ith
entrants particu larly high either early on in the h istory o f the indu stry or during
phases o f tech nological discontinu ities (and giving the indu stry a typical Sch u m p
eter M ark I fashion in these p eriods o f rapid and radical change). Large, vertically
integrated produ cers w ere m o re co m m on in Japan and Europe (M alerba 1985;
Langlois and Steinmueller 1999). T h u s, in these countries a Schum peter M ark
II m ode characterized the industry. In sem iconductors, other m ain actors have
played a m ajo r role. T h e m ilitary w as one o f the m ajo r factors responsible fo r the
growth o f the A m erican industry, com pared to Europe and Japan, because it
supported the en try of new firms and pro vid ed competent firms w ith a large and
394 FRANCO MALERBA
14.6 In s t i t u t i o n s
In all sectoral system s, institution s p lay a m a jo r role in affectin g the rate o f techno
logical change, the organ izatio n o f in n o vative activity, an d p erfo rm an ce. T h ey may
em erge either as a result o f deliberated plan ned d ecision b y firm s o r oth er organiza
tions, o r as the unpredicted consequence o f agents' in teraction .
Som e institution s are sectoral, i.e. specific to a sector, w h ile oth ers are national.
T h e relationship betw een national in stitu tion s and sectoral system s is quite im port
ant in m ost sectors. N ation al in stitu tion s have d ifferen t effects o n sectors. For
exam ple, the patent system , p ro p e rty rights, o r an titru st regu latio n s have different
effects as a consequence o f the different features o f the system s, as surveys and
em p irical analyses have sh ow n (see fo r exam p le L evin et al. 1987). H ow ever, the same
in stitu tion m a y take d ifferent features in d ifferent cou n tries, an d thu s m ay affect the
sam e sectoral system differently. F o r exam ple, the w e ll-k n o w n d iv ersity between the
first-to -in ven t and the first-to -file rules in the patent system s in the U nited States
and in Jap an had m ajo r consequences o n the b eh avio r o f firm s in these two
countries. O ften, the characteristics o f n ation al in stitu tio n s fa v o r specific sectors
that fit better the specificities o f the national in stitu tion s. T h u s, in certain cases,
som e sectoral system s becom e p red o m in an t in a c o u n try becau se the existin g insti
tu tions o f that co u n try p ro vid e an en viron m en t m o re su itab le fo r certain types o f
sectors and not for others. For exam ple, in France, sectors related to pu b lic dem and
have grow n co n sid erab ly (C hesnais 1993). In o th er cases, n atio n al in stitu tion s m ay
constrain d evelopm ent o r in n o vation in specific sectors, o r m ism atch es between
national and sectoral institutions and agents m ay take place. T h e exam p les o f the
different types o f interaction betw een n ation al in stitu tion s and sectoral evolution in
various advanced countries in D osi and M alerb a (1996) are cases in poin t.
S E C T O R A L S Y S T E M S OF I N N O V A T I O N
395
The relationship between national institutions and sectoral systems is not always
one-way, as it is in the case of the effects of national institutions on sectoral variables.
Sometimes, the direction is reversed, and goes from the sectoral to the national level.
In fact, it may occur that the institutions of a sector, which are extremely important
for a country in terms of employment, competitiveness, or strategic relevance, end
up emerging as national, thus becoming relevant for other sectors. But in the process
of becoming national, they may change some of their original distincti ve features.
Again, m ajor differences em erge across sectors, as in the case o f pharm aceuticals,
software, m achine tools, and telecom m unications, for exam ple. In pharm aceuticals,
national health system s and regulations have played a m ajor role in affecting the
direction o f technical change, in som e cases even blocking or retarding innovation.
In addition, patents have played a m ajo r role in the app ropriability o f the returns
from innovations. In softw are, standards and standard setting organizations are
important, and IP R play a m ajo r role in strengthening appropriability. However, the
emerging open source m ovem en t aim s to create a new segm ent o f the software
industry which is characterized by new d istribution m ethods and b y cooperative
production activities based on vo lu n ta ry association. This has reduced the possibil
ity o f m aintaining p ro p rietary control over data structure, thus inducing entry and
more com petition (Stein m u eller 2004). In m achine tools, internal and regional
labor markets and local institution s (e.g. local banks) have played a m ajor role in
influencing international advantages o f specific areas. Trust-based, close relation
ships at the regional level have over a long tim e ensured a sufficient financing o f the
innovation and o f the exp an sion plans o f fam ily businesses in G erm any and Italy
(Wengel and Shap ira 2004). Finally, in telecom m unications, the roles o f regulation,
liberalization/privatization, an d standards have been o f m ajor im portance in the
organization and perfo rm an ce o f the sector. A s discussed in D alum and V illum sen
(2001), liberalization and p rivatization have had m ajor effects on the behavior and
performance o f incum bents and have transform ed the structure o f the industry. A n
example o f the role o f institution s is given b y G S M in Europe.
14*7 T h e D y n a m i c s a n d T r a n s f o r m a t i o n
of S e c t o r a l S y s t e m s
As m entioned above, at the base o f the dynam ics and tran sform ation o f sectoral
systems lies the in terp lay am o n g evo lu tio n ary processes (such as variety creation,
replication, and selection) that d iffer fro m sector to sector (N elson 19955 M etcalfe
1998). Processes o f variety creation may refer to products, technologies, firms, and
institutions, as well as firm strategies and behavior and could take place through
entry, R8cD, innovation, and so on (Cohen and M alerba 2002). Sectoral systems
differ extensively in the processes o f variety creation and o f heterogeneity among
agents. The creation o f new agents— both new firm s and non-firm organizations—
is particularly important for the dynamics o f sectoral systems. As examined by
Audretsch (1996) and Geroski (1995), am ong others, the role o f new firms differs
drastically from sector to sector (in terms o f entry rates, com position, and origin),
and thus has quite different effects on the features o f sectoral systems and their
degree o f change. Sectoral differences in the level and type o f entry seem to be closely
related to differences in the knowledge base; level, diffusion and distribution of
competences; the presence o f non-firm organizations (such as universities and
venture capital); and the working o f sectoral institutions (such as regulations or
labor markets) (Audretsch x996; M alerba and Orsenigo 1999; M cKelvey 1997; Ger
oski 1995). Processes o f selection play the key role o f reducing heterogeneity among
firm s and may drive out inefficient or less progressive firm s. T h ey m ay refer to
products, activities, technologies, and so on. In addition to market selection, in
several sectoral systems non-m arket selection processes are at w ork, as in the cases of
the involvement o f the military, the health system, and so on. In general, selection
affects the growth and decline o f the various groups o f agents and the range o f viable
behaviors and organizations. Selection m ay greatly differ across sectoral systems in
terms o f intensity and frequency. Theoretical w ork (see M etcalfe 1998) and empirical
work on “ competence destroying” innovation, industrial dynam ics, firm s’ entry and
exit, and mergers and acquisitions have shed light on several aspects o f selection.
Changes in sectoral systems are the result o f coevolutionary processes o f their
various elements, involving knowledge, technology, actors, and institutions. Nelson
(1994) and Metcalfe (1998) have discussed these processes at the general level by
focusing on the interaction between technology, industrial structure, institutions,
and demand. These processes are sector-specific and often path-dependent. Here,
local learning, interactions among agents, and networks may generate increasing
returns and irreversibilities that may lock sectoral systems into inferior technolo
gies.3 In addition, the interaction between knowledge, technology firm s, and insti
tutions are also shaped by country-specific factors.
In general, one could say that changes in the knowledge base and in the relevant
learning processes of firms induce deep transform ations in the behavior and struc
ture of the agents and in their relationships between one another. Overall market
competition and market structure depend on the strategies and fortunes of individ
ual companies, which are linked to different national contexts o r to the international
scene. Firms have diverse reactions in order to try to increase their fit and to survive
in their particular environment. These environm ents keep changing, not least due to
innovations and choices made by all the constituent competitors: some of these
environm ents are national, others increasingly international.
S E C T O R A L S Y S T E M S OF I N N O V A T I O N
397
Over the past decades, com puters have had m ajor r e v o lu t io n a r y processes, quite
different from one another. In m ainfram es, coevolution has been characterized by
large systems requiring u se r-p ro d u c e r relationships, centralization o f user in fo r
mation systems, and extensive sales and service efforts b y large vendors. M arket
structure was h igh ly concentrated and suppliers were vertically integrated. A d o m
inant design (IBM /360) em erged in the grow th phase o f the segment and a market
leader (IBM ) dom inated the in d u stry early on, with a coordinating role over the
platform and the ability to steer the direction o f technical change. The U S govern
ment played a role in early su p p o rt for technological exploration and was a m ajor
buyer o f early com puters. In m in icom pu ters and m icrocom puters, coevolution has
been characterized b y tech nological change focused on dedicated applications in the
case o f m inicom puters o r on system s that increased ease o f use and a lower price/
performance ratio (in the case o f m icrocom pu ters). The relationships w ith custom
ers have required m uch less post-sales effort and service. M arket structure was
characterized by high entry early on, and then by increasing concentration in
platforms in both m in icom p u ters and m icrocom puters. In com puter networks,
connectivity and co m p atib ility led to m odular, open, and m ultiform client/server
platforms. Technical change follow s a variety o f directions w ith an upsurge in the
number o f potential technologies associated w ith the relevant platform s. Interde
pendencies and externalities have increased. D ivided technical leadership has
emerged, in that no single firm has been able to govern change and coordinate
platform standards.
This exam ple is quite different from coevolution in other sectoral systems. In
pharmaceuticals, the nature o f the process o f drug discovery (discussed in Section
14.4) had im portant consequences on the patterns o f com petition and on m arket
structure. U ntil the m o lecu lar b io lo g y revolution, dom inant firm s persisted as
leaders. The m o lecu lar b io lo g y revolution induced deep changes in the incentive
structures w ithin firm s and universities, w ith the advent o f university spin-offs and
the emergence o f the specialized new b iotech n ology firm s. In this process o f
adaptation and change, d ifferent dynam ic processes led to different patterns
o f com petition and p erform an ces (M cKelvey, O rsenigo, and Pam m olli 2004).
In telecom equipm ent and services, the early separation o f the radio spectrum for
use in one-w ay broadcastin g and tw o -w ay telephony gave rise to an oligopolistic
structure that persisted fo r quite a long tim e (D alum and V illum sen 2001). The
convergence w ithin IC T and betw een IC T and b roadcasting-audio-visual and the
emergence o f the Internet origin ated a m ore fluid m arket structure w ith a lot o f
different actors w ith different specializations and capabilities, and new types o f
users. This in turn greatly expanded the boundaries o f the sector b y creating new
segments and new o p p ortu n ities, and also by creating national differences in the
organization o f in n ovation . M oreover, the em ergence o f the Internet has generated
more pressure in favor o f open standards and has led to the rise o f new actors (such
as ISP and content p ro vid ers). In softw are, since the early 1980s, the spread o f
39^ FRANCO MALERBA
netw orked co m p u tin g, em bedded softw are, the In tern et, th e d evelo p m en t o f open-
system architecture and o pen source, and the gro w th o f w e b -b ased netw ork com-
p u tin g has led to the decline o f large co m p u ter p ro d u ce rs as d evelo p ers o f integrated
h ard w are an d softw are system s and to the em ergence o f a lo t o f specialized software
com panies. A lso, softw are d istrib u tio n has greatly ch an ged , fro m licensing agree
m ents in the early days, to the rise o f in d ep en d en t so ftw are ven d o rs, to price
d isco u n ts fo r package softw are, and, w ith the d iffu sio n o f th e C D -R O M and the
Internet, to sh arew are and freew are (this last on e p a rtic u la rly relevan t w ith Linux)
(D ’A d d erio 20 0 1). In m ach in e too ls, a m a jo r d riv in g fo rce fo r coevolutionary
processes is the d em an d fro m advanced cu sto m er sectors, n a m e ly the autom otive,
aeronautics, and defense indu stries, and the in creasin g use o f electron ic devices.
The em ergence o f new clusters that span several sectors, su ch as internet-soft
w a re-tele co m , b io tech n o lo g y -p h a rm a ceu ticals, an d n ew m aterials, is one o f the
m ost relevant current tran sfo rm atio n processes in sectoral system s. H ere a great role
is played b y the integration an d fu sio n o f p re v io u sly sep arated know ledge and
tech nologies an d b y the new relations in v o lv in g u sers, co n su m ers, firm s with
d ifferent specializations and com petences, an d n o n -firm o rg an izatio n s and insti
tutions grou n d ed in p revio u sly separated sectors.
14.8 P o l ic y Im p l i c a t i o n s
14*9 T h e C h a l l e n g e s A h e a d
T h is chapter has claim ed that in n o vatio n greatly d iffers acro ss sectors in terms
o f sources, actors, features, b ou n d aries an d o rg an izatio n . It has p ro p o sed an inte
grated an d co m p arative w ay to lo o k at sectors based o n the sectoral systems
fram ew ork.
Som e rem arks have to be advanced here in w ay o f co n clu sio n . T h e discussion o f
sectoral system s has show n that there cou ld be several levels o f sectoral aggregation,
an d that the choice o f one depends on the goal o f the analysis. W h ile the discussion
here has been very b ro ad in term s o f sectors in o rd e r to em ph asize linkages,
interdependencies, and tran sfo rm ation , fo r d ifferen t research goals the level o f
d isaggregation co u ld be m uch higher, at the level o f p ro d u c t g ro u p s. Still, we may
talk ab ou t system s o f in n o vation in this respect.
G eograp h ical bou n daries are a key d im en sio n to be co n sid ered in analyses o f
sectoral system s. N ation al b ou n d aries are n ot alw ays the m o st ap p ro p riate ones
fo r an exam in atio n o f structure, agents, an d co evo lu tio n . O ften , sectoral systems
are h igh ly localized and frequently define the sp ecializatio n o f lo cal areas (as in the
case o f m achinery, som e trad ition al indu stries, and even in fo rm a tio n technology).
For exam ple, m ach in ery is concentrated in specialized region al areas. Sim ilarly
sectoral specialization and local agglo m eratio n have o verlap p ed in R ou te 128 (for
m in icom p u ters) and in Silicon V alley (fo r p erson al co m p u te rs, softw are, and
m icroelectronics) (Saxenian 1994). M oreover, in the co n text o f transnational
econ om ic integration, the sector m ay m atter as m u ch o r m o re th an the national
system .
D ifferences across countries in sectoral system s have been relevant and have
affected countries international p erform an ce. In general, on e co u ld claim that
those coun tries that did not have effective sectoral system ch aracteristics did not
p erfo rm w ell in international m arkets. T h e sam e h old s fo r th o se co u n tries that tried
to replicate the success o f w orld leaders b y m im ick in g so m e o f the features o f the
sectoral system s o f the leading countries, w ith o u t h avin g the ap p ro p riate set o f
actors, linkages, and institution s. B y contrast, those co u n tries that have tried to
specialize in subsectors w ith products, know ledge, an d in stitu tio n al requirem ents
S E C T O R A L S Y S T E M S OF I N N O V A T I O N 401
that match their specific institutional fram ew ork have been successful (Coriat,
Malerba, and M o n to b b io 2004),
Finally, this chapter has tried to show h ow relevant a sectoral system approach is
for an understanding o f the features, determ inants, and effects o f innovation, in
terms o f research and policy. The p olicy aspect has been discussed in the previous
section and will not be repeated here, but research on sectoral systems m ay prove
very fruitful and has to m ove along several lines o f advancem ent.
N otes
1. In fact sectoral systems often have more than one technology, while the same technology
(as in the general purpose technology case) may be used by many different sectors.
2. Sim ilarly in addition to firm and non-firms organizations, also agents at lower and higher
levels o f aggregation such as individuals or consortia o f firms may be the key actors in a
sectoral system.
3. For example, sectors with competing technologies such as nuclear energy (Cowan 1990),
cars (and their power sources— Foreman-Peck 1996), metallurgy (ferrous casting— Foray
and Grubler 1990) and multimedia (VCR— Cusumano et al. 1992) show interesting
examples o f path-dependent processes.
R eferen c es
O r sen ig o , L., and P isan o , G. (1999), “ The Pharm aceutical In d u stry and the Revolu-
tion in M olecular Biology,” in M o w ery and N elson 1999: 26 7-312.
H u g h es , E P. (1984), I he Evolution o f Large T echnological System s,” in W. Bijker,
I. Hughes, and T. Pinch (eds.), T h e S o c ia l C o n stru c tio n o f T e ch n o lo g ica l System s, Cam
bridge, M ass.: M IT Press.
S E C T O R A L S Y S T E M S OF I N N O V A T I O N 405
Klepper , S. (1996), “ Entry, Exit, G row th and Innovation over the Product Life Cycle”
A m erican E co n o m ic R e v ie w 86: 562-83.
Langlois, R«, and S t ein m u e ller , E. ( 1 9 9 9 )> I h e Evolution of Competitive Advantage in
the Worldwide Sem iconductor In d u stry ” in M ow ery and Nelson 1999:19-78.
Levin , R., K lev o ric k , A ., N elso n , R., and W in t e r , S. (1987), “A ppropriating the Returns
from Industrial R & D ,” Brookings Papers on Economic Activity 3: 783-831.
Lundvall, R. A. (1993), National Systems o f Innovation, London: Pinter.
----- and J ohnson , B. (1994), “The Learning Economy,” Journal of Industry Studies 1 : 23-42.
McK elvey , M . (1997)» “ Using Evolutionary T h eory to Define Systems o f Innovation,” in
Edquist 1997: 200-22.
----- O rsenigo , L , and P a m m o l l i , F. (2004), “ Pharm aceuticals as a Sectoral Innovation
System,” in M alerba 2004,
----- A lm , EL, and R ic c a bo n i , M . (2002), D oes C o -lo ca tio n m a tter?K n o w le d g e collabo ration
in the Swedish Biotechnology-pharmaceutical Sector, Working Paper ESSY, http://www.ce-
spri.it/ ricerca/ESSY.htm
Malerba , F. (1985). T h e S e m ic o n d u c to r B u sin ess , M adison, Wis.: University o f W isconsin
Press.
------ (1992), “ Learning by Firm s and Incremental Technical Change,” E co n o m ic Journal 102:
845-59*
------ (2002), “ Sectoral System s o f Innovation and Production,” R esearch P olicy 31: 247-64.
* ----(ed.) (2004), S ectoral System s o f In n o v a tio n : C on cept , Issues and Analyses o f S ix M a jo r
Sectors in E u ro p e , C am bridge: Cam bridge U niversity Press.
------ and O rsen igo , L. (1996), “ Schum peterian Patterns o f Innovation,” C a m b rid g e Jo u r n a l
o f Econom ics 19: 47-65.
* ----------- (1997), “ Technological Regim es and Sectoral Patterns o f Innovative Activities,”
In d u stria l a n d C o rp o ra te C h a n g e 6: 83-117.
------------- (1999), “ Technological Entry, Exit and Survival: An Em pirical Analysis o f Patent
Data,” R esearch P o lic y 28: 643-60.
----- ------ (2000), “ Knowledge, Innovative Activities and Industry Evolution,” Industrial
a n d C o rp o ra te C h a n g e 9: 289-314.
------ and T o r r isi , S. (1996), “ The D ynam ics o f M arket Structure and Innovation in the
Western European Softw are Industry,” in M ow ery 19 9 6 :16 5-9 6 .
* M a r sili , O. (2001), The Anatomy and Evolution o f Industries: Technological Change and
Industry Dynamics, Cheltenham: Edward Elgar.
------ and V ersp a g e n , B. (2002), “ Technology and the D ynam ics o f Industrial Structures: An
Empirical Mapping o f Dutch Manufacturing,” Industrial and Corporate Change 11(4)*
791-815.
Mazzoleni , R. (1999), “ Innovation in the M achine Tools Industry: A Historical Perspective
o f the D ynam ics o f C om parative Advantage,” in M ow ery and Nelson i 9 9 9 : 169-216.
M etcalfe , S. (1998), E v o lu tio n a ry E co n o m ics a n d C re a tiv e D estru ctio n , London: Routledge.
M owery , D. (ed.) (1996), T h e In te rn a tio n a l C o m p u te r S o ftw a re In d u stry : A C o m p a ra tiv e
S tu d y o f In d u s tr y E v o lu tio n and Structure, Oxford: Oxford University Press.
* ----and N e l s o n , R. (1999), The Sources o f Industrial Leadership, Cambridge: Cambridge
University Press.
N e l s o n , R. (1993), National Innovation System s : A Comparative Study, Oxford: Oxford
University Press.
40 6 FRANCO MALERBA
*____ (i994), “ The Coevolution of Technology, Industrial Structure and Supporting Insti
tutions,” Industrial and Corporate Change 3: 47-64.
N elso n , R. (1995), “ Recent Evolutionary Theorizing about Economic Change, Journal of
Econom ic Literature 33: 48-90.
— - —and R o s e n b e r g , N. (1993)» “ Technical Innovation and National Systems, in R, Nelson
(ed.), N ation al Innovation Systems, Oxford: Oxford University Press, 3-22.
------and W i n t e r , S. (1982), A n Evolutionary Theory o f Econom ic Change, Cambridge, Mass,:
Belknapp Press.
O w en -S m it h , J., R ic c a bo n i , M., P a m m o l l i , F., and P o w ell , W. W. (2002), A Comparison
of US and European University-Industry Relations in the Life Sciences,” Working Paper
ESSY, http://wwwxespri.it/ricerca/ESSY.htm
*Р а ѵ іт т , K. (1984), “ Sectoral Patterns of Technical Change: Towards a Taxonomy and a
Theory,” Research Policy 13: 343-73.
PACE (1996) (Policy, Appropriability and Competitiveness of European Enterprises), Brus
sels: European Commission.
R iv a u d -D a n s e t , D. (2001), “ The Financing of Innovation and the Venture Capital, the
National Financial and Sectoral Systems,” Working Paper ESSY. http://www.cespri.it/
ricerca/ESSY.htm
R obson , M „ T o w nsend , J., and P a v it t , K. (1988), “ Sectoral Patterns of Production and
Use of Innovation in the U.K.: 1943-1983,” Research Policy 17 :1- 14 .
R o se n b e r g , N. (1982), Inside the Black Box, Cambridge: Cambridge University Press.
------(1998), “ Technological Change in the Chemicals: the Role of University-Industry
Relationships,” in Arora, Landau, and Rosenberg 1998:193-230.
S a x e n ia n , A. (1994), Regional Advantages, Cambridge, Mass.: Harvard University Press.
S c h e r e r , M. (1982), “ Interindustry Technological Flows in the U.S.,” Research Policy 11:
227-46.
S t e i n m u e l l e r , W. E. (2002) “ Embedded Software: European Markets and Capabilities,”
Working Paper ESSY, http://wwwxespri.it/ ricerca/ESSY.htm
------(2004), “ The Software Sectoral Innovation System,” in Malerba 2004.
T e e c e , D., and P isan o , G. (1994), “The Dynamic Capabilities of Firms: An Introduction”
Industrial and Corporate Change , 3: 537-56.
T e u b a l , M., Y inn o n , X, and Z u sco v itc h , E. (1991), “ Networks and market creation,”
Research Policy 20: 381-92.
F i r o l e , J. (1988), The Theory o f Industrial O rganization , Cambridge, Mass.: MIT Press.
T orrisi , S. (1998) Industrial Organisation and Innovation: A n International Study o f the
Software Industry, Cheltenham : Edw ard Elgar.
T u sh m an , M. L., and A n d e r s o n , P. (1986), “Technological Discontinuities and Organiza
tional Environments,” A dm inistrative Science Q uarterly 14: 311-47.
U t t e r b a c k , J. (1994), M astering the D ynam ics o f Innovation, Boston: Harvard Business
School Press.
V on H ippel , E. (1988), The Sources o f Innovation, Oxford: Oxford University Press.
Wengel , and S h a p i r a , R, (2004), “ Machine Tools: The Remaking of a Traditional
Sectoral Innovation System?” in Malerba 2004.
W inter , S. (1984), Schumpeterian Competition in Alternative Technological Regimes,”
Journal of Econom ic Behaviour and Organisation 5: 287— 320,
_ _ ( 19g7y "Knowledge and Competence as Strategic Assets,” in D. J. Teece (ed.), The
C o m p e titiv e C h a lle n g e : Strateg ies f o r In d u s t r ia l In n o v a tio n a n d R e n e w a l, Cambridge,
Mass,: Ballinger, 159-84.
C H A P T E R 15
I N N O V A T I O N IN
« Т Л ТДT . T p Г U ”
ЛЬшвЯШ ▼ V iJm im MR ^ m Jb * *Ш я*
INDUSTRIES
NICK VON T U N Z E L M A N N
V IR G IN IA A C H A
15.1 In tr o d u c tio n
m otor vehicles are included (O ECD 2003), so even if they could be expanded the
impact on GD P would be quite small. Governments need to give m ore thought to
the activities which generate most o f the output and em ploym ent o f their countries
and the best targets for “ dynamic comparative advantage for growth.
In this chapter we will consider not just tradition al4low-tech industries but also
those classified by the OECD as “ m edium -tech” We combine them hereafter as
“ low- and medium-tech” (LM T) industries. Our reason for banding the two to
gether is that both are being driven by similar factors which somewhat distinguish
them from the high-tech industries. These are frequently “ m ature” industries, where
technologies and market conditions may change m ore slowly. They can include non
m anufacturing activities, such as exploration for new resources in the oil and gas
industry. We naturally avoid much discussion o f services as being the subject o f the
following chapter, but this ignores the growing interrelationships o f service and
production/manufacturing activities in many relevant areas, so we include some
consideration o f the services sector.
Section 15.2 untangles the relationship between the technologies and markets
which comprise an “ industry” — effectively we reject O ECD -type classifications in
favor o f alternative sectoral taxonomies. This permits a m ore constructive analysis of
the key drivers o f change at the industry/sector level as between dem and and supply
factors in Section 15.3. Section 15.4 elucidates the roles o f firm strategies and
structures in the LM T industries. The implications o f this discussion for the evolu
tion o f industry structure, especially entry, are considered in Section 15.5, and
illustrations from some “ low-tech” industries are given in Section 15.6. Section 157
provides implications for government policy, which we believe requires radical
rethinking, and Section 15.8 concludes with a call for revising the academic agenda.
One simple consequence is that even “ old” products can be produced by, or partly
consist of, elements drawn from what had previously been a totally different set o f
activities. Equally, markets have become more blurred through the bundling of
goods and services (e.g. sales o f music products via the Internet).
As a result, conventional classifications o f sectors as high- or low-tech (etc.), as
long practised by the OECD, are becoming less and less useful for academic analysis,
though their sway still prevails in government policy making (see Section 15.7
below). To be fair, the OECD (2003) is coming— rightly— to place greater emphasis
on the “ knowledge-intensity” o f industries, based on criteria such as their use of
capital inputs from R&D-intensive industries, This necessitates rethinking the kinds
of taxonomies that help us to comprehend structural change (see Ch. 6 by Smith, this
volume).
There are many who believe that aU of the science and technology needed to dig oil and
gas out of the ground has been already discovered and deployed. Relatively low R&D
intensity figures attest to the same; over the period 1995 to 1997, the top teii oil majors
worldwide averaged an RBcD intensity o f 0.52 per cent However the accounting
category of R&D does not capture the full scope of investment into the search for
novelty and the applications thereof in the upstream petroleum industry Мшу
exploration activities that involve, in the words of the Frascati Manual, an Appreciable
element of novelty and the resolution of scientific and/or technological uncertainty^
and “ contribute to the stock of knowledge'" are captured under different budgetary
headings, notably exploration costs (Acha 2002). Mansfield (1969: 53) was also
concerned that geological and geophysical exploration were excluded from the R&D
definitions for precisely these reasons.
In fact, the annually growing numbers (in their hundreds per company) o f technical
and academic papers produced by the leading oil operating and service companies
reflect a substantially developed research program for the evolution of the science and
technologies underpinning this industry. This fact is better reflected in the number of
Fh.D. qualified staff from these firms actively working on a better understanding of the
composition and dynamics of this planet. This scientific endeavor in planetary science
does not go unnoticed even outside the industry. NASA has enlisted the help of
upstream oil industry companies in developing new technologies for drilling on
Mars (Babaev 2000). Likewise space scientists are collaborating with the geologists
and geophysicists who have been delineating marine space impact craters on Earth,
including one off the UK whose only natural analog exists on Jupiter"s ice moon,
Europa.
advantage in low-growth sectors (such as processed foods), but its production lies
towards the high-tech end o f these low-growth sectors (like applying biotechnology
to food processing).
precisely because they are “ labor-intensive” (though intensive in cheap skilled labor
rather than cheap unskilled labor).
Peneder (2001) provides a tripartite classification o f manufacturing industries, at
the 3-digit level o f disaggregation. One o f his taxonomies rests on factor intensity
(mainstream Le. average; labor-intensive; capital-intensive; marketing-driven; tech
nology-intensive), another on labour skills (low-skill; medium-skill blue-collar;
medium-skill white-collar; high-skill), and the third on external service inputs
(from knowledge-based services; from retail and advertising services; from transport
services; and from other industries). Only one o f the ninety-nine manufacturing
industries he lists (aircraft and spacecraft) has the classic high-tech profile of being
technology-intensive and predominantly using high-skill and knowledge-based
services; conversely there are labor-intensive industries which utilize high skills
(e.g. machine tools) and others utilizing knowledge-based service inputs (some
branches o f metallurgy). His classification underlines the great variety o f observable
combinations.
not essentially changed since the first large-scale use o f rubber; this promotes a global
oligopolistic structure (Box 15.2 discusses the tire industry more fully). But what
firms* in tires or elsewhere* choose to do as a result of strategic decision making—the
endogenous element— would rest on the profitability o f their strategic expenditures
in the face o f similarly strategic investment behavior by their competitors.
The modern tire industry has its origins in the nineteenth century and* by and large* its
development has mirrored that of the automotive industry. The world tire industry has
evolved to address several markets which each have different characteristics* direct
customers, and potential for growth. Over a century of development, it has segmented
across a wide number of markets, including automotive, aerospace, bicycle, and
locomotive tires. The industry is relatively highly concentrated; in 2000, the top ten
global tire manufacturers accounted for 83 per cent of the sales of the top seventy-five
tire manufacturers globally (www.tirebusiness.com/statistics).
Technology is applied by the tire manufacturers to reduce costs, to differentiate the
product line and to focus on greater value-adding activities (Acha and Brusoni 2003).
Facing a global market where it is more and more difficult to make a profit, the leading
manufacturers are continuously focusing on reducing costs through reducing
throughput and labor costs (including the long-awaited introduction of robotics),
innovations in processing technologies and source product (a new polyurethane tire
polymer), and in the method itself.
Beyond influencing the cost and ease of production, tire manufacturers have
invested in research and technology to also help them to move away from the
“commodity trap/5where products can only compete on price. Product differentiation
has occurred as companies offer a variety o f tire profiles and even colours to match cars
(the latter was led by Kumho, a Korean tire manufacturer that has successfully broken
into the top ten manufacturers). Such differentiation is certainly much more than
cosmetic; manufacturers have successfully incorporated new sensors into tire assem
blies and developed run-flat tires. Leading tire manufacturers are now looking to move
up the value chain by manufacturing entire tire assembly systems rather than simply
supplying the tire itself
O f course all o f this could be achieved through the support o f suppliers from the
high-tech sectors o f electronics and chemicals, yet the tire manufacturers themselves
patent in these areas and lead developments as applied to their business. Patenting in
technologies related to tires has increased, and most dramatically so for the tire
manufacturers (Acha and Brusoni 2003). The top ten tire manufacturers worldwide
had an average RScD intensity o f 4 per cent in 2000. Moreover, their patents are
applicable to a large number o f International Patent Classification (IPC) subclasses
(an average of forty-four subclasses over the period 1990 to 2000), indicating some
complexity in the nature o f these firms5knowledge bases. These firms have broadened
their focus to address the crucial interface between their chemical knowledge base (Le.
rubber and other chemical compounds) on the one side, and mechanical engineering
and electronics (e*g. sensors) on the other.
NICK VON T U N Z E LM A N N AND V IR G IN IA A C H A
414
Drawing on this reasoning, Davies and Lyons (1996) classified country strengths
on the basis o f dividing industries into four categories: those with high R8cD (i.e.
technology) expenditures; those with high advertising (i.e. sales) expenditures;
those with both; and those with neither. They showed that Western Europe was
relatively strong in the second and third o f these categories, i.e. those where both
R8cD and advertising were high (like pharmaceuticals) or advertising alone was
high. The latter includes particularly some industries norm ally regarded as “ low-
tech,” especially food processing (the industry selected by Sutton in his 1991 book to
validate his theory). The Sutton approach and associated taxonom y can be especially
useful for analyzing LM T industries, because supply (technology) is combined with
demand (product) aspects in a rigorous way.
15.2.5 Summary
Attempts to appraise innovation through adopting conventional sectoral classifica
tions can be quite misleading. Innovation is rapid in particular segments o f both
high-tech and LM T “ se c to rs e v e n if more segments o f the high-tech sectors display
such rapid innovation (for evidence see Ch. 6 by Smith in this volume). It is
admittedly possible to detach the high-tech segments from LM T “ industries” and
label them as new high-tech sectors, as was done for artificial fibers in textiles when
they arose to compete with natural fibers in the early twentieth century (see Section
15.6), but the final products remain very similar so this looks specious. Approaches
that blend the technology dimension and the product dim ension, such as those by
Sutton or by Peneder, appear to be not only m ore analytically satisfying but better
able to account for observed empirical differences between countries and regions. As
suggested below, they are also better able to account for dynam ic paths o f industrial
evolution over historical time. Furthermore, they are in our view a m ore advisable
platform for policy than simple OECD-style definitions o f high- or low-tech. They
need however to be supplemented by technology-oriented distinctions among
sectors such as the Pavitt taxonom y to provide a better grasp o f the nature of
structural change and competitiveness.
15.3 T h e K e y D r i v e r s
I he drivers o f change as they affect low, medium, and high-tech sectors can be
amilarly envisaged from the side of the products or from the side o f the technologies,
I N N O V A T I O N IN “ L O W - T E C H " I N D U S T R I E S 415
15.3.1.2 N ew Tastes
In addition to quality upgrading, consumers may switch their demand patterns to
goods which have new characteristics. While high-tech sectors may have greater
innate capacity to spawn product innovations, LM T industries may be faced with a
greater necessity to do so. Sectors such as food, energy generation and automobiles
have to confront intense pressure from communities and from governments to
produce safer and more environmentally friendly items. The same pressures extend
to the processes by which they produce these outputs. Less essential but often more
lucrative for LM T industries have been shifts in their product mixes to reflect the
changing composition o f consumers, for example the implications o f demographic
change (gender relations, ageing, etc.). These create new niches in which firms in
416 NICK VON TUNZELMANN AND VIRG INIA A C H A
low-tech industries can experience some resurgence. For exam ple, a leading Japan
ese toy manufacturer, Bandai, recently launched a new doll product range ( Primo-
puel") targeted not at children but at “ em p ty nesters" (women without children at
home) by embedding sophisticated electronic sensors and program m ing within this
“ real baby5' doll in an effort to develop a higher value market.
Both the potential for developing higher-quality products (“ quality ladders") and
new products therefore offset the seemingly inevitable m aturation o f older indus
tries, and give rise to new production and trade patterns (Grossm an and Helpman
1991). Indeed, over time, through adding value in processing and in new products,
the declines in demand for the products o f these industries have been less marked
than might be expected. In the case o f tires, companies have developed new model
profiles (e.g. “ fat" tires) to shift demand options from simple requirements (four
tires per vehicle) to a series o f options (different tires for different occasions), and to
differentiated product lines o f increasing quality (e.g. “ run-flat" tire systems).
The challenge for innovation strategies becomes how and how well firms in LMT
industries can alter their products and services and leverage the outcomes through
introducing better products or new products, as explored in Sections 15.4 to 15.6
below.
arise most commonly (though not invariably) in “ upstream” activities— in the cases
mentioned, in equipment and capital goods, in motive power and in basic mater
ials— from which they trickle down to user industries.
In LMT industries there is usually relatively little formal learning by science and
technology, at least at the firm level. Instead, innovation- and adoption-related
learning activities operate in practical and pragmatic ways by doing and using, The
bulk of the new technologies on which they draw, including these general-purpose
technologies, are developed by separate companies (rarely subsidiaries), specialized in
the relevant technological fields. However the downstream LMT industries need to
have “absorptive capacities” to make productive use of these upstream developments.
As shown by Cohen and Levinthal (1989, l 99o)> absorptive capacities are best incul
cated by doing some o f the innovative activities oneself, even if this means replicating
what has already been done elsewhere. Thus food-processing companies involved in
advanced (“ third-generation” ) biotechnology do not carryout much ofthe associated
research themselves but are often prominent in patenting in less advanced (“ second-
generation” ) biotechnology— this seems to provide them with the necessary absorp
tive capabilities. In addition, the national systems of innovation literature has drawn
attention to how formal science may congregate in national or regional laboratories in
such industries, instead o f being internalized within firms (Nelson 1993). The applic
able science produced by these organizations (e.g. public laboratories) may be generic,
but the applications themselves will require firm-specific absorptive capacities, gener
ated not just from formal R8cD but from broader-based innovative activities that
include engineering, continuous improvement processes, and organizational innov
ations such as integrated service and supply
15.4 F i r m s a n d C o r p o r a t e C h a n g e
i n LMT I n d u s t r i e s
This section raises issues concerning strategies and structures at the firm level, issues
o f scale and scope in both Carge firms and SM Es, and discusses different kinds of
integration.
other radical technological changes, which were bound to affect all their carefully
separated divisions (Mowerу 1995).
necessitated by radical change (D o d gso n and Rothw ell 1995). Q uestions o f access to
new technologies and on w hat term s then becom e param ount. This becom es sig
nificant in those L M T indu stries w hich are dom inated b y SM Es, which include many
of the more traditional areas. W hilst SM E s are frequently seen as a cause o f hope in
high-tech sectors, they are often (perhaps unfairly) seen as a m atter for despair in
LMT sectors.
directly producing only a m inority o f the value added by their “ system/5 Moreover
system integrators could emerge at varying points on the value chain, so that vertical
links might increasingly be occurring between various system integrators, That is,
each integrator would be surrounded by a network o f suppliers and related activities'
Within these complex structures o f interrelationships, new power balances could
emerge, In the lower-tech industries especially, like textiles and some branches of
food, the manufacturing stages o f the “ chain55were squeezed as power tended to shift
downstream to the final stages and even to the retailers, The growing competition in
diversified products in conjunction with the rising incomes and influence of con
sumers, often dissatisfied with what they are getting, is driving this trend. It remains
to be seen whether this pattern will proceed further in the direction o f buyer-driven
rather than producer-driven chains (Gereffi 1999).
15.5.3.2 D yn a m ic C om petition in T im e
A cross the full range o f industries, the m o d ern era is su p p o sed to be characterized by
co m p etition that has intensified because o f glo balization an d becau se o f m ore rapid
I N N O V A T I O N IN “ L O W - T E C H ” I N D U S T R I E S 425
change m m arket dem and. In L M T industries the pace o f change and com petition
may be less intense, as m arket leaders seek to retain their predom inance by brand
loyalty. In the tobacco industry, the trajectory o f a new product launch is norm ally to
begin with a new bran d m arketed at low prices and with high tobacco content. Once
loyalty is obtained the produ cer relaxes the prices and/or the m aterial content until
eventually consum ers lose faith, w hereupon a new brand is launched on the same
terms to eclipse it. Equally, brand loyalty can be extended through globalized
marketing.
Thus m arketing-based ap p ro p riab ility offsets som e o f the pressures to develop
new products. H ow ever firm s in L M T industries are arguably alm ost as susceptible
as those in higher-tech industries to innovations that accelerate the developm ent
times (“ cycle tim e” ) an d rate o f application and diffusion for new technologies. In
the clothing industry, Benetton com bined its production netw ork o f small suppliers
with a sophisticated IC T system for feedback from custom ers, achieving rapid global
growth in a relatively m ature in d u stry (Belussi 1987). For the m ore basic production
technologies, the L M T industries are distinguished less by the pressure to innovate
than by the difficulties encountered in app lyin g those basic technologies to purposes
that m ay be very different fro m their origin al intentions. H aving to confront such
difficulties am plifies rather than dim inishes the im portance o f speed to m arket for
competitiveness even in L M T industries. The key issue here is surely the real-tim e
achievability o f econom ies o f scope— do integrated firm s or outsourced networks
effect this better?
15.6 In n o v a t io n in S o m e L o w -Tech
In d u st r ies
terms o f corporate structures as well as technologies and are thus often thought o f
not as textiles proper, but this hardly seems reasonable. In m ore recent times
computerized technology is making inroads into the still m ore fragmented clothing
sector, while genetics is assisting advances in textiles. M icrofibers o f exceptional
fineness have evolved for particular market niches and have begun penetrating
standard clothing segments.
It is indeed the case that the recent developments have been slow to diffuse, but this
probably has less to do with technological limits than with organizational aspects. The
textile-clothing industry is still largely based on a pre-industrial vertical structure that
is highly segmented and allows countries with low wage rates to enter through the use
o f relatively simple technologies, such as sewing machines. To leverage new textile
technologies in the way we have described, change in production technologies may
have to be linked to product changes, as for instance with microfibers.
The demand side normally offers greater prospects for leverage than the supply
(technologies) side in the textiles and clothing sector. As incomes rise, consumers are
willing to pay extra for fashionable brand-names. M any o f the observed changes in
what is fashionable rest on styling rather than technological innovation, though
naturally that does not imply any lesser need for creativity. Even here, however, new
technology may help bridge the gap between designers, producers, and their markets.
An important innovation by Benetton was the firm s use o f IC T to learn quickly about
changes in market tastes in order to instruct their suppliers (Belussi 1987).
15*6.2 Food-processing
A second industry frequently, and often inappropriately, classified as low-tech is
food manufacturing. The “ industry” is characterized by a huge variety o f organiza
tional forms, related principally to the extent to which marketing can be leveraged
(Sutton 1991). It has been custom ary to see the industry as “ supplier-dom inated”
according to Pavitfs taxonomy, resting on a dependence on suppliers o f the ma
chinery for production, but this is changing nowadays in some critical ways. As such,
the industry is becoming more market-driven, but one cannot reasonably leave
technology out o f the picture.
Like all industries, the technical efficiency o f this industry comes from the
knowledge bases it draws upon, but here there exists an unusual am ount o f variety
to go with the variety o f organizational structures. The range o f expertise from
science (e.g. food microbiology) through to engineering has to span not just
production conditions but sanitation, quality assurance, environmental acceptabil
ity, and so on. The range of new technologies currently being drawn upon in
this industry covers almost the whole spectrum. The traditional reliance on suppliers
o f machinery is being overtaken by needs for technologies from advanced
I N N O V A T I O N IN 4L OW - T E C H >J I N D U S T R I E S 427
15*8 C o n c l u s i o n s
The underlying argument o f this chapter is that low-tech sectors do not lack for
technological opportunities, nor indeed for appropriability and other factors asso
ciated with benefiting from technological innovation. Indeed, our principal analyt
ical conclusion is that in the modern world there are no true low-tech sectors.
Instead what we observe is a varying degree o f permeation o f high technologies into
low-tech and medium-tech as well as into high-tech sectors.
The primary issue for further research is thus the need to explore much more
critically how industries and sectors ought to be classified. The OECD itself appears
NICK VON T U N ZE L M ANN AND V IR G IN IA A C H A
N otes
1. Fram es should not be confused w ith organization al “ cu ltu re” ; the latter com prises the
values established at the group level and interpreted by in d ivid u als, w hereas frames
com prise the interpreted relationships between ph en o m en a (and their associated values)
and are com posed o f individuals, m ediated at a group le v e l
R eferen c es
A c h a , V. L. (2002), “ Framing the Past and Future: The Development and Deployment of
Technological Capabilities in the Upstream Petroleum Industry,” Unpublished D.Phil.
Thesis, SPRU, University of Sussex.
and B r u so n i , S. (2003), “ C o m p lexity and In d u stry E vo lu tion : N ew in sights from an
O ld Industry, in Best Papers Proceedings o f 2002, E u ropean A sso ciatio n for Evolutionary
Political E co n o m y Conference, 2003.
B a b a e v , H . (2000),
Scientists, Engineers Team ing up to D evelop M ars D rillin g Technol
ogy,” reprinted in O il a n d G a s Jo u r n a l 98 (17).
B a ld w in , J. R. (1995I T h e D y n a m ic s o f In d u s t r ia l C o m p e titio n : A N o r th A m e r ic a n Perspective ,
Cambridge: Cambridge University Press.
and S a b o u r in , D. (2002), A dvanced Technology Use and Firm Perform ance in
C anad ian M an u facturin g in the 199ns, In d u s t r ia l a n d C o rp o ra te C h a n g e 11: 761—90.
B e l u s s i , F. (1987), Benetton, In form atio n le c h n o lo g y in P ro d u ctio n an d Distribution:
A Case Stu d y o f the Innovative Potential o f T raditional Sectors,” S P R U O ccasional Papers
25, SPRU , U niversity o f Sussex.
I N N O V A T I O N I N " L O W - T E C H ” I N D U S T R IES 431
N elso n , R, R. (ecL ) (1993), National Innovation Systems: A Comparative Analysis, New York:
Oxford University Press.
OECD (1994). The Measurement of Scientific and Technical Activities: Proposed Standard
Practice for Surveys of Research and Experimental Development (Frascati Manual: 1993)»
Paris: OECD,
------ (2003), Science, Technology and Industry Scoreboard 2003 - Towards a Knowledge-Based
Economy Paris: OECD.
Q r l i k o w s k i , W, J. (2000), “ Using Technology and Constituting Structures: A Practice Lens
for Studying Technology in Organizations,” Organization Science 11: 404-28.
------ and G a s h , D. C. (1994), “ Technological Frames: Making Sense of Information Tech
nology in Organizations,” ACM Transactions on Information Systems 2 : 174-207.
*P a v it t , K. (1984), “ Sectoral Patterns of Technical Change: Towards a Taxonomy and a
Theory” Research Policy 13: 343-73.
*P e n e d e r , M. (2001), Entrepreneurial Competition and Industrial Location, Cheltenham:
Edward Elgar.
P en ro se , E. T. (1959), The Theory of the Growth of the Firm, 3rd edn., Oxford: Oxford
University Press, 1995.
P isc it e l lo , L. (2003), “ Generation and Applicability of Technological Competencies: An
other way of Measuring Coherence in Corporate Diversification,” paper for Conference in
Honour of Keith Pavitt, Politecnico di Milano.
P o r t e r , M. E. (1985), Competitive Advantage: Creating and Sustaining Superior Performance,
New York: Free Press.
R o sen ber g , N. (1963), “ Technological Change in the Machine Tool Industry, 1840-1910”
Journal of Economic History 23: 414-46.
------(1976), Perspectives on Technology, Cambridge: Cambridge University Press.
R u m elt , R. (1974), Strategy; Structure and Economic Performance, Boston: Harvard Business
School Press, rev. edn., 1986.
S to n em a n , Р. L., and D a v id , P, A. (1986), “Adoption Subsidies vs. Information Provision as
Instruments of Technology Policy”, Economic Journal, Conference papers, 96:142-50.
S u tto n , J. (1991), Sunk Costs and Market Structure, Cambridge, Mass.: MIT Press.
*------(1998), Technology and Market Structure, Cambridge, Mass.: MIT Press.
T id d , J., B e ssa n t , J., and P a v it t , K. (2001), Managing Innovation: Integrating Technological,
Market and Organizational Change, Chichester: Wiley.
* v o n T u n zelm a n n , G. N. (i995)> Technology and Industrial Progress: The Foundations of
Economic Growth, Cheltenham: Edward Elgar.
W o m ack , J. P., J o n es , D. T. and Roos, D. (1990), The Machine that Changed the World, New
York: Rawson.
C H A P T E R 16
I N N O V A T I O N IN
SERVICES
IAN M ILES
Output and employment in the service sectors have grown significantly throughout
the industrial world since the 1950s. Table 16 л shows that as early as the 1970s, services
434 IAN M ILES
Table 16.1 Share ( % ) of gross value added Inservices in total GDP in EU countries
constituted more than half o f the value added in European Union countries, and by
the new century they contributed over two-thirds.
HORECA (hotels, restaurants and catering) is dom inated by food preparation and
delivery, and includes other elem ents o f hospitality, entertainm ent, and com fort.
Social and collective services such as public adm inistration and health and educa
tional services are delivered largely or entirely through the state, though patterns o f
organization v ary a great deal over tim e and across countries: the back-office
operations o f such bodies can be h igh ly IT-intensive. Business services include
practical support w ith logistics and office and building services, as well as support
for adm inistrative m atters (such as law and accountancy) and technology
support, such as com pu ter and engineering services. This sum m ary by no m eans
exhausts the variety o f activities and technologies included in the services sector.
Services m arkets also are diverse, spanning consum ers, businesses, and the public
sector and its clients. The tran sform ation s effected by these services industries
operate on such diverse “ raw m aterials” as hu m an clients (as well as som e other
biological organism s, e.g. veterinary services), physical artifacts (they m ay be
repaired, m aintained, stored, tran sported, tested, integrated into larger system s. . . ) ,
and data, sym bols and in fo rm atio n (that m ay be processed, stored, telecom m uni
cated, etc. b y services like financial industries as well as b y com puter and com m un i
cations services, etc.). T h e technical skills dem anded o f the w orkforce range from the
minimal ones used in fast fo o d outlets and office cleaning to the professional
qualifications o f m arket researchers and architects, and the scientific and engineer
ing credentials o f sta ff in specialized R & D firm s.
This diversity m eans that any generalizations about the nature o f services and
innovation in services m ust be qualified b y num erous exceptions. Som e services are
more like m an ufactu rin g in term s o f som e param eters— som e are technology
intensive (e.g. m edia, telecom m unications), som e w ork w ith m aterial artifacts
(e.g. rapid pro to typ in g, repairs). A n d the operations o f m any m anufacturing
firms include a great m an y “ services” activities (e.g. transport and logistics, office
work, m arketing and aftersales). N evertheless, a set o f com m on features charac
terizes m any services and differentiates them from m anufacturing.
For exam ple: m an y services products are intangible ones, w hich m akes it harder to
store, transport, and exp ort them than is true o f m anufacturing products. H istoric
ally, m any service in n o vation s have been difficult to protect via patent m echanism s,
although this situation has begun to change in som e sectors such as com puter
software and services, or F IR E (See H all 2003; G raham and M ow ery 2003, FhG . 151
2003). Services are typ ically interactive, involvin g high levels o f contact between
service supplier and client in the design, produ ction, delivery, consum ption, and
other phases o f service activity. Service products are often produced and consum ed
in the course o f su p p lier-clien t interaction at a particular tim e and place ( coter-
minality” ). In novations m ay focus on this interaction as m uch as on conventional
product and process characteristics, and m ay rely less on technical know ledge and
more on social and cultural nous. M an y services are highly inform ation-intensive,
with a preponderance o f office-based w ork or com m unicative and transactional
43б IAN MILES
o peratio n s, such as telem arketing. Som e service p ro d u cts are d eliverab le electronic
ally, such as text-based reports, T V p ro g ram s, m u sic reco rd in gs, c o m p u te r software,
an d w ebsites. T h e in fo rm atio n al co m p o n en ts o f m a n y o th er services are subject to
IT -based in n o vation . B u t in oth er respects, the con trasts w ith in the service sector are
as sign ifican t as those d ifferen tiatin g the sector fro m m a n u fa c tu rin g . A fter all, the
sector in cludes the m o st concentrated, k n o w led g e-in ten sive, an d IT-intensive
sectors in m o d e m in d u strial econ om ies (b an kin g, p ro fe ssio n al services, etc.), as
w ell as the least (retail, cleaning, etc.).
R&D performance by the U.S. service-sector industries underwent explosive growth between
1987 and 1991, driven primarily by computer software firms and firms performing R&D on a
contract basis. In 1987, service-sector industries performed less than 9 per cent o f all R&D
performed by industry in the United States. During the next several years, R& D performed in
the service sector raced ahead o f that performed by U.S. manufacturing industries, and by
1989, the service sector performed nearly 19 per cent o f total U.S. industrial R&D, more than
double the share held just two years earlier. By 1991, service-sector R&D had grown to
represent nearly one-fourth o f all U.S. industrial R&D. Since then, R&D performance in
U.S. manufacturing industries increased and began growing faster than in the burgeoning
service sector... Unlike the United States, Japan has yet to see a dramatic growth in service-
sector R & D . . . R&D in Japan s service-sector industries reached 4,2 per cent o f the total R&D
performed by Japanese industry in 1996 and 4.5 per cent in 1997
R&D within the EU s service sector has doubled since the mid-1980s, accounting for about
11 per cent o f total industrial R&D by 1997* Large increases in service-sector R&D are apparent
in many EU countries, but especially in the United Kingdom (19.6 per cent of its industrial
I N N O V A T I O N IN S E R V I C E S 437
R&D in 1997)3 l^ly (15.3 percent), and France (10.0 percent). (National Science Foundation
2002: ch. 6)
This picture o f service-sector R & D investm ent underm ines the view that innov
ation in services results solely from the sector’s adoption o f m anufacturing
innovations.1 A bun dant case-stu d y evidence also highlights innovation by services
firms (see e.g. A ndersen et al. 2000; T idd and H ull 2003). Indeed, som e service
industries invest h eavily in R8cD and pursue R & D program s that are at least as
sustained as are those o f m an u factu rin g. Som e useful inform ation on corporate
R&D can be gleaned from co m p an y annual reports and accounts, and C om pan y
Reporting Ltd. produces an annual analysis o f these data for the U K 's D epartm ent o f
Trade and In dustry (D T I). The 2003 analysis (D T I 2003) reports that ITservices rank
fifth among all industries in w o rld w id e R & D spending in 2003, and sixth am ong U K
industries in 2003 R & D spending. The analysis in this report o f R & D spending by IT
firms indicates that M icro so ft ranks eleventh in w orldw ide tabulations o f corporate
R&D spending, British Telecom is the fifth largest R & D spender in the U K , and
Reuters the eleventh. Fu lly 6 per cent o f the total R & D spending b y the world $
largest R & D investors— the top 700 com panies, w hose R & D spending exceeded
£35 million in 2002— is accounted for b y firm s classified as m ainly involved in
software and IT services. (T h e equivalent figure for the U K alone is 5 per cent.)
Data from the C o m m u n ity In n o vatio n Survey (C IS) also indicate that services
firms are m ajor innovators, and that it is not just “ high-tech” services (like software
and telecom m unications) w hich p lay significant roles in technology developm ent.
Although there are in n ovative activities in all branches o f services, the innovation
surveys depict low er levels o f Innovation and R & D investm ent w ithin services, on
average, than w ithin m an u factu rin g. The adoption by services firm s o f technologies
produced in other sectors is indeed a m ajor form o f innovation in the sector, m uch
of which thus displays “ su p p lier-d riven ” characteristics.
d iv isio n o f lab o r and h igh er levels o f technology. In fact m a n y services have been
h ig h ly standardized and tech n o logy-in ten sive fo r a lo n g tim e c o n sid e r railways
an d con ven tion al teleco m m u n icatio n and b ro ad cast services.2 B u t it is app aren t that
the grow th o f large-scale firm s in o th er service sectors is asso ciated w ith a form o f
increased stan d ard ization . M c D o n a ld ’s an d o th er fast fo o d restau ran ts are one
fam iliar exam p le o f such firm s in the service sector. Fast fo o d ch ain s also display a
m easure o f cu sto m izatio n , in that their p ro d u cts are co m p o se d o f v a rio u s co m p o n
ents, o r m odules, w h ich can be co m b in ed in n u m ero u s w ays a cco rd in g to custom er
d em an d ; n ew m odu les m a y be added to increase v ariety and su p p o rt o th er form s o f
in n o vatio n .
T h e in d u strializatio n o f services has been criticized as gen eratin g low -quality,
lo w -skill jo b s (M cD o n ald izatio n , o r M cjo b s, to so m e co m m en tato rs). O ther types
o f in n o vatio n and reorgan ization in services m ay p ro d u ce n ew fo rm s o f social
exclusion. F o r exam ple, the use o f call centers an d o th er elem ents o f banking
a u to m atio n often occu rs in tandem w ith the closu re o f tra d itio n al outlets, such as
h igh street branches. O u t-o f-to w n h yperm arkets have h ad an ad verse effect on high
street sh o p p in g areas an d their environs. Fears have been expressed that consum er
e-com m erce w ill have sim ilar effects, since these n ew ch an n els o f cu sto m er contact
m a y not be available to all consum ers. Such co n cern s m a y trig g er consum er,
regu lato ry o r legislative responses that w ill affect service m ark ets an d innovation
strategies.
B u t exactly w hat is con n oted b y the “ in d u strializatio n ” o f services cited b y Levitt?
A fter all, m an y m an u factu rin g firm s n o w em phasize flexib le sp ecialization , mass
cu sto m izatio n , reintegration o f h igh ly atom ized d iv isio n o f lab or, a n d the like. Som e
o f these trends are m ak in g m an u factu rin g m o re like services an d so m e are being
em ulated b y large services co m pan ies, even as oth er service o rg an izatio n s continue
to fo llo w m o re classical in d u strializatio n trajectories.
O ne w ay in w h ich services have em ulated m a n u fa ctu rin g is in the ad o p tio n and
d evelopm en t o f an organ ization al in n o vatio n — q u ality co n tro l p ro ced u res. In ser
vice firm s, as has been tru e o f m an y m an u factu rin g firm s, atten tio n to q uality has
often served as a trigger to in n o vatio n , b y req u irin g firm s to v ie w th eir services as
consistin g o f a n u m b er o f co m p o n en t parts to w h ich q u a lity co n tro l prin ciples can
be applied. T h is is typ ically used to id en tify areas w h ere there is w eakness in service
p erfo rm an ce and w here change thus needs to be engineered. In fo rm a tio n Technol
o gy is often in trodu ced in the context o f im p ro v in g cu sto m er service quality,
especially in speeding responsiveness v ia m eans such as call centers. U n derstan ding
the co m po n en ts o f a service also often generates in sigh ts in to w ays in w h ich these
co m pon en ts can be tran sfo rm ed o r recon figu red into n ew service b u n d les (Sundbo
1998). M o d u larizatio n underpin s m uch services in n o va tio n , since d eco m p o sitio n o f
service processes and /or produ cts m ay sp u r process in n o va tio n an d the identifica
tion o f new p ro d u cts and pro d u ct co m b in ation s. T h is typ e o f in n o vative activity
does not necessarily rely on R & D investm ent,3 th o u gh in the sectors w h ere there has
I N N O V A T I O N IN S E R V IC E S 439
been much talk o f u n b u n d lin g — softw are and telecom m unications— there is
much conventional R & D .
Another m ajor im petus to “ services industrialization” has been the application o f
IT. This has m ade it possible to autom ate elem ents o f the back-office w o rk o f m any
service firms and large firm s in o th er sectors— for instance through the use o f
document processing, em ail, E nterprise Resource Planning software and systems,
etc. IT is widely applied in back-office w ork, and in the m anagem ent and execution
of customer-facing services such as m ail and telephony. M any o f these IT applica
tions make possible the pro vision o f custom ized services through recom binations o f
standard service m odules. B ack-office autom ation through IT also has changed the
spatial location o f service activities. Telephone call centers— dedicated offices where
the work revolves aro u n d the com puter-assisted answ ering o f telephone calls,
normally for the p ro visio n o f rou tin e custom er service inform ation— are a case in
point.4 Over the last decade we have seen the relocation o f call centers in the U K to
lower-wage areas o f the cou ntry, and m ore recently to the Indian subcontinent— a
specific m anifestation o f the process o f “ offshoring.” There is considerable debate as
to the extent to w hich higher-level office service w ork will also be subject to
offshoring— som e quite sophisticated softw are activities have follow ed m ore basic
programming w o rk in the m ove to overseas locations. The new international
division o f labor clearly involves a redistribution o f service as well as m anufacturing
and extractive activities.5
* the historically specific nature o f the story (after assim ilating new IT and going
through an R P C process, w ill services then follow the classic product cycle?).
Uchupalanan (1998, 2000) m ounted a system atic critique o f the R P C approach,
tracing the history o f five IT in n o vatio n s10 through all firm s in the banking services
sector in Thailand. He uncovered a diverse range o f innovation strategies and
trajectories that were far richer than the R P C account. The banks were influenced
by the strategies o f co m petitors w ith respect to each given innovation, b y their
experiences w ith earlier in n ovation s (and their plans for others), and b y pressures
from regulators and the m arket. The interrelation o f m arket com petition, firm
circumstances and in n ovation dynam ics m eant that the R P C “ sto ry” o f innovation
processes was rarely applicable in this context. At best it was one o f a num ber o f
possible patterns o f developm ent.
The R PC approach, w hich has proven influential in m any recent analyses o f
services innovation, m ay neglect n o n -IT innovations and innovations undertaken
in earlier periods o f service innovation. But this fram ew ork's attention to the
increasing innovative activities o f service firm s, as well as the historic om ission o f
their innovative activities from m ost scholarship, m eans that the R P C is an im p o rt
ant contribution, p ro vid in g a starting point for further research on services in n ov
ation.
Other studies o f services in n ovation place m ore em phasis on the relation between
service firm s and their clients. O ne line o f w ork uses the term “ servuction” (services
production, intended to high light analogies with m anufacturing production activ
ities) to describe the activities and procedures involved in producing and sustaining
supplier-client relations and delivering “ service” — rough ly what we call “ interac
tivity” above. Belleflam e et ah (1986) classified a sam ple o f innovations from service
companies in term s o f w hether these centered on servuction, production, or a
combination o f both (w ithout assum ing that these w ould necessarily be IT-centered
innovations),11 and fo u n d n u m erou s exam ples o f all types o f innovation. Tordoir
(1996) studied profession al services, and noted very different patterns o f interaction
between suppliers and clients, in trodu cin g a distinction between “ jo b b in g” relations
(where the service su p plier provides a relatively standardized service), and “ spar
ring” ones (where the su p plier and client negotiate the details o f what service is to be
provided, and h ow ). M o re recent studies have addressed the question o f how service
innovation differs according to the degree to w hich services are standardized or
specialized to specific clients (cf. H ip p et al. 2000, 2003). The m ore interactive
services w ould seem to require greater exchange o f know ledge between service
firm and client, and the learning processes involved can be a fertile basis for
innovation: how ever, m an y o f the innovations are liable to be o n e -o ff solutions for
particular problem s, o r elaborate custom izations o f m ore generic solutions.
The R P C approach argues that IT foased innovation begins w ith back-office
processes and then m oves fo rw ard into functions involvin g m ore custom er contact.
4 42 IAN MILES
16.3 I n n o v a t i o n S y s t e m s a n d t h e
O r g a n i z a t i o n of I n n o v a t i o n
IT pro vid es a generic tech n ology that can be w id ely ap p lied across service sectors— a
substantial break w ith past experience, w here even tech n o logies like autom obiles
and typew riters, alm ost u n iversally adopted in certain services b ran ch es, w ere rarely
I N N O V A T I O N IN S E R V I C E S 445
applied in others.14 T h o u gh services vary in the extent and speed with which they are
adopting these innovations, and som e o f them are m ore relevant to certain branches
of services than to others, m any technological opportunities are opening up for all
services. Even personal services such as hairdressing are subject to innovations—
there are shops w here video tech nology and PC s are used to give clients an im pres
sion o f how they w ou ld look w ith different hairstyles, for instance. And IT is a very
configurational technology, w hich challenges users to develop new software and
interfaces, and to b rin g diverse elem ents o f technologies together in new ways.
Their increased tech n o logy-in ten sity has not yet transform ed all o f the features o f
innovation in services. M an y services are argu ably laboring under a heritage derived
from past periods w here few generic technologies found ready application in their
activities. The low tech n ology-in ten sity that previously characterized services meant
that many services firm s paid little attention to strategies for the m anagem ent or
adoption o f innovation. Lim ited use o f advanced technology w ould also mean that
most service firm s w o u ld have little incentive to be linked into innovation systems
that would connect them w ith those responsible for generating the new technolo
gies. And in the w id er in n o vation system s them selves, few o f the innovation-related
facilities offered b y institution s such as university departm ents, research institutes,
and government laboratories are tailored to the requirem ents o f services. It is thus
not surprising that few service firm s m ake m uch use o f these resources currently
(Institute o f In n ovation Research 2003).
There are, as alw ays w hen discussing services, exceptions, For exam ple rail,
broadcasting, and telecom m u nications services, m any o f which were state-owned
and very large organizations for m uch o f their existence, were closely related to
manufacturers. These large services firm s frequently ran their own laboratories,
testing sites, technical train in g program s, etc. O ther exceptions include very large-
scale service firm s in sectors like financial and retailing services. Especially in the
latter case, superm arket chains have becom e adept m anagers o f their ow n supply
chains, and are often active in dictating production processes and products—
including innovations— in their agricultural and m anufacturing suppliers. For
example, their suppliers m ay be requested to adopt specific environm ental or anim al
husbandry practices, to use ecom m erce techniques, etc. As well as influencing their
suppliers, such firm s have been pioneers in som e sorts o f technology for their own
use— for exam ple, data w areh ou sin g and data-m ining m ethods.15 Another, and
particularly im portant, class o f exceptions, are the business services that are, as we
shall see, im portant actors in in n o vation system s, contributing to innovation across
the economy. C onsultancies, training organizations, and m any firm s helping to
service new technologies are ju st a few o f the agents involved here- not to m ention
specialized R & D and design services themselves!
But relatively few service firm s and sectors have strong links w ith national or
regional innovation system s. T h e organization o f innovation is a new developm ent
in m any service firm s— for instance, in a set o f recent interview s the author found
446 IAN MILES
presence and use of KIBS enhances the performance of economic sectors and
regio n s.23
K IB S in clu d e all kinds o f business services that are fo u n d ed u p o n h ig h ly special
ized kn o w led ge— social and in stitu tion al k n o w led ge in m a n y o f the traditional
p ro fessio n al services* o r m o re tech n o logical an d tech n ical k n o w led ge. These
firm s typ ically have v ery h igh levels o f q u alified staff* su ch as u n iv e rsity graduates.24
S o m e K IB S are based on adm inistrative* legal* m arketing* o r sim ila r knowledge.
O thers are directly based on scientific an d tech n o lo gical k n o w led g e— testing* proto
typing* en viron m en tal services* engineerin g con su ltan cy, etc. T h e knowledge
requirem en ts fo r tech n o lo gy users are b o u n d to be m o re ch allen g in g w here new
tech n o lo gy is involved. Firm s are less lik ely to have acq u ired the k n o w led ge neces
sa ry to und erstand , m aster, and utilize n ew p ro d u ct an d p ro cess o p p o rtu n ities.25
T h u s, m an y technical K IB S focus on new tech n o logical o p p o rtu n itie s— exam ples
include W eb and Internet* softw are and co m p u ter services; o th ers on the production
and transfer o f know ledge ab ou t new tech n o logy; in fo rm a tio n an d tra in in g services*
fo r exam ple.
T ech nology-related K IB S sectors are a m o n g the m o st active in n o vato rs in the
econom y* as indicated b y C IS an d other data (e.g. D T I 2003, w h ich includes data
in d icatin g the h igh levels o f R8cD expenditure* patenting* etc. o f softw are and IT
services).26 We saw above that Tether and Sw ann (2003)* u sin g C IS 3 d ata fo r the UK*
fo u n d that m o st services are p o o rly linked w ith the p u b lic elem ents o f innovation
system s (e.g. in term s o f so u rcin g in fo rm a tio n fro m an d c o llab o ra tin g w ith univer
sities). B u t they also rep ort that som e technical K IB S w ere o u tsta n d in g ly w ell linked
to these co m p o n en ts o f in n o vatio n system s. S o m e tech n ical services (e.g. contract
R & D ) had u n iq u ely h igh levels o f in teraction w ith the p u b lic scien ce base* higher
than any m an u factu rin g sector; however* IT services an d m o re p ro fe ssio n al services
tended to have lo w levels o f contact* relying m o re o n p ro fe ssio n al association s and
the like to refresh their know ledge.
M a n y K IB S p lay im p o rtan t roles in in n o vatio n processes in their client firm s and
sectors. These roles m ay not be altogether n ew — som e h ave existed fo r m any
decades. B u t there has been considerable grow th o f e m p lo y m e n t an d output in
K IB S sectors* w h ich can o n ly m ean that their use has also expanded* an d suggests
that their significance for in n o vation across the e c o n o m y w ill also have m ush
room ed,
W hat are these roles? K IB S m ay p ro vid e the firm w ith general in fo rm a tio n about
its internal o p eratio n s and external environ m en ts. T ech n o lo g y and innovation-
related in fo rm atio n often form s p art o f this. K IB S m ay sim p ly p lay a role in
id en tifyin g the nature o f a p articu lar p ro b lem o r class o f p ro b lem s co n fro n tin g the
firm (fo r instance* that com petitors are lau n ch in g p ro d u cts w ith new fu n ction al
ities* that regulations m ay m ean that processes w ill have to generate less o f a
p articu lar pollutant* etc.). K IB S m ay p ro p o se w ays o f so lv in g a technological
problem (fo r exam p le 1 ecom m en d in g that p articu lar strategies are undertaken for
I N N O V A T I O N IN S ER V IC E S 451
product or process innovation). They may provide advice (for example, recom
mending a specific technological solution), or actually implement such a solution
on a “turnkey” or long-term basis (as in the case of systems integrators and facilities
managers). The KIBS relate technological knowledge to the specific problems
encountered by the client. German survey data indicate that technology-related
KIBS are more likely to produce specialized products that are tailored to client needs
than were the other services studied (Hipp et al. 2003).
In addition to expertise in specific industrial, technological, or functional
domains, KIBS professionals require skills in interpersonal communication, presen-
tation o f materials, “ im pression m anagem ent ” and the like. These are fairly rare
capabilities, and their co m b in ation is rarer still; for this reason labor costs and wages
are high in m ost K IB S. Toivonen (2001) finds that in Finland effectively all o f the
KIBS studied require co m b in ation s o f generic and sector-specific skills. C om m on
requirements were m arketing and sales skills; social skills such as sensitivity to
others, willingness to share know ledge and m otivate others, capacity for self
renewal, IT-related skills, and sector-specific know ledge related to o n e s particular
expertise, to the processes and business m echanism s characteristic o f the K IB S, and
knowledge about the industries and organizations o f clients.
16.4 C o n c l u s i o n s
give services a prominent role in such policies, and their specific requirements may
be overlooked in current programs. (There is a dearth of research concerning how
innovation policies affect services, let alone analysis of service- related policies.) One
exception is Finland, where KIBS are being treated as important actors in the
innovation process. We have seen that KIBS are significant elements of innovation
systems, and as policy adopts a more systemic viewpoint, we can expect their role to
b ecom e m o re o f a focus in policy.
T h e greater p o licy salience o f services in n o va tio n is, in tu rn , lik ely to stim ulate
m o re detailed research into the area. It w ill be im p o rta n t to m o ve b ey o n d d ocu
m en tin g the grow th o f K IB S , to exam in e precisely h o w th ey fu n c tio n as know ledge
sources and interm ediaries: w h at sorts o f exch ange o f k n o w led g e take place in what
w ays an d at w hat steps o f the in n o vatio n process; h o w these are m an ag ed o n service
and client sides o f the equ ation ; w h at skills and cap ab ilities are req u ire d fo r effective
and in n o vative so lu tio n s to be im plem ented . T h ere are still v e ry fe w studies that
address these fu n d am en tal q u estio n s.27
N otes
1. This view has been widely expressed, for instance, with services being classified as
supplier-driven in Pavitt’s original taxonomy o f types o f innovation (1984)— though in
a subsequent paper (1994) he put software services into the “ specialized supplier” group,
and added a category o f “ information intensive” firms which included finance, retailing,
travel and publishing.
2. Hipp et al. (2000) show that a surprisingly large proportion o f German service firms
consider their outputs to be largely standardized. Least standardization was reported by
business services such as technical and computer services.
3. For just one example, consider Jones (1995) on in-flight catering services.
4. By 2000, about 1 per cent o f the UK workforce was employed at telephone call centers.
5. For a series of studies on service internationalization and innovation— still a largely
unexplored topic— see Miozzo and Miles (2003).
6. CL Miles and Matthews (1992), Roach (1988) for early analyses. The data, derived from
input—output tables, allow us to examine investment patterns across different industries.
IT investment from services sectors constitutes a greater share of this investment than
services output constitutes of all output. And for the UK in the 1980s, Miles and Matthews
found that the proportion of sectoral investment devoted to computers and telecommu
nications equipment was 5.7 and 4.6 per cent respectively for services, 4.1 and 4.0 per cent
for manufacturing. (The expenditure on telecommunications equipment among services
at this time was heavily dominated by telecommunication services firms; and in general
the high levels of 11 investment are driven by specific sectors, such as finance.)
7. Not that the problem is not reported by manufacturers— 1 found U K public sector
organizations in the 1980s to be vociferously complaining that their projects were plagued
by departui e o f IT staff to financial services. Beyond the affluent West similar concerns are
I N N O V A T I O N IN S E R V I C E S
453
also raised, for instance Computerworld Singapore (7(3); 20) documented in October
2000 that financial services computer professionals there were consistently paid more
than comparable staff in other sectors.
8. This model, while proving extremely influential and a useful starting point for analysis, is
widely viewed as o f limited applicability even to manufacturing.
9. US researchers have also suggested such developments: see e.g. Faulhaber, Noam, and
Tasley (1986). But see Bronwyn H alls Ch. 17 in this volume, which indicates that
“followers” may spend a long time in that state before becoming innovative in their
own right.
10. Interbranch on-line services; autom ated teller m achines; credit cards and associated
services; remote banking, and electronic funds transfer at point o f sale.
11. Gallouj and Weinstein (1997) provide a useful review, comparing “ servuction” to a
number of other formulations. Though there have been few “ servuction” studies focus
ing on Innovation, the conceptual approach continues to be developed, e.g. by Gadrey
and de Bandt (1994).
12. Note that bar-code scanners require the cooperation of manufacturers in bar-coding
their products. Similarly, financial service innovations like credit and debit cards require
the cooperation o f retailers in accepting these cards and using validation systems.
13. The BBC website at http://www.bbc.co.uk is a case in point. New BBC radio and TV
channels are only available digitally, and remarkable volumes of archived content and
text are also online.
14. Probably the only universal innovations were ones involving the construction and
maintenance o f buildings and technologies o f heating and lighting— and telephones.
Such innovations typically involved relatively little user learning. Even where new skills
were required (e.g. automobile driving skills— which in any case were typically acquired
for everyday purposes) these technologies required relatively little configuration to meet
the needs o f specific users. Indeed, limited scope for configuration was presented, and
the relevant engineering services (e.g. garages) were a matter o f high-street crafts rather
than industrial laboratories.
15. NIST in the USA (TASC 1998) examined how technology-intensive services in several
sectors deal with technology barriers. Some barriers were technology-specific— thus the
high risks o f complex technology led to needs for technical expertise not available to
most individual firms. Collaborative R8cD was undertaken with other firms in their
industry (and often with manufacturing sector partners, too), typically to gain access to
complementary research or technical skills. IT development barriers were overcome by
codevelopment projects with manufacturing suppliers. Such collaborations were not in
general oriented to knowledge resulting from basic research. IT implementation barriers
were associated with high costs o f configuring and employing systems (often IT imple-
mentation cost four to five times more than acquiring the hardware and software). Other
barriers were “market-related” — thus high transaction costs associated with the sys-
temic nature o f new IT, which led to emphasis on standards-related activities to reduce
barriers to IT development and implementation. Standards and protocols were often
central to innovation strategies.
16. See the report by FhG-ISI (2003), which deals especially with services patenting activ
ities. The general argument is that the patent mechanism is not very appropriate to many
service innovations, and thus the firms are not particularly oriented to these types of
454 ia n m il e s
intellectual property. Nevertheless, the study found numerous service firms active in
patenting activity and the development of new IP strategies.
17* Again, there are striking exceptions to this rule some service firms in areas like IT and
consultancy are pioneers in innovation and knowledge management techniques.
18. The sectors studied excluded public and consumer services such as retail and HORECA;
and microbusinesses and very small firms were also excluded.
19. Tether and Swann (2003) present detailed information on UK firms in CIS3 data, where
different types of manufacturing and service are contrasted in some detail.
20. Nearly 70 per cent o f the innovating manufacturers conducted R 5cD.
21. Other evidence supporting this view is presented in Miles 1999a; Tether and Swann 2003.
22. One— but only one— o f the sources o f growth in business services, in particular, is the
outsourcing o f activities previously undertaken in-house by firms and organizations in
other sectors.
23. Some authors even describe them as forming a “ second knowledge infrastructure.” The
traditional primary knowledge infrastructure is mainly a matter o f Higher Education
Institutions (HEIs) and government laboratories/public research and technology
organizations (RTOs).
24. It is difficult to attribute causality to the correlational data involved here, but some very
different types o f study yield broadly similar results. Researchers like Antonelli (2000)
and Tomlinson (2000)— using slightly different input-output datasets, and statistical
methods— show an association between the use o f KIBS as intermediate inputs, and the
performance o f the user sectors. Doubt is cast on the methodology by ECORYS-NEI
(2003), however. Peneder (in European Commission 2000: ch. 4) found that clusters of
industries characterized by high KIBS use performed particularly well. Hansen (1994)
reported that the growth performance o f the economies o f US cities was related to the
size o f the KIBS sectors in these economies. An interesting study by Muller (2001)
examines relations between KIBS and SMEs at regional level, suggesting benefits in
terms o f innovation on both sides.
25. Tether and Swann (2003) show that in the UK, at least, these are the sectors with the
highest graduate-intensity. There are clear differences between those KIBS with high
levels o f science and engineering graduates (e.g. Technical and IT services) and those with
other classes o f University graduate (e.g, consultancy and marketing services).
26. CIS2 data show that acquisition o f external technology through use o f consultancy
services is the second most frequent mechanism used by manufacturing firms. The
most important mechanism is direct equipment purchases.
27. Tomlinson (1999) analyzed U K survey data, finding that KIBS staff are more likely than
are others to learn new things, to receive training, to work with computers, and to move
between different types o f work. Labor mobility is often emphasized as a means for
diffusing knowledge around the economy. Tomlinson argued that KIBS provide an
alternative perhaps a superior— means. (The survey suggests that people moving
between jobs fare poorly on these indicators of “ life long learning.” Whether this result
is peculiar to the UK and/or to the recession underway at the time o f the study and the
downward mobility it induced, requires further study, since the argument is very
provocative.)
28. Cf. ECORYS-NEI (2003) for a review o f literature on business services and their clients.
Several relevant studies are presented in the recent collections by Dankbaar (2003) and
1 1 dd and Hull (2003).
I N N O V A T I O N IN SERVIC ES
455
Re f e r e n c e s
opment Process Analysis in Service Activities, Brussels, EC, FAST. Occasional papers no 116.
Bessant, J., and R u s h , H. (2000), “ Innovation Agents and Technology Transfer” in Boden
and Miles 2000:156-69.
Boden , M., and M i l e s , I. (eds.) (2000), Services and the Knowledge Based Economy London:
Continuum.
Bo lisa n i , E., S c a r s o , E., M i l e s , I., and B o d e n , M. (1999), “ Electronic Commerce Imple
mentation: A Knowledge-Based Analysis,” International Journal of Electronic Commerce
3(3); 53-6 9.
*B ryso n , J. R., and D a n i e l s , R W. (eds.) (1998), Service Industries in the Global Economy
2 vols., Cheltenham: Edward Elgar.
Cox, D., G um m ett , P., and B a r k e r , K. (2001), Government Laboratories— Transition and
Transformation, Amsterdam: IOS Press.
*D a n k b a a r , В. (ed.) (2003), Innovation Management in the Knowledge Economy London:
Imperial College Press.
Davies, A. (2003), “Are Firms Moving ‘Downstream" into High-Value Services,” in Tidd and
Hull 2003:321-41.
D en H e r t o g , P. (2000), “ Knowledge Intensive Business Services as Co-Producers o f Innov
ation,"" International Journal ofJnnovation Management 4(4): 491-528.
-----and B ild er beek , R. (2000), “ The New Knowledge Infrastructure: The Role o f Tech
nology-Based Knowledge-Intensive Business Services in National Innovation Systems,"" in
Boden and Miles 2000: 222-46.
D e p a r t m e n t o f T r a d e a n d I n d u s t r y (2003), The 2003 R&D Scoreboard (data provided by
Company Reporting Ltd), London: DTI available at: http://www.innovation.gov,uk/
projects/rd_scoreboard/introfr.html
ECORYS-NEI and CRIC (2003), Business Services: Contribution to Growth and Productivity in
the European Union, Report to European Commission DG Enterprise, Rotterdam:
ec o r ys .
E u r o p e a n C o m m i s s i o n (2000), European competitiveness report 2000, Brussels: Commis
sion Staff Working Paper EN TR DT 2000/045/A1 Competitiveness, available at: http:// eur-
opa.eu.int/comm/enterprise/enterprise_policy/competitiveness/doc/compet_rep_ 2000/
cr-2000_en.pdf
S au ve , P., and M atto o , A, (eds.) (2003), Domestic Regulation and Service Trade Liberaliza
tion, Oxford: Oxford University Press/World Bank.
Soete, L., and MiozzO) Nt. (1989)> Trade and Development in Services. A Technological
Perspective, Working Paper No. 89-031. Maastricht: MERIT.
S undbo , J. (1998), The Organization of Innovation in Services, Roskilde: Roskilde University
Press.
------ (2000), “ Organization and Innovation Strategy in Services” in Boden and Miles 2000:
109-28.
------ and G a l lo u j , R (2000), '"Innovation as a Loosely Coupled System in Services,” in
S. Metcalfe and I. Miles (eds.), Innovation Systems in the Service Economy, Dordrecht:
Kluwer.
TASC (1998), The Economics o f a Technology-Based Service Sector, National Institute of
Standards & Technology Program Office, Strategic Planning and Economic Analysis
Group, January 1998 NIST Planning Report: 98-102.
T e t h e r , B. S. et al. (2001), “Analysis o f CIS Data on Innovation in the Service Sector: Final
Report,” report to European Commission DG12, CRIC, University o f Manchester (112 pp.)
available at: http:// www.kiet.re.kr/ffies/econo/20021230-inno.pdf
—— H ip p , C , and M il e s , I. (2001), “ Standardization and Particularization in Services:
Evidence from Germany,” Research Policy 30 :115-38.
------ and S w a n n , G. M. P. (2003), “ Services, Innovation and the Science Base: An Investi
gation into the UK's 'System o f Innovation' Using Evidence from the UK's Third Com
munity Innovation Survey,” presented at the International Workshop: Innovation in
Europe: Empirical Studies on Innovation Surveys and Economic Performance, Institute of
Socio-Economic Studies on Innovation and Research Policy, National Research Council,
and University o f Urbino, Faculty o f Economics; Rome, 28 January 2003.
*T id d , J., and H u ll , F. M. (eds.) (2003), Service Innovation: Organizational Responses to
Technological Imperatives and Market Opportunities, London: Imperial College Press.
T o iv o n en , M. (2001), “ Megatrends and Qualification Requirements in the Finnish Know
ledge Intensive Business Service Sector,” in M Toivonen (ed.), Growth and Significance of
Knowledge Based Services, Helsinki: Uusimaa TE Centre Publications 3.
T o m lin so n , M. (2000), “ Information and Technology Flows from the Service Sector:
a U K -Japan Comparison,” in Boden and Miles 2000: 209-221.
T o rdo ir , P, P. (1986), “ The Significance o f Services and Classifications o f Services,” in
P. Coppetiers, J.-C. Delaunay, J. Dyckman, J. Gadrey, F. Moulaert, and P. Tordoir, The
Functions of Services and the Theoretical Approach to National and International Classifica
tions, Lille: Johns Hopkins University Center.
(1996), The Professional Knowledge Economy: The Management and Integration of
Professional Services in Business Organizations, Dordrecht: Kluwer.
U c h u p a la n a n , K. (1998), Dynamics of Competitive Strategy and IT-based Product-Process
Innovation in Financial Services: The Development of Electronic Banking Services in Thai
land, D.Phil. thesis, University o f Sussex, Falmer, Brighton.
(2000), Competition and ІГ -based Innovation in Banking Services,” International
Journal of Innovation Management 4(4): 455-90.
C H A P T E R 17
INNOVATION AND
DIFFUSION
B R O N W Y N H. H A L L
1 7. 1 I n t r o d u c t i o n 1
In 1953, a young female M acaq u e m o n key in the south o f Japan washed a muddy
sweet potato in a stream before eating it. T his obvious im provem ent in food
preparation was im itated q u ickly by other m onkeys and in less than 10 years it
became the n o rm in h er im m ediate group; by 1983, the m ethod had diffused
completely. In 1956, the sam e m onkey innovated again, inventing a technique in
which handfuls o f m ixed sand and w heat grains were cast u pon the sea, so that the
floating cereal could be skim m ed from the surface. A gain, by 1983, this m ethod o f
gleaning wheat had diffused alm ost com pletely throughout the local populations o f
Macaques.2 Besides the o b vio u s fact that hum ankind does not have a m o n o p o ly on
innovation, these exam ples illustrate a couple o f things about the diffusion o f
innovations: first, w hen they are clearly better than what went before, new ideas o f
how to do things w ill u sually spread via a “ learning by observing” process, and
second, the process can take som e tim e; in these cases it took thirty years, and the life
cycle o f the M acaque m o n k ey is som ew hat shorter than ours (Kawai, W atanabe, and
Mori 1992).
Turning to the w o rld o f h um ans, it is safe to say that w ithout diffusion, innovation
would have little social o r econom ic im pact. In the study o f innovation, the w ord
diffusion is co m m on ly used to describe the process b y w hich individuals and firm s in
4бО B R O N W Y N Н. H A L L
in the history o f diffusion o f many innovations, one cannot help being struck by two
characteristics o f the diffusion process: its apparent overall slowness on the one hand,
and the wide variations in the rates o f acceptance o f different inventions, on the other.
(Rosenberg 1972:191)
E m p irical m easu rem en t an d stu d y since then has c o n firm e d this view. This
chapter and the references in clu d ed in it review the d iffu sio n o f a num ber o f
in ven tion s and in n o vative processes, fro m the b o ilin g o f w ater to preven t diarrheal
diseases to m o b ile teleph ony in E u rop e. B o th these stu d ies an d the figu res showing
d iffu sio n rates in v ario u s countries d em o n strate the tru th o f R o sen b erg 's statement.
T h e studies go fu rth er than sim p ly n o tin g the speed an d v a ria tio n o f diffusion, in
that they correlate the rates o f ad o p tio n w ith ch aracteristics o f the technologies and
their poten tial adopters in an attem pt to exp lain the sp eed o f d iffu sio n and the
ultim ate acceptance o f the new p ro d u ct. B esides the w id e v a ria tio n in acceptance o f
in n o vation s, a second im p o rtan t ch aracteristic o f the d iffu sio n p ro cess is the way in
w h ich it interacts w ith the in n o vative process. T h is has p erh ap s been a som ew hat less
studied aspect o f d iffu sio n , o w in g to the d ifficu lty o f co llectin g system atic data, but
case studies aboun d. R osen b erg (1982), a m o n g oth ers, has em p h asized the fact that
the d iffu sio n o f in n o vation s is often accom p an ied b y le arn in g ab o u t their use in
d ifferent en viron m en ts, and that this in tu rn feeds b ack to im p ro vem en ts in the
origin al in n o vation .
W h y is d iffu sio n som etim es slow? W h y is it faster in so m e co u n tries or regions
than others, and fo r som e in n o vation s than fo r others? W h at facto rs exp lain the wide
variatio n in the rate at w h ich it occurs? T h is ch ap ter p ro v id e s a historical and
co m p arative perspective on d iffu sio n that lo ok s at the b ro a d determ in an ts, eco
n o m ic, social, and institu tion al. The w ays in w h ich the d ifferen t so cial scientific
disciplines think about d iffu sio n is discussed and a fra m e w o rk is presented for
stu d yin g its determ inants. Som e o f the em p irical evid en ce o n these determ inants
IN N O V ATIO N AND DIFFUSION 461
is reviewed, and a range o f exam ples given. The chapter concludes with a discussion
of gaps in our understanding and future research questions.
17. 2 C o n c e p t u a l F r a m e w o r k s
The diffusion o f innovations has been studied from a num ber o f different perspec
tives: historical, sociological, econom ic (including business strategy and
marketing), and netw ork theoretical. The choice o f approach is often dictated by
the use to which the results w ill be put, but there is no doubt that insights from one
perspective can in fo rm the research in another discipline. Perhaps a key exam ple o f
this is the w ay in which historical study o f the developm ent and spread o f certain
major inventions has affected h ow econom ists understand the role o f the diffusion
process in determ ining the dynam ics o f p ro d u ctivity change, a topic I return to later
in this chapter. First, I lay out som e o f the fram ew orks that have been used by
different disciplines fo r the analysis o f diffusion.
The sociological and organizational literature is exem plified b y Rogers" well-
known book, Diffusion o f Innovations, now in its fourth edition. In this book, he
reviews the subject p rim arily from a sociological perspective, but one that is
informed by research on organizations, the role o f econom ic factors, and the
strategies o f firm s and developm ent agencies. Rogers provides a useful set o f five
analytic categories that classify the attributes that influence the potential adopters o f
an innovation:
(1) The relative advantage o f the innovation.
(2) Its com patibility, w ith the potential adopter’s current w ay o f doing things and
w ith social norm s.
(3) The co m plexity o f the innovation,
(4) Trialability, the ease w ith w hich the innovation can be tested b y a potential
adopter.
(5) O bservability, the ease w ith w hich the innovation can be evaluated after trial.
M ost o f these attributes are recognizable in one form or another in the m any
analyses o f specific in n ovation s that have been undertaken by researchers in the past,
albeit under different nam es. F o r exam ple, both trialability and observability are
characteristics that speak directly to the level o f uncertainty faced b y a potential
adopter. The latter characteristic is a key feature o f the real options m odel o f
technology choice w hich is discussed later in this chapter and which underlies
some o f the w o rk on tech nology adoption by business firm s. C om plexity as a
determ inant is clearly related to the econ om ist’s notions o f cost and com plem entary
462 B R O N W Y N Н. H A LL
challenge and included such concepts as network effects in their m odels (see the
discussion in Box 17.1). N evertheless, the factors and m echanism s considered in
most o f their studies typ ically fall short o f m any that other disciplines might
consider im portant, such as social connectedness. A n interesting early debate on
this topic that reflected different view s o f the determ inants o f hybrid-corn adoption
in the United States was conducted by a pioneering econom ist in the study o f
diffusion and a n u m ber o f sociologists including Rogers in the pages o f Rural
Sociology (Babcock 1962; G riliches 19 60a and b, 1962; H avens and Rogers 1961;
Rogers and H avens 19 6 2).5 L o o kin g back at this debate from today, a reasonable
conclusion is that both econ om ic and non-econom ic factors probably mattered for
the diffusion o f hybrid corn, although econom ic factors by themselves did a pretty
good job explaining variatio n across states.
As an exam ple o f m icro eco n o m ic analysis o f the adoption decision in a m odern
technological setting, consider the decision to replace a w ired physical connection to
the Internet with a w ireless one, either at hom e or in an office. Benefits m ight include
the ability to w o rk on the n etw ork th rou gh ou t o n e s house or w orkplace rather than
at a fixed lo catio n such as a desk, and the absence o f w ires. T h e y m ig h t also include
the fact that several m em b ers o f the h o u seh o ld can be o n lin e at the sam e tim e using a
single telephone connection. T h e costs in clu de the pu rch ase o f a base statio n and the
services o f a tech nician to install it, but th ey m a y also in clu d e the tim e o f the user
(adopter) spent reco n fig u rin g his o r her co m p u ter an d en su rin g th at all the com
m u n icatio n tools needed are w o rk in g . C o sts m igh t also in clu d e the acq u isition o f
n ew softw are, o r the tim e spent train in g oth er m em b ers o f the h o u seh o ld o r office in
its use. W ere we to enrich this sto ry to in clu d e the a d o p tio n e n viro n m en t, w e m ight
focu s o n such factors as w h eth er n eigh b o rs o r colleagues alread y h ad undertaken
such an installation, the extent to w h ich it has been ad vertised b y the su p p lier o f the
tech n o lo gy (o r the extent to w h ich it has been “ sp o n so re d ” b y a g o vern m en t agency
o r lead in g firm ), and even the state o f d evelo p m en t o f the n ew tech n o lo g y and the
o p eratin g system necessary to use it (a co m p le m e n tary in p u t). N o te also that m ost o f
these factors have been ch an gin g rap id ly over tim e.
A s alluded to earlier, the first em p irical stu d y o f the d iffu sio n o f tech n o lo g y b y an
econ o m ist w as G rilich es’ (1957) stu d y o f the d iffu sio n o f h yb rid c o m seed in the
M id w estern U n ited States. T h is stu d y em phasized the role o f e c o n o m ic factors such
as expected pro fits and scale in d eterm in in g the v a ry in g rates o f d iffu sio n across the
M id w estern states. A t the sam e tim e, it fo u n d that the v a ria tio n in in itial start dates
fo r the process depended on the speed w ith w h ich the seed w as cu sto m ized fo r use in
p articu lar geograph ic areas. T h at is, d iffu sio n dep en d ed to a certain extent on the
activities o f the su p p liers o f the tech n o lo gy in ad ap tin g it to lo cal co n d itio n s, again
h igh ligh tin g the ten d en cy fo r the fu n d am en tal ch aracteristics o f the tech n ology to
change som ew h at d u rin g the a d o p tio n process. T h is th em e is repeated throughout
the h isto ry o f in n o vatio n . B ru lan d (1998, 2002) fin d s that the n in eteen th -cen tury
d evelo p m en t o f the N o rw eg ian textile in d u stry w as g reatly facilitated b y the
tech n o lo g y tran sfer activities u nd ertaken b y the m o stly B ritish m a c h in e ry suppliers
in the fo rm o f train in g, in creasin g the su p p ly o f skilled w o rk ers in N orw ay.
T h e m arketin g literature on d iffu sio n is p rim a rily fo cu sed o n tw o questions: how
to encou rage co n su m ers and cu stom ers to pu rch ase n ew p ro d u c ts o r technologies,
and h o w to detect o r forecast success in the m arketplace. T h at is, it o ften looks for
factors that can be in flu en ced in o rd er to increase the n u m b e r o f agents that will
ch oo se a p articu lar p ro d u ct. For this reason, the literatu re tends to em ph asize factors
such as m ed ia in fo rm a tio n o r the role o f social n etw orks and ch an ge agents, as well
as the characteristics o f the p ro d u ct itself, rather than in d iv id u a l ad o p ter factors
such as ed u catio n and incom e levels that are less subject to m a n ip u la tio n b y the
m arketin g organ ization . The w o rk h o rse m odel in m ark etin g fo r m a n y years has
been the Bass (1969) m o del, w hich assum es that m ass m ed ia are im p o rta n t early on
in the d iffu sio n process but that as tim e passes, in terp erso n al co m m u n icatio n
becom es far m o re im p o rtan t. E stim atio n o f this m o d el on a n u m b e r o f consum er
d urables has revealed that interperson al c o m m u n ica tio n p lays a m u ch bigger role
than the m edia in d iffu sio n (R ogers 1995). For an in terestin g d iscu ssio n o f the
contrast between the econ om ic and m arketing view s and a com parison o f models
from the two literatures, see Zettelm eyer and Stonem an (1993). Recent work on
identifying and forecasting success in the m arketing literature is illustrated by
Golder and Tellis (1997)- I defer discussion o f their m odel to later in this chapter
when I discuss som e o f the findings obtained b y Tellis, Strem ersch, and Yin (2002)
using this m ethodology.
The activist view o f d iffu sion taken by the m arketing literature is also that pursued
by specialists in tech nology policy, w ho are generally interested in encouraging the
adoption o f particular new technologies for w elfare-enhancing reasons, either
because it serves particu lar pu blic po licy goals (such as encouraging the boiling o f
water to reduce disease in less-developed countries) or because certain technologies
are viewed as conferring externalities on society as a w hole (such as the adoption o f
Internet use or vaccin ation against a com m unicable disease). In understanding the
variation across countries in d iffu sion , variables describing their institutions and
culture have proved essential in som e cases (but not all, see the discussion o f Tellis,
Stremersch, and Y in 2002 in Section 17.5).
17.3 M o d e l i n g D i f f u s i o n
The most im portant th in g to observe abou t the decision to adopt a new invention is
that at any point in tim e the choice being m ade is not a choice between adopting and
not adopting but a choice betw een ad optin g now or deferring the decision until later.
It is im portant to lo o k at the decision in this w ay because o f the nature o f the benefits
and costs. By and large, the benefits fro m adopting a new technology, as in the
wireless com m unications exam ple given above, are flow benefits that are received
throughout the life o f the acquired innovation. H owever, the costs, especially those
of the n on-pecuniary ‘T eam in g” type, are typically incurred at the tim e o f adoption
and cannot be recovered. There m a y be an on goin g fee for using som e types o f new
technology, but it is u sually m uch less than the initial cost. Econom ists call costs o f
this type “ sunk.” T h at is, ex ante> a potential adopter weighs the fixed costs o f
adoption against the benefits he expects, but ex post, these fixed costs are irrelevant
because a great p art o f them have been su n k and cannot be recovered.
The argum ent that ad o p tion is characterized b y sunk costs im plies two stylized
facts about the adoption o f new technologies: first, adoption is usually an absorbing
state, in the sense that w e rarely observe a new technology being abandoned in favor
of an old one.6 T h is is because the decision to adopt faces a large benefit m inus cost
hurdle; once this hu rd le is passed, the costs are sunk and the decision to abandon
466 B R O N W Y N Н. H A LL
requires givin g up the b en efit w ith o u t regain in g the cost, so even i f the gross benefit
is reduced relative to w hat w as expected, the net ben efit is still lik ely to be positive.
S eco n d , u n d er u n certain ty ab ou t the benefits o f the n ew tech n o logy, there is an
o p tio n value to w aitin g b efore sin k in g the costs o f a d o p tio n , w h ich m a y tend to delay
a d o p tio n .7
A n im p o rtan t exception to the rule that a d o p tio n is n o rm a lly an ab so rb in g state is
the p o ssib ility o f fads o r fashion s, w h ich m igh t be d efin ed as th in gs su ch as the “ hula
h o o p ” craze o r vario u s types o f w eigh t-loss diets, w h ich d iffu se ra p id ly and then
d isap p ear after a tim e. The experien ce o f a w ave o f a d o p tio n fo llo w ed b y a wave o f
disuse seem s to be som ew h at m o re likely in the case o f in n o v a tio n s in “ practice,”
such as m edical practice or business practice, than in the case o f p h ysical products,
p o ssib ly because in the latter case the costs that are su n k are o u t o f p o ck et costs paid
to others, w hereas in the fo rm er m u ch o f the cost (alth o u g h b y n o m ean s all) comes
in the fo rm o f the ad o pter's tim e an d effort. T h at is, the p o ssib ility o f sun k costs
m a y lo o m larger to the ad o p ter w hen d en o m in ated in d o lla r o r eu ro sym bols.
N elson et al. (2002) discuss this p h en o m en o n m o re fu lly an d g ive so m e exam ples
(such as the q u ality circle m o vem en t). T hese au th o rs place co n sid erab le em phasis
o n the d ifficu lty in these cases o f getting feedback that the in n o va tio n truly is
an im p ro vem en t. R elatively lo w su n k costs co m b in ed w ith u n certain benefits
w ill m ean that the d ecision to ad o p t is m o re easily reversible in the case o f practices.
Strang and Sou le (1998) also discuss the cyclicality o f fash io n s in business
practices.
It is a w ell-k n o w n fact that w h en the n u m b er o f users o f a n ew prod u ct or
in ven tion is plotted versus tim e, the resu lting cu rve is ty p ic a lly an S-shaped or
ogive d istrib u tio n . T h e n o t v e ry su rp risin g im p lica tio n is that a d o p tio n proceeds
slo w ly at first, accelerates as it spreads th ro u g h o u t the p o ten tial ad o p tin g p o p u la
tio n , and then slow s d o w n as the relevant p o p u latio n b eco m es satu rated. In fact, the
S-sh ap e is a natu ral im p licatio n o f the o b servatio n that a d o p tio n is u su ally an
ab so rb in g state. Figu re 17 .1, w h ich represents the d iffu sio n o f electric m otors in
U S m an u factu rin g betw een 1898 and 1955, show s su ch a cu rve. In 1898, the share o f
m an u factu rin g h o rsep o w er p ro d u ced b y electric m o to rs w as ab o u t 4 per cent. It
increased stead ily and sm o o th ly betw een 1900 and ab ou t 1940, at w h ich p o in t nearly
all h o rsep o w er is p ro d u ced b y electricity. S atu ratio n ap p ears to be reach ed at around
90 per cent, p resu m ab ly because fo r som e specialized uses, o th er typ es o f m otors are
preferred.
L o o ked at in term s o f the benefits and costs o f tech n o lo g y a d o p tio n , a range o f
sim p le assu m p tio n s w ill generate this curve. T h e tw o lead in g m o d els explain the
d isp ersion in a d o p tio n tim es using tw o d ifferent m ech an ism s: c o n su m e r heterogen
eity, o r co n su m er learning. The heterogen eity m o d el assu m es that d ifferent con
su m ers expect to receive d ifferent benefits fro m the in n o va tio n . I f the d istrib u tion o f
benefits over co n su m ers is n o rm al (or a p p ro x im a te ly n o rm a l, that is unim odal
with a central tendency), the cost o f the new p ro d u ct is co n stan t o r declines
INNOVATION a n d d iffu sio n 467
Year
monotonically over tim e, and it is assum ed that consum ers adopt when the benefit
they receive for the prod u ct is greater than its cost, the diffusion curve for the
product will have the fam iliar S-shape.
An im portant alternative m odel is a learning o r epidem ic m odel, which is m ore
popular in the sociological and m arketing literatures (the Bass m odel is an exem
plar), but has also been used b y econom ists. In this m odel, consum ers can have
identical tastes and the cost o f the new technology can be constant over tim e, but not
all consumers are in form ed ab ou t the new technology at the sam e time. Because each
consumer learns about the tech n o logy fro m his or her neighbor, as tim e passes, m ore
and more people adopt the tech n ology du ring any period, leading to an increasing
rate o f adoption. H ow ever, eventually the m arket becom es saturated, and the rate
decreases again. This too w ill generate an S-shaped curve for the diffusion rate.8 In
general, com bining this m odel w ith the previous m odel sim ply reinforces the
S~shape o f the curve. C o ld e r and Tellis (1997) define a concept they call “ take-off,”
which is their attem pt to id en tify the p oin t at which the em pirical diffusion curve
appears to have its greatest inflection relative to the initial growth rate.9 For the data
in Figure 17.1, this poin t w o u ld be in about 1910. Because for m any consum er
products the existence o f such a p o in t is a good predictor o f eventual success, the
focus o f their w ork is to id en tify predictors o f this point.
Regardless o f the details o f the m echanism generating the probability distribution
of adoption tim es, the question w hich concerns both social scientists and those
interested in encouraging the spread o f new technologies is the question o f what
factors affect the rates at w h ich these events occur. A second and no less interesting
question is what are the determ inants o f the ceiling at which the S-curve asym ptotes.
That is, when w ou ld we expect this ceiling to be less than 100 per cent o f the potential
# 8 BRONWYN H. HALL
user base? T h e next section o f this ch ap ter review s these facto rs an d som e o f the
em p irical evidence co n cern in g their im p o rtan ce.
17.4 D e t e r m i n a n t s of t h e
D iffusion Rate
Year
From the considerations reviewed earlier in the chapter, one can derive a list o f
factors that m ight be expected to influence the d iffu sion o f innovations. These can be
classified into four main groups, those that affect the benefits received, those that
affect the costs o f adoption, those related to the industry or social environment, and
those due to uncertainty an d information problems. Alternatively, using the clas
sification system o f Rogers, one can identify the first and second as combining to
yield relative advantage an d com plexity, the third as com patibility, and the fourth as
being determined by trialability and observability.
For exam ple, Saloner and Shepard (1995) exam ine the adoption o f ATM machines
by banks, under the assum ptions that consum ers prefer a larger network o f ATM
machines to a sm aller and that banks respond to consum er preference. These
authors do indeed find that banks with m ore branches adopt an ATM network
sooner, even after con trollin g for overall bank size, and argue that this confirm s that
a higher netw ork valu e leads to earlier adoption o f a new technology, other things
equal.10 This exam ple illustrates both the im portance o f networks and also the role
o f large firm s as interm ediates between technology and consum ers in sponsoring
particular standards for networks.
A fam ous exam ple o f the role o f "n etw o rk externalities” in consum er adoption o f
new technologies is the V H S/B eta com petition, which resulted eventually in a single
standard for video recorder/players in a large part o f the w orld. M ost observers
attribute this outcom e to the consum er desire for a large range o f software in the
form o f pre-recorded tapes to go w ith this hardware, and to the fact that V H S had an
initial early advantage in the length o f program that could be recorded. See Park
(2002) for details on the diffusion o f this technology to consum ers.
Although netw ork effects (particularly those from networks that diffuse know
ledge about o r experience w ith an innovation) have always been view ed as im portant
for the diffusion o f innovations, especially in the sociological literature, recent w ork
in econom ics has focused on the role played by standards in accelerating or slowing
the diffusion process, as in the V H S/B eta exam ple (D avid 1985; Katz and Shapiro
1985; A rthur 1989; E con om ides and H im m elberg 1995). The central message o f the
modern econom ic literature on standards and netw ork externalities is that con
sumers and firm s receive benefits from the fact that other consum ers and/or
firms have chosen the sam e tech nology that they have. These benefits are viewed
as being o f tw o kinds, direct and indirect. D irect netw ork benefits are those that
arise because they allow the adopter to com m unicate with others using the same
technology. Exam ples are the choice o f fax com m unication technology or the choice
o f w ord processor docum ent form at. Indirect benefits arise from the fact that
adoption o f a p ro d u ct that uses a particular technological standard by a greater
num ber o f people increases the p rob ab ility that the standard will survive and that
goods com patible w ith that standard w ill continue to be produced. The V H S/Beta
example alluded to earlier can be view ed as an exam ple where indirect network
benefits w ere very im po rtan t, although direct benefits presum ably also play a part
(the benefits from being able to loan a video m ade on one's ow n m achine to a friend
or neighbor).
The close connection between technological standards and network externalities
comes from the fact that standards create a number o f effects all o f which go in the
direction o f making it more likely that a good will exhibit network externalities.
First, a technological standard increases the probability that communication
between two products such as telephones, instant messaging services, or a CD player
and a CD, will be successful Second, standards ease consumer learning and
472 B R O N W Y N H, H A L L
encourage adoption when the same or sim ilar standards are used in a range o f
products. The use o f a particular standard, such as a W indows operating system, by
others in a consumer's network, also helps learning and will encourage adoption,
because o f the relative ease with which a new adopter can obtain advice from those
nearby. Third, a successful standard increases the size o f the potential market for a
good, which can be im portant in lowering the cost o f its production and in
increasing the variety and availability o f com plem entary goods. Besides the VH S/
Beta example referred to earlier, an example o f this latter effect m ight be the wider
availability o f software for the W indows operating system, in com parison to M acin
tosh OS or Linux.
A lth o u g h standards have alw ays m attered fo r d iffu sio n , the in creasin g im p o rt
ance o f d igital an d in fo rm a tio n tech nologies have in creased th eir salien ce and led to
a v ariety o f “ standards b attles" and to strategic b e h a v io r o n the p art o f firm s that
h op e to in fluen ce their ad o p tion . E arlier exam p les o f stan d ard s battles are the
co m p etitio n betw een A C and D C m ethods o f d istrib u tin g electricity (D avid
19 9 0 a), an d the failu re o f gas-p o w ered refrigerators to succeed in the m arket despite
their app aren t efficiency, because o f the sp o n so rsh ip o f electric p o w e r b y G E and
W estinghouse (R ogers 1995). N evertheless, it is clear that the im p o rta n ce o f this
p h en o m en o n has increased recently, w ith increase in in fo rm a tio n an d co m m u n i
cation tech nologies. C o n sid er fo r exam p le, the battle b etw een N etscap e an d M icro
soft Internet E x p lo rer fo r d o m in an ce in the w eb b ro w ser m arket.
T h e increase in the im p o rtan ce o f standards that has acco m p a n ied the grow th in
im p o rta n ce o f the in fo rm atio n and teleco m m u n icatio n s in d u stries has led to a wave
o f eco n o m ic m o d elin g. T hese m odels in co rp o rate the in creasin g retu rn s p h en o m
enon that results fro m the p o sitive feedback fro m installed base to a d o p tio n b y other
consum ers. A n early effort is that b y A rth u r, E rm o liev, an d K a n io v sk i (1983), which
em phasizes D a v id 's insigh t that w h ere there are m u ltip le p o ssib le stand ard s, small
events early in the process that fav o r one o f the stan d ard s can lead to an adoption
process that settles on an in ferio r standard. B y ad d in g h etero g en eity in consum ers'
tastes o r lo calization in in fo rm a tio n spillovers, later researchers h ave p ro d u ced m ore
co m p lex m o d els o f d iffu sio n in the presence o f n etw o rk extern alities that results in
m o re than on e stand ard su rv iv in g in the m arket even in the presen ce o f increasing
returns in ad o p tio n (B assan in i and D osi 1998; W endt an d v a n W estarp 2000).
In d u strial o rgan izatio n and strategy theorists have centered th eir m o d e lin g efforts
on the im p licatio n s o f increasing returns in a d o p tio n fo r co m p etitive strategy and
m arket structure. E xam ples o f this literature include K atz an d S h ap iro (19 8 5,198 6 ,
l 994 )y Farrell and Salo n er (1992), and Sh ap iro an d V arian (1999). In a series o f papers,
Katz and S h ap iro have explored the im p licatio n s o f co n su m er a d o p tio n b eh avio r in
the presence o f n etw ork externalities for the strategic in teractio n s a m o n g firms
o fferin g co m p etin g produ cts. In general, the theoretical literatu re o f w h ich these
pap ers are an exam p le identifies m u ltip le possible eq u ilib ria a m o n g firm s com peting
in such en viron m en ts, so that it is d ifficu lt to draw firm co n clu sion s.
I N N O V A T I O N AND D IFFU S IO N
Farrell and Saloner study the speed o f diffusion (relative to the socially optimal rate)
when the good in question is subject to network externalities, so that early adopters
ignore the consequence o f their adoption on future adopters and on the users of the
previous technology. They show that in this setting, diffusion can be either too fast
(excess momentum) or too slow (excess inertia). Finally, the book by Shapiro and
Varian draws out the implications o f these various theoretical models for the produc
tion and marketing o f information goods (broadly defined), many of which exhibit
the properties that give rise to network externalities. They describe strategies for
competing in markets where network externalities are important and where it is
important to win standards battles because losing them means business failure.
U n ited States (also see Figu re 17 .1 and M o w e ry an d R o sen b erg 1998)* T h e installation
o f electric p o w er in a facto ry requ ired a com plete red esign o f its la y o u t an d a change
in task allo catio n , w h ich m eant that ad o p tin g this n ew te c h n o lo g y w as a rath er costly
process, an d tended to o ccu r slow ly, o r w hen green field in vestm en t w as being
undertaken* D a v id argu es that a sim ilar reo rgan izatio n o f w o rk flo w takes place
w h en co m p u ter tech n o lo g y is in tro d u ced into the w o rk p lace o r w h en Internet-
b ased pro cessin g replaces telephone o r m ail o rd er p ro cessin g. R ecent p ro d u ctivity
grow th evidence in the U n ited States app ears to co n firm the v ie w that m ajor
tech n o lo gical-o rg an izatio n al change takes tim e fo r its effects to be felt (G o rd o n
2003; Econom ist 2003).
Shaw (2002) has d o cu m en ted this k in d o f p h e n o m en o n in the replacem ent o f
m an u al m o n ito rin g o f p ro d u ctio n lines in co n tin u o u s h ot steel p ro d u c tio n lines by
co m p u terized p u lp it o p eratio n . N o t o n ly does this in vo lve a su b stan tial investm ent
in h igh tech n o lo g y equ ip m en t, b u t it also requ ires few er w o rk ers w ith substantially
h igh er cogn itive skills. W here th ey used to be on the p ro d u c tio n lin e w orkin g
ph ysically w ith the m ach in ery, they are n o w in sm all ro o m s (“ p u lp its” ) above the
line, m o n ito rin g and ad ju stin g the process u sin g co m p u te r tech nology.
T ech n o logy p ro d u cers o ften try to subsidize the a d o p tio n o f n ew tech n ologies by
p ro v id in g free train in g and o th er help to (poten tial) users an d b y ch arg in g reduced
in tro d u c to ry rates fo r a certain p erio d . A n o th er sy m p to m o f the desire o f innovating
firm s to reim b u rse new cu stom ers for their su n k costs in p re v io u s tech n o logies is the
w id esp read practice am o n g softw are firm s o f o fferin g c o m p e titive u pgrades to
ow n ers o f rival p ro d u cts as well as to the ow n ers o f th eir o w n p ro d u cts. For a
m o re com plete d iscu ssion o f strategies used b y tech n o lo g y p ro d u ce rs to encourage
d iffu sio n and increase the installed base o f their p ro d u ct, see S h a p iro an d Varian
(1999).
Because m ost o f the costs o f a d o p tio n are fixed, firm s' ch oices to change or
in tro d u ce tech nologies m a y be in flu en ced by their o w n scale an d b y the m arket
structu re o f the in d u stry w ith in w h ich they operate. A n in terestin g exam p le o f this
p h en o m en o n is given b y Paul D av id in a series o f pap ers on the in tro d u c tio n o f the
m ech an ical reaper in U S and B ritish agricu ltu re in the n in eteen th cen tu ry (D avid
1975a and b). H e argues p ersu asively that a d o p tio n w as delayed in B rita in relative to
the U nited States fo r tw o reasons: first, because the reaper w as a fixed cost invest
m ent, profi tab ility required a farm and fields o f a certain size; secon d, because it was
i n co m p atib le w ith the typ ically B ritish pattern o f sm all fields d iv id ed b y hedgerow s.
In ad d itio n to the difference betw een cou ntries, he also fin d s that d iffu sio n was
delayed in the U S itself until the price o f lab o r rose to a level that m ad e the invest
m ent in the reaper (a lab o r-savin g device) profitab le.
In the p resen t-d ay context, a sim ilar em p irical fin d in g can be fo u n d in m any
studies o f d iffu sio n , M a ju m d a r and V enkataram an (1998) lo o k ed at the replacem ent
o f m echan ical sw itch in g by electronic sw itch in g in the U S telecom m u n icatio n s
industry and fo u n d that larger firm s ad o p ted first, p re su m a b ly b ecau se the costs
I N N O V A T I O N AND D I F F US I ON 475
per customer were somewhat lower. Note that even when technology adoption
involves an investment in equipment that is proportional to the existing size o f the
firm, the requirement that the firm have sufficient absorptive capacity, and the need
for worker training or other complementary changes may create a fixed cost that is
not proportional to firm size.
As in the case o f investment in innovation, firm investment in new technologies is
also sensitive to financial factors. As was suggested earlier, the decision to adopt new
technology is fundamentally an investment decision made in an uncertain environ
ment, and therefore we should not be surprised to find that all the arguments for a
relationship between sources o f finance and choice o f investment strategy that have
been advanced in the investment literature have a role to play here. Chapter 9 by
O'Sullivan in this volume reviews these financial factors in some detail. For example,
Mansfield (1968) reports that the adoption o f diesel locomotives by railways depends
somewhat on their liquidity, implying these firms faced a higher cost o f external than
internal finance.
(2002). She looks at the decision by US plants to adopt three advanced manufacturing
technologies, and finds that plants operating in industries with lower degrees o f
demand and technological uncertainty and a thicker resale market (higher resale
prices for used machinery) are more likely to adopt these technologies. She argues
that this confirms the importance o f uncertainty in the decision; if adopting a new
technology corresponds to the exercise o f an option, we expect adoption to happen
more often in industries with lower uncertainty and lower sunk costs.
Market structure can affect tlie decision to adopt in two distinct way si via seller
behavior and via buyer behavior. Highly concentrated providers of new technology
will tend to have higher prices, slowing adoption, but they also have the ability to
determine a standard more easily, increasing the benefit o f adoption. If two or more
oligopolistic firms are competing to offer different standards, we may in fact get too
rapid adoption o f a new technology, because o f the incentives they face to price below
cost in order to build market share (Farrell and Saloner 1992). In the case o f potential
adopting firms, market concentration affects both their ability to pass through any
costs to consumers and also the incentives they face in incurring the costs o f adoption.
Many of the issues raised by the tension between the fear of displacement and the
exercise of market power here are familiar from the literature on monopolists'
incentives to innovate (for example, see Gilbert and Newberry 1982).
Along with market size and structure, the general regulatory environment will have
an influence, tending to slow the rate o f adoption in some areas due to the relative
sluggishness o f regulatory change and increasing it in others due to the role of the
regulator in mandating a particular technological standard. As an example o f
the former situation consider the use o f plastic pipe for plumbing, which lowers
construction cost, but has been slow to diffuse in many localities due to existing
building codes. As an example o f the latter, Mowery and Rosenberg (1982) have
written about the extent to which airline regulation by the Civil Aeronautics Board
in the United States was responsible for promoting the adoption o f new innovation in
airframes and jet engines, in its role as standard setter and coordinator for the industry.
An important example o f the unintended consequences o f regulation for diffu
sion is the difference between the United States and Europe (and Japan to some
extent) in the diffusion o f household Internet use. Historically, pricing in the US
telecommunications industry has permitted unlimited local calling at a single
monthly rate, whereas pricing for local calls in other countries has usually been
proportional to usage. These policies are largely determined by regulatory bodies,
but once in place, are difficult to change because consumers and firms adapt to them.
In the absence o f direct connection to the network such as is available in large
institutions, household Internet use requires the ability to connect over local phone
lines for extended periods o f time. The marginal cost o f the Internet for households
is therefore to a great extent determined by the cost o f local calling, so diffusion o f the
Internet along with email and instant messaging use has been far more rapid in the
United States than in other countries that are just as developed. Only with the recent
advent o f ISDN service charged by the month in some European countries has
household Internet use begun to spread there. In contrast, the diffusion o f various
“ text-messaging” services on wireless phones, which are a form o f communication
popular with teenagers and similar to the Internet instant messaging widely used in
the US, has been more rapid in Japan and Europe. Relative costs o f the two forms o f
instant com m un ication , w hich in turn are due to regulatory reasons, are probably
the m ain explanation fo r the differences.
478 B R O N W Y N H. H A LL
17.5 C u l t u r a l a n d S o c i a l D e t e r m i n a n t s
Econom ic factors like these can go a long way toward explaining differences in rates
o f diffusion (Griliches 1957 and subsequent authors) but other factors m ay also be
important; For example, m any have stressed differences in cultural attitudes towards
risk and simple ‘"newness.” 12 These characteristics can vary within cultures as well as
between them, leading to dispersion in adoption rates that are not accounted for by
the economic variables. Am ong others, Strang and Soule (1998) provide a useful
discussion o f the cultural basis o f diffusion.
Rogers (1995) cites a number o f situations where compatibility with existing social
norms has strongly influenced the adoption o f health-related innovations such as
the boiling o f water for consumption or various types o f contraceptives in under
developed countries, whose relative popularity depends greatly on local religious
and cultural mores. He cites as example an instance where a strong traditional
distinction made between the qualities of cold and hot water discouraged the use
of the very simple preventive measure of boiling water destined for human con
sumption in order to prevent diarrheal diseases.
On the other hand, for consumer household durables, Tellis et al. (2002) find that
variables such as gender, cultural attitudes, religion, etc. have little predictive power
for "‘take-off’ on average (across European countries) in the presence o f lagged
market penetration. When these variables are considered separately as predictors,
""industriousness” (which is measured by a climate variable) and ""need for achieve
m ent” ( which is measured by the ratio o f Protestants to Catholics in the country)
speeds diffusion, and a measure o f ""uncertainty avoidance” slows diffusion. This
study is noteworthy in that it includes economic, cultural, and communication
variables jointly in the same predictive equation.
17.6 C o n c l u s i o n s
Traditionally, diffusion is one o f the three pillars on which the successful introduc
tion o f new products, processes, and practices into society rests, along with inven
tion (a new idea) and commercialization/innovation (reducing the invention to
practice). In some ways it is the easiest part o f the process to study, because it is
m ore predictable from observable factors than the other two. Certainly countless
studies o f the diffusion o f individual innovations exist, and even exhibit some
commonalities (see the references in this chapter and in Rogers 1995), such as
INN OV ATION AND DIFFUSION 479
the familiar s-shaped curve, and the importance o f both economic factors and social
networks,
Although many have criticized the linear model that lies behind the division of
innovative activity into three parts as oversimplified, it remains true that without
invention it would be difficult to have any thing to diffuse, so that the model still
serves us as an organizing principle, even if we need to be aware of its limitations.
Nevertheless, an important insight from the many historical case studies of individ
ual inventions has been the extent to which the diffusion process enhances an
innovation via the feedback of information about its operation or utility under
varying conditions and across different users, information that can be used to
improve it. A second major finding from this literature has been the possible
feedback from differences in the rate or scale of adoption across geographic areas
to the rates of improvement in the innovation.
In the introduction to this chapter, Rosenberg’s observations on the slowness and
variability o f the diffusion o f different innovations were cited. The studies reviewed
in this chapter have identified some explanations for these observations, such as the
size of sunk costs (trialability), the adaptations and improvement necessary to make
the invention useful after its initial conception, and the inherent slowness o f
interpersonal communication networks in spreading information. In the case of
major innovations such as electricity or the computer, some studies have empha
sized that the necessi ty o f reorganizing the workplace to take advantage of the new
innovation means that diffusion will be greatly delayed, and also that the expected
gains from innovation may take time to be realized.
Several areas stand out as potentially fruitful for future research. First, most of the
studies conducted to date have been methodologically rather simple; the most
ambitious have used a hazard model to correlate the time until adoption with
various characteristics o f the innovation and the adopter (depending on the par
ticular dataset). There is room for an approach that is more structural and grounded
in the choice problem actually faced by the adopter. One promising avenue for
modeling is the real options approach suggested by Stoneman (2001b); such a model
would yield a hazard or waiting time model rather naturally, while explicitly
incorporating the effects o f uncertainty on the decision.13 The cumulative distribu
tion for adoption derived from a hazard model has the familiar S-shape.
Second, although many studies have described the process o f innovation
enhancement during its diffusion qualitatively, there has been relatively little
systematic collection o f data or modeling o f the process. Investigations of this type
would be very helpful in quantifying the importance o f this effect, which is similar to
but not the same as the well-known learning curve. One technological area where
this process has been very important and might be worth study is the area o f user-
driven software developm ent. '
Finally, an area o f research that is receiving increasing attention in a globalizing
economy is that concerned with international technology transfer.14 This literature
4$0 B R O N W Y N H . HALL
N o tes
1. University of California at Berkeley, Scuola Sant’anna Superiore Pisa, NBER, and the
Institute of Fiscal Studies, London. I am grateful to Beethika Khan for contributing some
of the literature review that lies behind the issues discussed in this paper, and other
contributors to this volume, especially my discussants, Kristine Bruland, John Cantwell,
and Ove Granstrand, for their very helpful comments. Finally I owe an immense debt to
the editors for their careful reading of multiple drafts of this chapter,
2. I am grateful to Chris E. Hall for calling this example to my attention. It is described in
McGrew (1998), where a more complete set of references to the anthropological literature
is given. A third feature of this example, perhaps not directly relevant to this chapter, may
be noted: the fact that once having innovated, innovators tend to innovate again.
3. As discussed in the introduction to this volume, the view that every adopter develops and
adapts an invention to his own use has led so me of the literature to refer to adoption itself
as “ innovation.” I will follow the more conventional practice of reserving the term
innovation for the first “public” use of a new product, process, or practice.
4. See Godinho and Fagerberg (Ch. 19 in this volume) on the role of adoption of new
technology in the catch-up process and in long run economic growth.
5. I am grateful to Paul David for calling some of these references to my attention.
6. Although see Rogers (1995) for some examples of innovations that failed to diffuse because
they were rejected after trial.
7. An option is a choice between doing nothing and paying a fixed amount to purchase an
uncertain return. It is real (as opposed to financial) if it involves investment in real assets.
In this setting, the investment is the adoption of a new technology, which has uncertain
benefits and costs that may change over time. The option value arises from the fact that
waiting may reduce the chance that the wrong decision is made.
8. For a good presentation of this class of models and their extensions, see Geroski (2000).
David (2003) provides an evolutionary interpretation of this mechanism.
9. For any particular parametric distribution function, this point might be defined at the
point where the curvature of the cumulative distribution (the second derivative) is
INNOVATION AND DIFFUSION 4&l
maximized. Such з point is well defined if it exists. It occurs when shout 20 per cent o f the
population has adopted in the case o f a logit and when about 15 per cent have adopted in
the case o f a norm al. C o ld er and Tellis (1997) define a non~parametric discrete version o f
this measure by looking at the current rate o f adoption as a share o f adoption to date.
10. On the adoption o f A TM systems, see also Hannan and M cDowell (1984a and b), who
emphasize the role o f bank size and industry concentration, which are chiefly cost side
and market structure considerations.
11. See Brynjolfsson (2000) for a su m m ary o f this work and further references.
12. For a discussion o f various cultural explanations, see M okyr (1990).
13. In the labor econom ics literature, Lancaster and Nickell (1980) developed a sim ilar
model for the probability o f obtaining a job when unem ployed (see also Lancaster 1990).
14. See Keller 2001 for a review o f this literature.
R e f e r e n c e s
ЗЗ2-7*
482 B R O N W Y N H. H A L L
INNOVATION AND
PERFORMANCE
In t r o d u c t i o n to P a r t IV
I N N OV A T I ON
AND ECONOMIC
GR OWT H
BART VERSPAGEN
18.1 I n t r o d u c t i o n
According to Maddison (2001), the world economy began to grow in roughly the
year ad 1000 following a long period o f stagnation (Figure 18.1).1 A marked increase
in growth rates occurred in the period 1600—1800, and, as the figure suggests, growth
has been increasing ever since.
Econom ic h isto ry addresses the issue o f the w ay in w hich the record o f econom ic
growth in Figure 18.1 is related to historical developm ents such as the Industrial
Revolution (see also B ru lan d and M ow ery, Ch. 13 in this volum e). A lthough the exact
impact that this has had on growth rates, particularly at the sectoral level, remains a
subject o f debate (e.g. Crafts 1985), it seems beyond dispute that a change ot
technology in the pure sense, coupled with organizational changes at various eve s
of aggregation, are the main driving factors behind the continuous increase о lving
standards entailed b y this process.
The historical perspective also nicely illustrates the fact that there is m uc m ore о
econom ic grow th than ju st the data on per capita incom e that are s o wi e У use У
economists, E co n o m ic grow th is an historical process o f structur c ange m
time (AD)
broadest sense, and only the most elementary aspects o f this process can be meas
ured by the data on production and income. The form o f structural change most
visible in the statistics is the changing sectoral m ix o f the economy. Chenery,
Syrquin, and Robinson (1986) have illustrated the regularity between the changing
sectoral com position o f the econom y and the increasing level o f productivity, while a
“ deeper” manifestation o f structural change in the long-run process o f economic
growth remains largely the dom ain o f historical research.
Although the argument that technological and organizational innovation are
responsible for this lengthy period o f gradually accelerating growth is appealing,
in fact econom ic theories explaining any such relationship are far from straightfor
ward. Growth theory, especially when focused on the issue o f technology, is a field
characterized by spirited scholarly debate. An im portant current debate is that
between the evolutionary approach and the more neoclassically inspired “ endogen
ous growth theory”.
This chapter argues that the gap between these two approaches is rooted in
fundamental differences in their basic worldviews. While the neoclassical tradition
adheres to a worldview in which cause and effect are clearly separable, and growth
is an ordered, steady state phenomenon, the evolutionary worldview is one o f
IN N O VATIO N AND ECONOMIC GROWTH 489
historical circum stances, co m p lex causal m echanism s that change over tim e, and,
above all, turbulent grow th patterns that appear to be far from a steady state.'
Before these tw o approaches are com pared (in Section 18.3), I discuss some o f the
perspectives found in the earlier literature (Section 18.2). This discussion includes
both highly applied m ethods from the m ainstream toolbox o f econom ics (growth
accounting and the literature on R & D and productivity) and applied w ork from a
post-Keynesian o r Schu m peterian perspective. Section 18.4 outlines a few lines for
future research.
1 8. 2
G rowth and T echnology:
Trad itio nal Economics A pproaches
While technological change and economic growth were at the core of the work of the
classical economists (think of Adam Smith or Karl Marx), these topics largely
vanished from the scene with the neoclassical revolution in economic thinking in
the late nineteenth and early twentieth centuries. The neoclassical growth models
that appeared h a lf a centu ry ago (So lo w 1956) treated technological change as an
exogenous phenomenon. Technology was an explanatory factor “of last resort” in
the sense that growth not explained by the variables included in the model was
assumed to be the result of exogenous technological change. However, when empir
ical work—so called “growth-accounting” (Abramovitz 1956; Solow 1957)2— indi
cated that the unexplained share of long run economic growth tended to be very
high, the interest in technological change and other possible explanatory factors not
taken into account by the modelers increased.
Following Solow (1957), growth accounting commonly starts from the assump
tion of so-called “ neutral technological change ” implying that technological change
improves the productivity of both labor and capital equally. Moreover, all markets
are assumed to be “ perfectly” competitive and in equilibrium. Economies of scale
are assumed to be insignificant.
These assumptions support the following approach to calculating the contribu
tion of “technological progress” to economic growth: subtract from the growth rate
of GDP the weighted growth rates of the capital stock and employment, using the
share of wages in GDP as a weight for employ ment, and subtract from one to get the
weight for the capital stock. What remains, the “ residual, is labeled total factor
productivity” (TFP) growth. This should, following Solow, be seen as the result of
technological progress. Although convenient, the strong assumptions underlying
these calculations are likely to be violated in practice, and the residual almost
certain ly includes m a n y m o re facto rs than ju st the c o n trib u tio n o f tech n ology. T h is
is w h y A b ra m o v itz (1956) called the resid u al “ a m easu re o f o u r ign o ran ce.
O ver the years, the grow th accou n tin g m eth od has been greatly refined. First, the
collection o f m ore refined statistical data allow ed m o re p ro d u c tio n factors to be
distinguish ed, e.g. h u m an capital, vario u s types o f lab or (different b y education al
level), different types o f capital, etc. In this way, the residual sh rin ks, attrib u tin g a
larger part o f it to the factors that are n o w better m easured (D en iso n 19 6 2 ).3 T h e
second line o f extension has been to refine the concept in a theoretical w ay, fo r exam ple
b y assu m in g that som e factors (capital) are q u asi-fixed , i.e. can n ot be reduced or
increased as a result o f sh o rt-ru n flu ctu ation s in o u tp u t grow th (e.g. M o rriso n 1986).
T h e T F P concept rem ains im p o rtan t in studies o f grow th b y eco n o m ists, as it
p ro vid es a “ p ro x im a te ” in d icato r o f the im p act o f tech n o lo gical ch an ge o n grow th.
N on eth eless, the p ro b lem s that rem ain in co n cep tu alizatio n an d m easu rem en t have
m ad e m an y sch olars in the field critical o f its use. P erh ap s the m o st fu n d am en tal
critiq u e is that m an y o f the factors go in g in to the grow th a cco u n tin g calcu latio n s are
interrelated b y causal links n ot accoun ted fo r b y the u n d erlyin g th eo ry.4
G ro w th th e o ry in the 1950s and 1960s w as b ased o n a sim p listic v ie w o f tech n o logy
as a “ p u b lic go o d ” Tech nological kn ow led ge o b v io u sly has so m e ch aracteristics o f a
p u b lic go o d , i.e. m o re than o n e firm can use the sam e piece o f k n o w led ge at the sam e
tim e (n o n -riv a lry ), an d once kn ow led ge is in the open, it is h ard to exclu d e specific
firm s fro m u sin g it (n o n -exclu d ab ility). In its extrem e fo rm , this v ie w leads to the
co n clu sio n that all kn ow ledge can be acquired extern ally as “ general kn ow ledge,”
an d firm s need not d evelop kn ow led ge them selves.
O th er im p o rtan t aspects o f tech nology, how ever, m ake it a p riva te rath er than a
p u b lic go o d (see also Fagerberg, C h . 1 in this vo lu m e ). P u re p u b lic g o o d s do not
require an y special effort o r special skills o n the side o f the co n su m er o r receiver o f
the services o f the go od . T h is is o b v io u sly n o t the case fo r tech n o lo g ical know ledge.
U sin g tech n o logical know ledge, even i f it stem s fro m the p u b lic d o m a in , requires
con siderab le skills and efforts o n the side o f the receiver o f this k n o w led ge. The
reason fo r this is that kn ow led ge has a stro n g ly cu m u lative an d o ften tacit character.
E v ery piece o f n ew kn ow led ge b u ild s to a large extent o n p re v io u s kn o w led ge, an d to
a p p ly know ledge requires that one have co m m a n d o ver the o ld er kn ow led ge on
w h ich the new kn ow led ge builds.
A n u m b er o f m odels developed d u rin g the 1950s and 1960s m ad e tech n o logy
en dogen ou s. In K ald o r (1957) this to o k the fo rm o f a so -called “ tech n ical progress
fu n ctio n ,” w h ich assu m ed a lin ear relation betw een grow th o f la b o r p ro d u ctivity
and the gro wth o f capital p er w orker. K a ld o r’s w o rk gave rise to a sp ecific trad ition ,
often labeled “ P o st-K eyn esian ism ” W ork in this trad itio n takes the role o f dem and
into account exp licitly.5
T h e P ost-K eyn esian trad itio n also em phasizes the role o f “ cu m u lati ve cau satio n ”
o r “ p ositive fe e d b a c k ” C o n tra ry to the neoclassical idea o f k n o w led ge as a public
go o d , these m odels assum e that know ledge is specific to the agents that d evelop it
IN N O V A T IO N AND ECONOMIC GROWTH 49I
and does not spill over easily to other agents or nations. This idea was applied to
regional growth in Kaldor (1970), and goes hack to Verdoorn (1949), Fabricant (1942)
and Young (1928). In this view, generating knowledge is mainly a learning process
deeply rooted in gaining experience with specific production processes and prod
ucts: learning-by-doing and learning-by-using are key concepts. Only those engaged
in the actual learning experiences will gain from it, and others, who do not profit
from experience, will be left behind.
The consequence o f this is a tendency for “ success to breed success” : those nations
(or regions, or agents) that are grow ing rapidly accum ulate experience and hence
learn faster than others. T his leads to a better com petitive position for those already
ahead and enables them to m ove fu rther ahead. Hence, the crucial tendency here is
one o f divergence, in w hich som e nations (regions) are able to grow rapidly while
others are left behind. A m odel o f regional grow th along these lines was presented in
Dixon and Thirlw all (1975):6
An important contribution in this post-Keynesian tradition is Cornwall (1977),
who argues that manufacturing is the leading sector in economic growth because of
the externalities it generates for other sectors. The motivation for this hypothesis is
consistent with the Schumpeterian idea that large innovations have a broad impact
across many sectors (see also Section 18.4.2 below). This is coupled with a view that,
for many countries, the inflow of knowledge from abroad is paramount (see also
Fagerberg and Godinho, Ch. 19 in this volume).
Attempts to generate m odels o f endogenous technological change were also
formulated in the neoclassical trad ition in the 1960s. A rrow (1962) introduced a
model o f learn in g-b y-d oin g as the source o f technological progress, and Uzawa
(1965) and Shell (1967) form ulated full-fledged grow th m odels with endogenous
technological change, w hich in m an y respects can be considered as the front-runners
of the wave o f “ endogenous grow th m odels” that em erged in the late 1980s and early
1990s (see Section 18.3.3).
The w ork on grow th accounting also contributed to the em ergence during the
1970s o f a purely em pirical approach to the issue o f growth and technology that
formulated and estim ated econom etric m odels o f the relationship between G D P and
R&D investment (e.g. G rilich es 1979, 1984). These studies em ploy a production
function that adds a “ know ledge stock” m easure (typically, cum ulative, depreciated
R&D investm ent) to the trad ition al factors o f labor and capital. Estim ates o f the
elasticities o f ou tp u t w ith regard to the variou s production factors suggest that
knowledge (R & D ) has a significant im pact on produ ctivity growth. This approach
has been used at variou s levels o f aggregation: firm s (e.g. G riliches and M airesse
1984)1 sectors (e.g. Verspagen 1995) or countries (e.g. G riliches 1986).
An important issue in this literature is the empirical identification of so-called
R&D spillovers. This goes back to the notion that knowledge is at least partly a public
good and can be used by others than the firm that developed it. In the context of
a production function, spillovers are incorporated by introducing two R&D
k n o w led ge stocks: o n e fo rm ed b y R & D u n d ertak en b y the firm (o r n a tio n , o r sector)
itself, and an o th er o n e fo rm ed b y R & D u n d ertak en b y o th er firm s (n atio n s, sectors;
see Lo s an d V erspagen 2000, fo r a m icro -level ap p lica tio n ). T h ese studies generally
conclude that the social rate o f return to R & D is larger th an the p riv a te rate o f return,
at a n y level o f aggregation. Firm s thus tend to ben efit fro m o th er firm s5 R & D , an d the
sam e h o ld s at the in tern atio n al level: o n e n a tio n s p ro d u c tiv ity g ro w th is to an
im p o rta n t extent d eterm in ed b y that o f others. D espite th eir e co n o m e tric so p h isti
catio n , how ever, these studies reveal little ab ou t the exact ch an n els th ro u g h w hich
sp illovers operate. These channels m ay in clu d e trad ed g o o d s, em p lo yee m obility,
tech n o lo g y alliances, o r even kn ow led ge that is usim p ly in the a ir ” 7
Two m ajor approaches emerged during the 1980s and 1990s as the dominant
approaches to the analysis o f the relationship between technology and growth.
These are the neoclassical approach, which is also dom inant in other fields o f
economics, and the neo-Schumpeterian or evolutionary approach. While the
neoclassical approach consists o f a relatively hom ogenous set o f interrelated
sub-approaches (models), the field o f neo-Schumpeterian or evolutionary econom
ics consists o f a more loosely connected set o f contributions. The evolutionary
approach includes formal models as well as m ore “ appreciative” or historical
approaches, as will be explained in m ore detail below. Even the label used to describe
this approach is not yet com m on understanding. Here, we will use, m ainly for
convenience, the short description o f “ evolutionary economics,” but we include
under this heading a broad category o f work, including what some have called
neo-Schumpeterian economics.
Both o f these approaches agree on basic issues such as the im portance o f innov
ation and technology for economic growth, as well as the positive role that can be
played by government policy for science and technology. Yet they disagree on the
behavioral foundations underlying these respective theories. These differences can
be characterized by saying that the neoclassical theory sacrifices a significant amount
o f realism in terms o f describing the actual innovation process in return for a
quantitative modeling approach that favors strong analytical consistency, while
the evolutionary approach embraces the micro com plications o f the innovative
IN N O V A T IO N AND ECONOMI C G RО W T H 493
process and applies a m ore eclectic approach. G iven these differences, it is useful to
start with an overview o f their analytics in term s o f the m icroeconom ic aspects o f
endogenous technological change and innovation.
turns out to be useless, simply because a prom ising technological idea never makes it
into a successful commercial application.
This has given rise to a distinction in the literature between incremental and
radical innovations. But this distinction obscures the fact that the size distribution of
innovations is not a dichotomy, but instead covers a continuous range o f innovation
sizes. Moreover, there is an im portant interaction and interdependence between
radical and incremental innovations. For example, the first workable steam engine
(the so-called Newcomen engine) was very large and had a lim ited applicability as
well as efficiency. It took more than fifty years for the next step to be taken, Le. James
Watts engine with a separate condenser. If we can characterize the impact o f some
innovations as major, basic, or ’‘radical,” it is only because o f a continuous
stream o f incremental innovations following the introduction o f a basic new design.
Box 18.2 Evolution and the blind watchmaker
Let us use Richard D aw kins's m etaphor o f the blind watchm aker to illustrate the
general idea behind economic growth as an evolutionary process. Dawkinss story
starts from the idea o f W illiam Paley, an eighteenth-century theologian. Paley argued
that certain objects, like a watch, are by their nature obviously created by conscious
design, whereas for others, like a rock, it is easy to believe that they “ have always been
around” His argument then went on to stress that nature contains many such objects
that are obviously created by conscious design, The m ost fam ous o f such objects
discussed by Paley is the human eye. He then used this argument to offer the propos
ition that the world must have been created by a conscious being (God).
Dawkins uses Paley s exam ples to argue that the watch m ay look as if it was carefully
designed (and in the case o f a watch it really was), but it might just as well have been
created by an evolutionary process that can be thought o f as a blind watchmaker. This
blind watchmaker is unable to design the watch by carefully planning it on a drawing
board and then implementing it using precision instruments. Instead, he operates
through the processes o f random mutation and natural selection. His approach is to
start with a simple device and add sm all and simple changes in a random way. These
changes are subjected to a real-world test, i.e., whether or not they lead to an
improvement in keeping the time. Only if they do so are they kept; otherwise they
are discarded. From a new design that incorporates such a successful small change, the
process may start again, and step-by-step a more complicated design emerges. In the
end, after a long and gradual process, a complicated artifact such as a watch may result,
Although this artifact looks as if it were carefully designed, it was instead the blind
watchmaker and his tools o f random mutation and natural selection that created it.
Carrying the metaphor over to economic growth and technology, our watchmaker is
blind because o f the strong uncertainty facing the individual economic decision maker.
No businessman can perfectly foresee the huge potential o f a new innovation when it
first emerges. But it is through a process o f incremental innovations, each one o f which
is implemented by an entrepreneur who sees some market for the newly resulting
artifact, that the full potential o f the technology unfolds. The incremental innovations
are the economic counterpart o f biological mutation. Natural selection has its coun
terpart in economic selection, i.e. markets that decide whether or not certain innov
ations become successful. Just as in biology, many of the “ m utations” (incremental
innovations) are not successful, and the selection process erases them from history.
The metaphor is thus concluded by arguing that, although the individual entrepre
neur has to cope with strong uncertainty and therefore cannot design a process that we
may call a technological revolution, the capitalist system, working by means of a
combination o f the creation o f novelty (innovation) and economic selection
(markets), can create “objects” that seem as if they have been carefully designed.
With hindsight, technological revolutions, such as the diffusion o f steam power or
Information and Communication Technologies (ICTs) may look as if they were
planned from the very beginning to create a “ new economy, but in reality, so it is
argued by evolutionary theory, these technological systems were created by the trial-
and-error method o f the blind watchmaker,
49 6 BART VERSPAGEN
technological development is limited by the paradigm. Still, there is some room for
choice within the paradigm, and these choices are governed by the specific circum
stances (e.g., scarcity o f a particular resource) in which the technology develops. This
development is termed a “technological trajectory”
Thus, in the paradigm/trajectory heuristic, a basic innovation can be thought of as
setting out developments in the techno-economic domain for a number o f years to
come, but the success of the paradigm, and hence of the basic innovation, depends
crucially on how well incremental innovation is able to adapt the paradigm to local
(e.g. industry, geographical and temporal) circumstances. These circumstances
include the skills and capabilities o f the workforce that has to work with new
machinery, as well as factors such as cultural aspects o f the society in which the
paradigm develops.
Another set of heuristics developed in the historical part o f evolutionary econom
ics relates to the temporal clustering o f innovations. This part o f the literature starts
from Schumpeter's observation that innovations “ are not evenly distributed in time,
but that on the contrary they tend to cluster, to come about in bunches, simply
M anagerial
C ap italism
the factory
system
The telegraph process
■.■r*
Networks'
Canal mania
A ssem bly
company belt
ham as A. Ed ison
The rise of
Information
lechanizatiori The rise of The rise of The A m e ri and Communi
of textiles, steam as a electricity as can System cation Techno-
production of pervasive a pervasive of M an u fac logy, - ■
pig iron technology technology turing
because first som e, and then m ost firm s follow in the wake o f successful innova-
tioiT(Schumpeter 1939: 75), A lth ou gh Schum peter was in fact referring to a ten
dency for increm ental in n o vation s to cluster follow in g a large innovation (this is an
idea not incom patible w ith the paradigm view sum m arized above), his idea has been
interpreted in the literature as im p lyin g that large (or “ basic” ) innovations cluster in
time (e.g. Mensch 1979: K leinknecht 1987). In this view, som e historical periods are
characterized by an above average rate o f (basic) innovations, while other periods
show a relatively low rate o f such activity.
Together, these tw o sets o f heuristics have interesting im plications for growth.
They suggest that technological in n ovation can introduce an uneven tem poral
pattern into econom ic grow th. In the early, exp loratory stages o f a paradigm , the
technology progresses rapidly, but the pace o f change slows when the paradigm goes
into its phase o f “ n o rm a l” developm ent, and slows still further when technological
opportunities becom e less n u m erou s (and the paradigm m ay start to break dow n as
a result o f this). The clustering-heuristic suggests variations over time in the rhythm
of growth sim ply because the rate at w hich large, influential innovations occur
differs over time.
One extreme interpretation o f this tem poral pattern o f innovation is the idea o f a
“long wave” in econom ic grow th, in w hich periodicity is bounded in a short range o f
50-60 years (e.g. K leinknecht 1987; Freem an and Lou^a 2001). Another view claims
that growth patterns are inherently turbulent, but w ith little regularity in terms o f
strict cycles. In any case, the evo lu tio n ary view argues that the uneven tem poral rates
of technological change m ean that the econ om y is alm ost always away from anything
that could be characterized as a steady state.
Theories and historical analyses o f this type propose a view o f the interactions
among technology, the econom y, and the institutional context. The institutional
environment is im po rtan t because it is both a facilitator o f and an im pedim ent to
technological change. M oreover, the institutional context is itself an endogenous
factor that changes under the influence o f technological and econom ic develop
ments. Although it is som etim es claim ed that theories o f this type suffer from
“technological d eterm in ism ” (i.e. a tendency for one-w ay causality from technology
to growth: see e.g. B ijk er et al. 1987), w o rk such as that o f Perez (1983) proposes an
interactive relationship am o n g institutions, the econom y, and technology that
emphasizes m utual causality.
is a result of search activities by firms, but search is initiated only when the firm's rate
af return falls below a certain (arbitrarily set) value. Search may take two different
forms: local search or imitation. In the first case, firms search for new, yet undiscov
ered techniques, each o f which has a probability o f being discovered which linearly
declines with technological distance from their current technology (hence the term
lo c a l search). In the second search process, imitation, a firm searches for techniques
currently employed by other firms but not yet used in its own production process.
Like most models in this tradition, the Nelson and Winter model has to be
simulated on a computer to obtain an impression of its implications. The model,
which is calibrated with the Solow (1957) data on total factor productivity for the
United States in the first half of the century, yields an aggregate time path for capital,
labor input, output (GDP), and wages (or labor share in output) that corresponds in
a qualitative sense to those observed by Solow. Based on these results, Nelson and
Winter argue that “ it is not reasonable to dismiss an evolutionary theory on the
grounds that it fails to provide a coherent explanation o f . . . macro phenomena" (p.
226). More specifically, they argue that although both the neoclassical explanation of
economic growth offered by Solow and the Nelson and Winter model seem to
explain the same empirical trends, the causal mechanisms underlying the two
perspectives differ greatly:
the neoclassical interpretation o f long-run productivity change. . . is based upon a clean
distinction between “ moving alon g5 an existing production function and shifting to a new
one. In the evolutionary theory. . , there was no production fu nction.. . . We argu e. . . that the
sharp “ growth accounting” split made within the neoclassical paradigm is bothersome
empirically and conceptually. (Nelson and Winter 1982: 227)
growth (they are specific to the “ evo lu tio n ary” technical change specification o f the
model). However, the im pact o f the savings rate can be com pared between the
various model setups. C o n lisk finds that using purely exogenous technical change
(as in the Solow m odel), or learning b y doing specifications (as in A rrow 1962), the
savings rate does not have an im pact u p on (long-run) econom ic growth. This result,
which is in fact also well k n ow n from standard neoclassical growth theory, marks an
important difference between these m odels and his m ore evolutionarily-inspired
specification.
The recent so-called “ h isto ry-frien d ly m odels” (M alerba et a l 1999) aim to bring
evolutionary m odels closer to em pirical reality by reproducing the historical evolu
tion of a particular industry, e.g. the com pu ter industry. To this end, they start with a
descriptive analysis o f in d u stry variables such as grow th, concentration, and em
ployment, and incorporate the insights fro m this analysis into a m odel the behav
ioral foundations o f w hich are consistent w ith the evolutionary view. This m odel is
calibrated and sim ulated to reproduce real-w orld trends as closely as possible. While
this approach generates em p irically relevant m odels, the sim ulations em ploy a
relatively narrow set o f param eter values. The w ork devotes little attention to a
more open-ended investigation o f w hich minimal set o f assum ptions is necessary to
generate certain aspects o f the structural evolution o f specific industries.
These more open-ended uses o f evo lu tio n ary m icro m odels could lead to a new
class o f models that em p lo y relatively sim ple, evolutionary m icroeconom ic founda
tions to generate a b roader range o f phenom en a in the evolutionary interpretation
of technology and grow th, rather than increasing the sophistication o f the m icro-
foundations. A m uch clearer focus on the salient m acro features and what really
drives them at the m icro level m ay result from this approach, which is necessary to
close the gap between the h istorical, evo lu tio n ary view and m odel building.
grow th to slow down or even cease in the long run. As growth proceeds, capital
accumulates, i.e, the capital stock increases, and hence an extra unit of investment
generates less and less growth. Exogenous growth or productivity (knowledge) had
been the traditional answer, but Romer (1990) and Grossman and Helpman (1991)
proposed to make technology endogenous by modeling the R&D process. Abstract
ing from technicalities (a survey is provided by Verspagen 1992), this can be
summarized as follows.
A ll the m odels assum e that R & D is essentially a lo tte ry in w h ich the prize is a
successful in n o vatio n . In the m o d el b y A g h io n an d H o w itt (19 9 2), this in n o vatio n -
prize buys the firm a te m p o ra ry m o n o p o ly o f su p p ly in g the b est-p ractice capital
g o o d used fo r p ro d u c tio n o f c o n su m p tio n goods. T h e te m p o ra ry m o n o p o ly v a n
ishes w h en the next firm m akes an in n o vatio n . H ence, the in n o v a tio n process is
m o d eled as a “ q u a lity ladder” o f in n o vatio n s, in w h ich each n e w in n o vation
supersedes the o ld one. In the in d u strial o rg an izatio n literatu re, this is called
“ vertical d ifferen tiatio n ” o f produ cts.
In the m o d el b y R o m e r (1990), the in n o va tio n prize bu ys the su ccessfu l firm a new
v a rie ty o f capital that w ill be d em an d ed b y p ro d u cers o f c o n su m p tio n g o o d s forever,
b u t has to com pete w ith all o th er varieties (in vented in the past, w ith the range
co n tin u in g to exp an d in the fu tu re as a result o f R & D ). In this m o d e l, varieties o f
g o o d s (in n o vatio n s) d o n o t go o u t o f the m arket. S u b stitu tio n b etw een variatio n s o f
g o o d s is govern ed b y a u tility fu n ctio n o r p ro d u c tio n fu n c tio n (d ep en d in g on
w h eth er in n o va tio n takes place in co n su m er g o o d s o r in term ed iate g o o d s) w ith a
“ constant elasticity o f su bstitu tion.” T h is is called “ h o rizo n ta l d ifferen tiatio n .”
M o re tickets fo r the R & D lo ttery can be b o u g h t b y d o in g m o re R & D , w h ich is o f
course a costly process. R elative to the e v o lu tio n a ry m o d els co n sid ered above, the
cru cial assu m p tio n Is that the o u tco m es o f the R & D pro cess can be characterized
realistically b y w eak uncertainty, i.e. the firm is able to estim ate the p ro b a b ility that it
w ill get the in n o vatio n -p rize given its level o f R & D sp en din g. W ith exp ected benefits
an d costs o f R & D k n o w n , the firm m a y m ake a c o st-b e n e fit an alysis an d derive an
o p tim al le vel o f R & D spen ding. T h is w ill, on average, co rresp o n d to a given am ount
o f in n o vatio n , and p ro d u ce a given grow th rate. A lth o u g h a d d itio n a l assum ption s
are necessary (e.g. w ith regard to the w o rk in g o f capital m arkets in w h ich R & D
exp en d itu res have to be fin an ced ), this m ech an ism is the k ey to gen eratin g en d o gen
ous grow th.
Before en d o gen o u s grow th is possib le in these m o d els, there is o n e essential
assu m p tio n ab ou t the nature o f tech n o lo g y that needs to be m ade. T h is is related
to the (partly) p u blic go od nature o f technology. In the n ew gro w th m o d els, this is
represented by the assu m p tio n that there are tech n o lo g y sp illo vers b etw een firm s in
the R&D process. T h e assu m p tio n takes tw o fo rm s, d ep en d in g o n w h ich flavor o f
m odel is used. In the horizontal d ifferen tiation type m o d els (also called “ love-
o f-v a rie ty m o d els), each in n o vatio n increases the level o f gen eral know ledge
available in the econom y, and this increases the p ro d u c tiv ity o f the R & D process
in n o v a t io n and ec o n o m ic gro w th
503
itself (Romer 1990). T h is assu m ptio n is necessary because o f the ever m ore severe
competition between the varieties o f capital goods, and the falling profit rates that
this causes. A tendency fo r R & D to be m ore productive (i.e. the costs o f R & D to fall)
offsets this falling pro fit rate, and keeps R & D feasible in the long run (see Grossm an
and Helpman 1991).
In the quality ladder m odels (vertical differentiation), each new innovation
destroys the m o n o p o ly o f the old innovator. H owever, the new innovator also builds
on the previous in n ovation , because the quality o f the new capital good is a fixed
increase over the previou s one. In other w ords, each new innovator is “ standing on
the shoulders o f g ia n ts” and know ledge spills over intertem porally from one
innovator to the next one. W ithout this spillover, endogenous growth would not
be possible.
The technological spillovers in endogenous growth m odels lead to increasing
returns to scale at the aggregate level. Even though the production functions o f firms
at the micro level are characterized by constant returns to scale, the R & D spillovers
that flow from one firm to the rest o f the econom y im ply increasing returns at
the aggregate level. In term s o f the expression for the aggregate growth rate o f the
economy, this feature o f the endogenous grow th m odels im plies that growth at the
country level depends (ceteris paribus) on the size o f the country. Taken literally, this
means that (ceteris paribus) larger countries will grow m ore rapidly. Related to
this issue is the fact that the basic endogenous grow th m odels are quite sensitive to
small changes in the m odel specification w ith regard to technology spillovers.
A slightly different specification o f the im pact o f “ general knowledge” on R & D
productivity will lead to either zero grow th in the long run, or to increasing growth
rates in time (G rossm an and H elp m an 1991).
Technological spillovers m ake endogenous grow th possible, but pose a challenge
for policy makers. W hen tech n ology generates positive externalities, the social
benefits o f R & D are larger than the private benefits (a rational firm investing in
R&D does not consider the benefits o f its R & D for its com petitors). Hence the
amount o f R & D investm ent “ generated b y the m arket” w ill be too low from a social
point o f view. Technology p o licy in the form o f R & D subsidies m ay bring the
economy to a higher, so cially optim al grow th path. A sim ilar conclusion is reached
in a model o f h um an capital and grow th in Lucas (1988). In A ghion and H owitt
(1992), there is also a negative externality: each new innovator destroys the rents o f
the existing m o n o p o list (this is called “ business stealing,” or, in line with Schum p-
eter (1939), “ creative destru ction ” ). In this m odel, private R & D investm ent also can
be too high from a social w elfare perspective, depending on which o f the two form s
of externalities (creative destruction or standing on the shoulders o f giants) is
stronger.
The development of this new class of models raises promise and problems. On the
positive side, it can be argued that this new growth theory takes seriously a number
of arguments about technological change previously championed by evolutionary
5©4 B A S T VEBSPAGEM
theorists but' ignored by mainstream economists, These include the notion that
R&D and technology are essentially stochastic phenomena (although evolutionary
theory would argue that the type o f uncertainty, i.e. weak uncertainty in which the
probability distribution is known, is still not very adequate), and the importance o f
technology flows between agents (spillovers) for growth in the long run, The
implication in many o f these models that technology policy matters for growth
also is relatively consistent with evolutionary theory, but may be less easily accom
modated by mainstream economic theories that emphasize the efficiency o f market
forces.
On the negative side, these new growth models still propose a view o f the
interaction between economic growth and technology that differs substantially
from that of evolutionary theory. The evolutionary view is one in which contingency
and more systematic factors blend together in the dialectical process of historical
time, but the new growth theory is still much closer to a Newtonian clockwork world
in which there is a certain degree of “weak” uncertainty. In other words, the new
growth theory still portrays the relationship between technology and growth as one
of a steady-state growth pattern, which can be “tweaked” relatively easily by turning
the knobs of the R&D process.
The evolutionary inclination, on the other hand, is that the nature of the growth
process is more complex and variable over time. While the importance attached to
the technology factor is shared with the new growth models, the belief that the
relation between technology and growth is easily tweaked is not. In the evolutionary
view, it is hard to predict exactly the impact of a policy measure, because it impacts
on a complex range of interrelated factors. Moreover, while relations between a
number o f factors may have been revealed by careful research for a specific instance
in time, it is to be expected that the nature of this relationship will change over time,
exactly because of the (co)evolutionary nature of the process.
A more recent branch of new growth theory is the group of models that comes
under the heading of “general purpose technologies” (GPT, Helpman 1998). A GPT
is defined in essentially the same way as a basic innovation or paradigm in the
evolutionary tradition. It consists of a basic technology (radical breakthrough), but
this needs to be developed in the form of a range of intermediate (capital) goods.
Within each GPT, the determinants o f productivity are essentially the same as in one
o f the variants of the new growth models discussed above: technological change
takes the form o f an ever-expanding range of capital goods, but this is time-specific
to the GPT. Thus, we see that at least two ideas from the evolutionary tradition are
captured: the idea o f differences in innovation size, and the idea that incremental
innovations are responsible for the diffusion of a basic technology.
The GPT model generates cyclical growth. In its simplest form, the cycle consists
o f two phases. In the “ low growth phase,” the new GPT has been discovered, but is
not yet in operation. New capital goods are being developed for it, and this activity
has been halted for the old GPT. Thus, economic growth is low, because the main
IN NOVATION AND ECONOMIC GROWTH 505
technology in use is no lon ger being developed. O nce enough capital goods are
available for the new G PT, its p ro d u ctivity outperform s that o f the old GPT, the old
GPT vanishes, and the econ o m y shifts into a “ high growth phase.”
The G PT m odel resem bles the evolutionary, Schum peterian idea o f long waves in
economic growth. But scholars in the latter tradition have m oved away from the
fixed and determ inistic cycle that characterizes the G P T m odel. Its clockw ork view o f
economic growth has been dom in an t in the neo-classical tradition since the Solow
model. One illustration o f the lim itations resulting from this view is the fact that, in
the GPT view o f the w orld , there is o n ly roo m for substitution between subsequent
paradigms. But econom ic and technological histories are filled with exam ples o f the
adaptation and survival, often in m o dified form , o f old paradigm s. For exam ple,
although the autom obile is typ ical o f the m ass-production paradigm , it still plays a
crucial role in the m odern “ In fo rm ation Econom y,” although IC T has indeed been
applied in the pro d u ction o f cars.
In conclusion, the evo lu tio n ary tradition and the neo-classical tradition have
converged som ew hat in the ph enom en a deem ed central within each analytic ap
proach. But they disagree on the essential nature o f the growth process. The
neoclassical th eo ry conceptualizes grow th as a determ inistic process in which
causality is clear-cut, and p olicies can be built on an understanding o f tim e-invariant
determinants o f grow th patterns. In the evolutionary view, on the other hand,
contingencies and specific historical circum stances play a larger role, and causal
mechanisms that prevail in one period m ay be subject to endogenous change in the
next. In such a w orld , designing p o licy is harder, but not im possible.
the estim ation results are sensitive to a sm all n u m b e r o f o b serva tio n s in the
large sam ple (L evin e and Renelt 1992). N on eth eless, this w o rk lead s to the co n clu
sio n that stead y state grow th rates d iffer betw een n ation s, G ro w th rates m ay
converge tow ard a co u n try-sp ecific stead y state gro w th p a th at best (so-called
co n d itio n al con vergen ce), lead in g to the d ivergen ce o f g ro w th path s a m o n g co u n
tries, G ro w th seem s to be h eterogen eou s am o n g co u n tries startin g fro m lo w levels o f
G D P p er capita, w ith so m e cou n tries falling b eh in d , and so m e co u n tries b ein g able
to catch up. T h is p h en o m en o n is discu ssed in m o re detail in F ag erb erg an d G o d in h o
(C h , 19 in this vo lu m e).
Jo n es (1995a an d b) has argu ed that the o b served e m p iric a l reco rd o n R & D and
grow th is incon sisten t w ith the theoretical p red ictio n s o f e n d o g en o u s grow th
m o d els (see B o x 18,3 on the wJon es critiq u e” and se m i-e n d o g e n o u s gro w th m odels).
H e observes that the p o stw ar em p irical eviden ce does n o t c o n firm the relationship
p ro p o sed b y R & D -b a se d en d o gen ou s grow th m o d els that an in crease in the n um ber
o f R & D w orkers leads to h igh er rates o f e co n o m ic grow th . Jo n e s notes that the
n u m b er o f R & D w o rk ers has increased since the 1960s, b u t gro w th rates ( o f total
facto r p ro d u ctivity) have either been constant o r d eclin in g d u rin g the sam e period.
T h e so-called “ Jo n es critiq u e” has led to still m o re w o rk in the e n d o g en o u s grow th
trad itio n since its pu b licatio n . Jo n es (1995a) suggests а so -called sem i-en d o gen ou s
grow th m o del, w h ich appears to be m ore consisten t w ith the e m p iric a l facts, but in
w h ich en d o gen o u s grow th o n ly takes place w h en the p o p u la tio n grow s.
Figure 1 8 ,3 illustrates the Jones critique for the United States and the European Union
during the 1 9 8 0 s and 1 9 9 0 s. The R&D-based endogenous growth models predict that
the growth rate of an economy, which is here approximated by total productivity factor
growth, depends on the number of researchers in R&D. We see a steady increase in the
latter, both in the US and the EU, but total factor producti vity growth does not display a
clear trend—instead it fluctuates widely around a roughly constant level. Does this
constitute evidence against the relationship between innovation and economic growth?
Jones suggests an alternative model, which differs from the R&D-based endogenous
growth models by Romer, Grossman and Helpman, and Aghion and Howitt by a
different specification of the invention process. Whereas these original R&D-based
growth models assumed that the growth rate of knowledge depends on the number of
R&D workers in a linear way, Jones assumes that there are decreasing returns to R&D
labor. This assumption is based on the idea that "the most obvious ideas are discovered
first, so that the probability that a person engaged in R&D discovers a new idea is
decreasing in the level ofknowledge ... [and] the possibility that at a point in time the
duplication and overlap of research reduce the total number of innovations” (Jones
1 9 9 5 a; 765)* in this so-called semi-endogenous growth model, endogenous growth is
only possible when the population grows.
INN О V A 1 ION AND E C O N O M IC GROWTH
507
№
C
О
R§. 18.3 The Jones critique. Total factor productivity growth trends are
flat while the number of researchers in R&D increases (R&D researchers
on left scale, tfp on right scale).
Source: Source for tfp data: Groningen Growth and Development Centre Total Economy Growth Accounting Database.
■'Ш шш -Ші Ш Ш - i ste: OECO М а ш Science andTertinology Indicators Database.
From the point of view of evolutionary growth theory, the Jones critique appears to
be the result o f the misguided emphasis on steady-state growth states in the R
based endogenous growth models. The assumed relationship between R&D labor, the
number of innovations, and resulting economic growth is based on assumptions ot
va^Mbrium bdiavior and weak, uncertainty, in the less mechanistic wo:utionary
woiii innovation, R&D, and growth are linked in a less rigid relationship that may
f aenge over time as a result of new and radical technological deve pments. n is
S i i wHdie specific relationship between R&D labor and TFP grewt о serve у ones
j|iB8 yweB be specific to the historical circumstances of the period, and may be subject о
change in the future.
are em b od ied in traded go od s, and, hence, that R & D w eigh ted b y trad e flo w s m a y be
used to m easure them . T h e em p irical analysis b y C o e an d H elp m an sh ow s th at the
co rrelatio n s betw een T F P grow th an d this m easu re are in d eed stro n g , suggesting
that trade is an im p o rta n t sou rce o f kn o w led ge sp illovers. H o w ever, subsequen t
co n trib u tio n s sh ow that oth er w eig h tin g schem es m a y p ro v id e d ifferen t in terpret-
ation s. Fo r exam p le, Lich ten b erg an d V an Pottelsbergh e (1996) sh o w th at Foreign
D irect In vestm ent (F D I) m ay be a carrier o f sp illo vers an d V erspagen (x997) show s
the im p o rta n ce o f in ter-sectoral sp illovers, w h ile K eller (1998) ls critical o f the
vario u s w eigh tin g schem es an d b en ch m ark s th em again st a ra n d o m w eig h tin g
schem e. These results also are sensitive to the m easu rem en t b y e m p irica l researchers
o f ab sorp tive cap acity in the sp illo ver-receivin g cou n tries.
A n interesting “ m erg er” betw een the em p irical trad itio n on p ro d u c tiv ity and
R & D , o n the one h an d, an d new grow th th e o ry on the o th er h an d , is the em p irical
m o d el b y E aton and K o rtu m (1999). T h is p ap er p ro vid e s a m o d el in w h ich in n o v
atio n and tech n o lo gy d iffu sio n are b oth d rivers o f co u n try -lev e l grow th . T h e m odel
is m o tivated b y em p irically observed trends, an d is estim ated w ith data o n tech n o l
o g y in d icators (patents, R & D ) and grow th. T h e results o f the estim atio n s sh o w that
both en d o gen o u s R & D and the d iffu sio n o f k n o w led ge betw een co u n tries co n trib
ute to grow th , alth o u gh the m ix betw een these tw o sou rces d iffers greatly between
co u n tries and tim e perio d s. T h is ap p ro ach an d its co n clu sio n s also has m u ch in
co m m o n w ith earlier tech n o lo g y gap m o d els such as that o f N elso n (1968), as
su rveyed b y Fagerb erg and G o d in h o (ch. 19 in this v o lu m e ). F ag erb erg an d V erspa
gen (2002) recently reassessed the p o st-w a r evidence fo r these typ es o f m o d els, and
co n clu d ed that, o ver tim e, in n o va tio n has b eco m e a m o re im p o rta n t source o f
grow th as co m p ared to the “ p u re” im itatio n o f fo reign tech n o logy. M o d e ls such as
that o f E ato n an d K o rtu m thus have great p ro m ise to gu id e n ew g ro w th th eo ry in a
d irection that has m u ch in co m m o n w ith the h isto rica lly -in sp ired e vo lu tio n a ry
ap p ro ach .
18.4 O u t l o o k for T h e o r e t i c a l R e s e a r c h
on I n n o v a t i o n a n d G r o w t h
Is further convergence o f the tw o traditions likely, as Heertje predicted for the end
of the (previous) century? O ne avenue for convergence is in the further analysis o f
the intertemporal variab ility o f grow th patterns. At least som e new growth m odels
(e.g. Aghion and H ow itt 1992) argue that tim e series o f econom ic growth show
variability and this is a m ain top ic in evolu tion ary m odels. The application o f
Pareto-type probability d istribution s, in w hich very large innovations have non-
negligible p ro b ab ility m ay b rin g the two approaches closer together, since they
provide an intuitive w ay o f m odelin g "stro n g uncertainty” (see e.g, Sornette and
Zajdenweber 1999).
Each o f the tw o approaches also contains a range o f im portant and interesting
lines o f research to be pu rsued. In the endogenous growth tradition, the returns to
purely theoretical w o rk seem to have slowed dow n, but im portant em pirical chal
lenges remain open. The m ost fru itfu l avenue o f research here seems to be further
theoretical refinem ent indu ced b y em pirical w ork on technology and growth, with
the explicit aim o f developing em pirically relevant m odels instead o f new explor
ations m otivated b y technical problem s w ith the existing models. For a long time,
empirical research has led the w ay in the m ainstream analysis o f technology and
growth, and this app roach still seem s to be the w ay forw ard.
Two main challenges confront the evolu tion ary tradition. The first is to develop a
research program that goes beyond ju st em ulating, although with a m ore plausible
micro-foundation, the results o f neoclassical analysis. Such an extension o f the
evolutionary research agenda could benefit from closer interaction with the non-
formal w ork in the evo lu tio n ary trad ition and greater reliance on historical research.
Evolutionary m odelers could seek to explain observed historical regularities in the
relation between grow th and technology.
A second challenge for evolutionary theorists is the development o f more practically
relevant models, for exam ple, with regard to specific policy advice. Evolutionary
theory rarely generates precise policy advice (see also Lundvall and Borras, Ch. 22 in
this volume), m ostly as a result o f the nature o f the theory that points to com plex
interactions and rather unpredictable dynam ics as im portant ingredients o f the
economic environm ent. To a certain extent, evolutionary theory will argue for a
change in the w ay p olicy is view ed, but m ore precise w ork on how this can be
implemented to achieve higher o r m ore sustainable econom ic growth remains
crucial
510 BART VERSPAGEN
1. The (older) data are necessarily rather imprecise, but the general trends are plausible on
the basis o f historical evidence. Note that since the vertical axis displays the logarithm o f
per capita income, a straight line would correspond to growth at a fixed rate, the slope o f
the line indicating the growth rate.
2. Solow (1957) is often quoted as the standard reference on growth accounting, but the
ancestry o f the method lies earlier (e.g. Tinbergen 1943 and Abramovitz 1956; for an
overview see Abramovitz 19 89 :13-15).
3. Well-known studies in this tradition are Denison (1962, 1966), Jorgensen (1967) and
Maddison (1987,1991): see Nadiri (1970) for an early overview o f the methodology.
4. Critical surveys o f the method can be found in Nelson (1973,1981) and Fagerberg (1988b).
5. Pasinetti 1993 analyzes growth and technology from a demand perspective.
6. An elaborate overview o f (empirical as well as theoretical) work on growth in the post-
Keynesian tradition is in McCombie and Thirlwall (1994)* A specific application to the
issue o f technology dynamics and growth is in Fagerberg (1988a).
7. Griliches (1992) provides a broad overview o f empirical studies estimating R8cD spill
overs; Cincera and Van Pottelsberghe (2001) provide a survey on international spillovers,
Van Pottelsberge (1997) on intersectoral spillovers.
8. An early attempt to develop a heuristic similar to the ones cited by Silverberg is in Sahal
(1981).
R efer en c es
A b r a m o v it z , M. A . (1956 ), “ Resources and Output Trends in the United States since 1870,”
American Economic Review 46: 5 -2 3 .
------ (1989), Thinking About Growth, Cambridge: Cambridge University Press.
A g h i o n , P., and H o w i t t , P. (1992), “A Model o f Growth Through Creative Destruction,”
Econometrica 60 (1992): 323-51.
A r r o w , K. J. (1962), “ The Economic Implications ofLearning by Doing,” Review of Economic
Studies 29:155-73.
B a r r o , R. J. (1991), “ Economic Growth in a Cross-Section o f Countries,” Quarterly Journal of
Economics 106: 407-43.
B i j k e r , W. E., H u g h e s , T P., and P in c h , X (eds.) (1987), The Social Construction of
Technological Systems, Cambridge, Mass.: M IT Press.
C h e n e r y , H. B ., S y r q u i n , M., and R o b i n s o n , S. (1986), Industrialization and Growth: A
Comparative Study, Oxford: Oxford University Press.
C h i a r o m o n t e , E , and D osi, G. (1993), “ Heterogeneity, Competition, and Macroeconomic
Dynamics,” Structural Change and Economic Dynamics 4: 39-63.
C in c e r a , M., and V an P o t t e l sb e r g h e , B. (2001), “ International R & D Spillovers: A
Survey,” in M. Cincera, Cahiers Economiques de Bruxelles 169: 3-32.
C o e , D. I., and H elp m a n , E. (1995), International R&D Spillovers,” European Economic
Review 39: 859-87.
An Aggregate Model o f Technical Change,” Quarterly Journal of
* C o n l i s k , J. (1989),
Economics 104: 787-821.
IN N O V ATIO N AND ECONOMIC GROWTH 511
Robertson,
Crafts, N. E R. (1985), British Economic Growth During the Industrial Revolution. Oxford:
Oxford University Press,
Denison, E. (1962), The Sources of Economic Growth in the United States and the Alternatives
Before Us, Washington: Committee for Economic Development.
-----(1966), Why growth rates differ>Washington: Brookings Institution.
Dixon, R. J., and T h irlw a ll , A. P. (1975), “A Model of Regional Growth-Rate Differences on
Kaldorian Lines,” Oxford Economic Papers n: 201-14,
Dosi, G. (1982), ‘'Technological Paradigms and Technological Trajectories,” Research Policy
11:147-62.
Eaton, J., and K ortum , S. (1999), “ International Technology Diffusion: Theory and Meas
urement,” International Economic Revim' 40: 537-70.
Fabricant , S. (1942), Employment in Manufacturing 1899-1939, New York: NBER.
Fagerberg , J. (1988a), “ International Competitiveness,” Economic Journal 98: 355-74.
----- (1988b), “ Why Growth Rates Differ” in G. Dosi, C. Freeman, R. R. Nelson, G. Silver-
berg, and L. Soete (eds)., Technical Change and Economic Theory; London: Pinter, 87-99.
----- and V e r s p a g e n , B. (2002), “ Technology-gaps, Innovation-diffusion and Transform
ation: an Evolutionary Interpretation,” Research Policy 31:1291-304.
Freeman, C , and Lou^A, F. (2001), As Time Goes By: From the Industrial Revolutions to the
Information Revolution, Oxford: Oxford University Press.
----- and Soete , L. (1997), The Economics of Industrial Innovation, 3rd edn., London and
Washington: Pinter.
*------------ (1990), “ Fast Structural Change and Slow Productivity Change: Some Paradoxes
in the Economics o f Information Technology,” Structural Change and Economic Dynamics
1:225-42.
Griliches , Z. (1979), “ Issues in Assessing the Contribution o f Research and Development to
Productivity Growth,” The Bell Journal of Economics 10: 92-116,
----- (1980), “ R&D and the Productivity Slowdown,” American Economic Review 70:343-8.
----- (1984), R&D, Patents and Productivity>Chicago: Chicago University Press.
----- (1986), “ Productivity, R&D and Basic Research at the Firm Level in the 1970s,” American
Economic Review j 6 : 141-54.
* --- (1992), “ The Search for R&D Spillovers,” Scandinavian Journal of Economics 94:
S29-S47.
----- (1996), “ The Discovery o f the Residual: A Historical Note,” Journal of Economic
Literature 34:1324-30.
----- and M a i r e s s e , J. (1984), “ Productivity and R&D at the Firm Level,” in Z. Griliches
(ed.), R&D, Patents and Productivity>Chicago: Chicago University Press, 339-74*
* G r o s s m a n , G. M., and H e l p m a n , E. (1991), Innovation and Growth in the Global Economy;
Cambridge, Mass.: M IT Press.
G r u e b l e r , A, (1990), The Rise and Fall of Infrastructures: Dynamics of Evolution and
Technological Change in Transport, Heidelberg: Physica-verlag.
H e e r t j e , A , (1994 ), “ Neo-Schumpeterians and Economic Theory, in L, Magnusson (ed.),
Evolutionary Approaches to Economic Theory, Dordrecht: Kluwer, 265—76.
INNOVATION AND
CATCHING-UP
JAN FAGERBERG
M A N U E L M. G O D I N H O
19.1 I n t r o d u c t i o n
T h e h isto ry o f cap italism fro m the In d u strial R evo lu tio n o n w ard s is on e o f increas
in g differences in p ro d u c tiv ity and livin g co n d itio n s across d ifferen t p arts o f the
globe. A cco rd in g to o n e source, 250 years ago the d ifferen ce in in co m e o r p ro d u c
tiv ity p er head betw een the richest and poorest c o u n try in the w o rld w as a p p ro x i
m ately 5 : 1 , w h ile to d ay this difference has increased to 4 0 0 : 1 (L an d es 1998).
H ow ever, in spite o f this lo n g -ru n trend tow ards d ivergen ce in p ro d u c tiv ity and
in com e, there are m a n y exam p les o f (initially) b ack w ard co u n tries that— at different
tim es— have m an aged to n a rro w the gap in p ro d u c tiv ity an d in co m e between
them selves and the fro n tier cou ntries, in o th er w o rd s, to “ catch up.” H o w d id they
do it? W hat w as the role o f in n o va tio n and d iffu sio n in the process? T h ese are am o n g
the q u estion s that we are go in g to discuss in this ch ap ter.1
The “ catch-up” question should be seen as distinct from the discussion o f
“ convergence ” although the two issues partially overlap. “ Catch-up” relates to the
ability o f a single country to narrow the gap in productivity and incom e vis-a-vis a
leader country, while “ convergence” refers to a trend towards a reduction o f the
overall differences in productivity and income in the world as a whole. The issue o f
I N N O V A T I O N A N D C A T C H I N G - UP 515
convergence has been central to the econom ists research agenda, in part because
some prom inent theoreticians form ulated m odels o f long-run growth im plying
such convergence (Solow 1956),2 O f course, if all countries below the frontier catch
up, convergence w ill necessarily follow, But if only som e countries catch up (and
perhaps forge ahead), w hile others fall behind, the outcom e with respect to conver
gence is far from clear (Abramovitz 1986), W hat the em pirics show is that, at best,
such convergence is confined to groups o f countries— or “convergence clubs”
(Baumol et a l 1989)— in specific tim e periods. Arguably, to explain such differences
in the conditions for catch-up through tim e, it is not enough to rely on general
mechanisms. Som e historical perspective is required.
During m ost o f the nineteenth century, the econom ic and technological leader o f
the capitalist w orld was the U nited K ingdom , with a G D P per capita that was 50 per
cent above the average o f other leading capitalist countries. However, during the
second h alf o f the century, the U nited States and G erm any both started to catch up
and substantially reduced the U K lead. T h ey did not achieve this growth b y merely
imitating the m ore advanced technologies already in use in the leading country, but
rather did so b y developing new ways o f organizing production and distribution, e.g.
by innovating (Freem an and Soete 1997; Freem an and Lou^a 2001). In the case o f the
US, this led to the developm ent o f a historically new and dynam ic system, based on
mass production and the d istribution and exploitation o f econom ies o f scale.
Germany introduced new w ays o f organizing production, particularly with respect
to R&D in the chem ical and engineering industries, that in the long run w ould com e
to have a very im portan t im pact. M ore recently, the very rapid catch-up o f Japan to
Western p ro d u ctivity levels d u rin g the first h alf o f the twentieth century was
associated with a n u m ber o f very im portant organizational innovations (such as
the Ujust-in-tim e system ” see B o x 19.1) that, am ong other things, totally transform ed
the global car industry, These innovations did not only benefit Japan, but diffused
(with a lag) to the established leader (the U SA ) and contributed to increased
productivity there.
As these b rie f exam ples show, successful catch-up has historically been associated
not merely w ith the adoption o f existing techniques in established industries, but
also with innovation, particu larly o f the organizational kind, and with inroads into
nascent industries. H ow ever, as is equally clear, this has been done in different ways
and with different consequences. I f we extend the perspective to the m ost recent
decades, as we w ill do in this chapter, this diversity in strategies and perform ance
becomes even m ore striking. In the next section, w e discuss som e o f the perspectives
that have em erged in the catching up literature. Section 19.3 extends the perspect
ive to the m ost recent decades, com pares cases o f successful catch-up to less
successful ones, and considers the lessons that m ay be drawn. Finally, Section 19.4
raises, by w ay o f con clu sion , the question o f what present day developing countries
can learn, particu larly w ith respect to policy, from the literature on innovation and
catching up.
516 JAN F A G E R B E R G A N D M A N U E L M. G O D I N H O
19.2 L e s s o n s fr o m t h e L i t e r a t u r e
under “otherwise suitable circ u m sta n c e s"4 largely “ a question o f the pecuniary
inducement a n d . . . opp ortu n ities offered by this new in du stry" (ibid, 192), Since
the latecomers could takeover the new technology ready-m ade, 1 without having to
share the costs o f its developm ent, this m ight be expected to be a very profitable affair
(ibid, 249)* This being the case, Veblen predicted that other European countries, e g
France, Italy, and Russia, w o u ld soon follow suit (he also m entioned the case o f
Japan).
While in Vebleffs interpretation, the G erm an catch-up was a relatively easy affair,
the economic historian A lexan d er G erschenkron (1962) took a different view,
emphasizing the d ifficu lty o f the m atter. W hile, he argued, technology at the time
Britain industrialized w as sm all scale, and hence institutionally not very dem anding,
these conditions were rad ically altered in the nineteenth century when Germ any
started to catch up. W hat G erschenkron particularly had in m ind was the seemingly
inbuilt tendency o f m odern tech nology to require ever larger and m ore com plex
plants (static and d yn am ic econom ies o f scale), with sim ilarly changing require
ments with respect to the physical, financial, and institutional infrastructure. He
argued that, because o f the high potential rewards from successful entry, and the
heavy transform ation (m odern ization) pressure on the rest o f the econom y it helped
to generate, it was o f p aram ou n t im portance for the latecom er to target such
progressive, dynam ic industries, and to com pete globally through investing in the
most modern equipm en t/plan ts.5 H ow ever, to succeed, catching-up countries had,
in Gerschenkron s view , to build up new “ institutional instrum ents for which there
was little or no cou nterpart in the established industrial cou n try" (1962: 7), The
purpose o f these institutional instrum ents w ould be to m obilize resources to
undertake the necessary changes at the new and radically enlarged scale that m odern
technology required. H is favorite exam p le6 was the G erm an investm ent banks (and
similar examples elsew here in Europe), but he also adm itted that, depending on the
circumstances, other types o f institutional instrum ents, such as, for instance, the
government (in the R ussian case),7 m ight conceivably perform the same function.
Gerschenkroffis w o rk is often identified w ith his focus on the role o f banks in
industrialization, although as pointed out by Shin (1996), it is possible to see it as an
attempt to arrive at a m ore general th eo ry about catch-up, focusing on certain
requirements that m ust be m et fo r successful catch-up to take place, as well as
different, though “ fu n ctio n ally equivalent," institutional responses (or catch-up
strategies). A n im po rtan t chain in G erschenkron’s argum ent is the em phasis on
the advantages o f targeting rap id ly grow ing, technologically advanced industries. It
should be pointed out, how ever, that for h im this was a generalization based on
historical evidence. T h u s it is not o bviou s that his recom m endations w ould be
equally relevant for later tim e periods/technologies. N either did he rule out that
there might be other paths to successful industrialization than the one he recom
mended, although he held that to be rather exceptional. For instance, he pointed to
Denmark as an exam ple o f a co u n try that m anaged to catch up w ithout targeting the
5l 8 I A N F A G E R B E R G A N D M A N U E L M. G O D I N H O
p ro gressive in d u stries o f the tim e, an d exp lain ed this w ith its close lin k s to the
rap id ly gro w in g B ritish m arket fo r ag ricu ltu ral p ro d u cts.
W e m a y use Veblen s an d G ersch en k ro n s accou n ts o f the G e rm a n catch -u p to
m ake a p re lim in a ry classificatio n o f catch -u p strategies. T h e ty p e d escribed by
Veblen assum es that tech n o lo g y is easily available/transferable, n o t v e ry d em an d in g
in term s o f skills o r in frastru ctu re, and that m ark et forces are able to take care o f the
necessary c o o rd in a tio n w ith o u t the large-scale in vo lvem en t o f extern al “ change
agents/’ In contrast, there is the G ersch en k ro n ian case in w h ich te c h n o lo g y tran sfer
is so d em an d in g in term s o f skills/in frastru ctu re that m ark et fo rces, i f left alon e, are
co n sid ered u n lik ely to lead to success, an d so so m e degree o f active in terv en tio n in
m arkets b y ou tsid ers, w h eth er private o rg an izatio n s o r p arts o f go vern m en t, is
d eem ed necessary.
leading sectors. R & D activity also flourished, partly for m ilitary needs, and was,
according to one source (Odagiri and Goto 1996), well above 1 percent of GDP in the
early 1940s.
The defeat o f Jap an in W orld W ar II changed the pow er structure in Japanese
society by eliminating two of the three contending power centers, the military and
the (owners of the) Zaibatsus, hence giving a boost to the bureaucracy that once
more took on the challenge of gearing the economy and the society at large towards
economic catch-up with the West. The sequence of events from the late nineteenth
century somehow repeated itself, with a very important role for the state (and—in
particular—the Ministry for Trade and Industry, MITI) in the early phase, and a
growing role for private initiatives (and business groups) as the economy grew
stronger (with no role left for the military). The new business groups that emerged,
the Keiretsus, were in som e cases based on the pre-w ar groupings (which had been
dissolved b y the A m erican o ccu p atio n forces), w hile in other cases they were totally
new constructions. T h is reorientation differed from the pre-w ar groupings in
respect o f a stronger role fo r banks, and a sm aller role for private investors/family
ownership. In the early phases, the banks were very dependent on credit from the
state, reinforcing the power of the bureaucracy in getting business to cooperate with
the government in its preferred catch-up path. A fter a few decades o f rapid growth,
the banks (and business m o re generally) grew m ore independent, and the role o f the
state diminished and to o k on m ore “ norm al,” W estern proportions.
The exact role o f the govern m en t versus private actors in the various phases o f
Japanese econom ic grow th is a m atter o f considerable controversy, and we shall not
attempt to resolve it here. Suffice it to say that governm ent/bureaucracy interven
tion, through activist econom ic, industrial, and trade policy (protectionism ), was
very im portant, especially in the early phases. A lthough not everything it touched
turned in to 41gold,” and som etim es its interventions were strongly resisted by private
business (and for perfectly go od reasons), there is no doubt that it contributed
significantly to focusing the attention o f private business to catch-up with the West.
An important elem ent in this catch-u p process (and the policies that were pursued)
was a very rapid but ord erly process o f structural change, through which industries
“of the past” were grad u ally phased out in favor o f technologically m ore progressive
industries, em phasizing in particu lar the com bination o f econom ies o f scale,
product differentiation, and rap id ly grow ing dem and, on the one hand, and con
tinuous im provem ents o f produ cts and processes through learning, on the other. In
this way, Japanese in d u stry soon rose to the pro d u ctivity frontier in its chosen fields,
first in the steel in d u stry and in sh ip-bu ild ing, and later in cars and (consum er)
electronics.9 A lth ou gh Japan ese in n ovation in the catch-up phase also included a
large num ber o f pro d u ct in n ovation s, especially o f the m in or type (adaptations to
demand), the m ain em phasis w as on process innovations, particularly o f the organ
isational type, that allow ed for sim ultaneous exploitation o f scale econom ices and
flexibility, leading to high throu gh -pu t, efficient inventory m anagem ent, high
520 JA N F A G E R B E R G AN D M A N U E L M. G O D I N H O
Henry Ford allegedly once said that a customer could get a car in any color he wanted as
long as it was black. This was the quintessential logic o f the Am erican system o f
manufacturing, based on standardized products, produced in long series for mass
consumption, by low-skilled (often immigrant) labor, controlled by a hierarchy o f
foremen, engineers, and managers.
The attempt to adapt this system to Japanese conditions after World War II led to
important modifications. First, the Japanese market was much smaller, so critical mass
could only be reached through exploiting demand diversity. Second, the Japanese labor
force was well educated, trained, and culturally homogeneous, and the differences in
status and pay between blue and white collar workers small. As a result of these
differences, the production system that evolved in Japan came to look very different
from that o f the USA (Freeman 1 9 8 7 ).
The kanban or “just-in-time” system, developed by the Japanese auto industry,
combines the advantages of mass production with flexibility in adjusting to changes
in the composition and level of demand (Aoki 1 9 8 8 ). What is going to be produced
(and when) is decided by the part of the firm close to the end users (market). Orders
are placed on a daily basis at the firm's production units, which have to deliver the
requested products “just-in-time.” This also holds for suppliers of parts, and the
system is referred to as the “zero inventory” method. However, “zero inventory”
implies that defect parts cannot be tolerated (because otherwise production would
be halted). To increase quality and eliminate defects, organizational practices such as
“total quality control,” originally borrowed from US industry, and “ quality circles”
were introduced. Eventually, a new organization of work emerged, with workers
rotating through different tasks and a much greater role for the individual worker
(and work-team) in surveying production and quality, than what was common in the
US auto industry. This new organization of work also meant more competent, com
mitted, and motivated workers.
Important efficiency improvements stemmed from these organizational innov
ations. By the late 1 9 8 0 s, Japanese manufacturing, particularly in the car industry,
was unrivalled in its efficiency (Womack et a l 1990). The time needed to produce a car
in Japan in 1989 was 16.8 hours, while the equivalent figures for the US and Europe
were, respectively, 25.1 and 36.2 hours (The Economist, 17 October 1992).
IN N O VA TIO N AND CATCH IN G-UP 52!
F o r instance, w hile Jap an ese catch -u p w as largely self"fin an ced , the S o u th K orean
catch -u p cam e to d epend h eavily on fo reign len d in g. H ow ever, su ch increased
debt exp o su re, w h ile p ro v id in g o p p o rtu n ities fo r catch -u p , m a y also m ak e countries
vu ln erab le, as sh ow n b y the fin an cial crisis in K o rea (an d to so m e extent in
o th er A sia n cou n tries) tow ard s the end o f the 1990s (see B o x 19 .2). A lth o u gh
co n tro versial (Sh in an d C h an g 2003), the crisis m a y also be seen as an illu stration
o f the im p o rtan t p o in t that p o licies an d in stitu tion s that w o rk ed w ell d u rin g the
catch -u p phase m a y n ot be equ ally w ell suited w h en this ph ase is co m p leted and
the fo rm er catch -u p c o u n try has to co m p ete w ith o th er d evelo p ed co u n tries on an
equal fo o tin g. A n o th er exam p le o f this, also fro m the fin an cial sector, com es
fro m Jap an . T h e Jap an ese fin an cial system w as d esign ed to generate large savings
a m o n g the general p u b lic and to fu n n el these to the large in d u strial con glom erates,
w h ich w ere the van g u ard s o f the catch -u p process, on p referen tial term s, and
as such it w as v e ry effective. H ow ever, w h en the catch -u p w as co m p leted , this
fin an cial m ach in e co n tin u ed to generate large savin gs, even th o u g h the profit
o p p o rtu n ities created b y the p o ten tial fo r catch -u p w ere largely gon e. T h is led
to excesses, crises, and d ep ression . H ence, fro m b ein g a v e ry v a lu a b le asset,
the c o u n try ’s fin an cial system actu ally tu rn ed into a co n sid erab le b u rd e n for the
Jap an ese econom y.
19.2.3 A Macro-View
The third strand of catch-up research mentioned above operates on the macro-level
and asks questions of the extent to which the catch-up or convergence actually
occurred, for whom and how this may be explained. As mentioned in the introduc
tion, an important finding in this literature is that the long-run trend since the
British Industrial Revolution points to divergence, not convergence, among capital
ist economies. It has also been shown that these trends differ considerably between
time periods. For example, one such period in which the conditions for catch-up
appear to have been especially favorable (and during which many countries man
aged to narrow the gap in productivity and income vis-a-vis the leader) is comprised
of the decades following the end of World War II, what Abramovitz (1986,1994) has
called “the post-war catch-up and convergence boom.” He suggested that such
differences in performance over time and across countries might, to some extent,
be explained with the help of two concepts, technological congruence and social
capability. The first concept refers to the degree to which leader and follower country
characteristics are congruent in areas such as market size, factor supply, etc. For
instance, the technological system that emerged in the USA around the turn of the
century was highly dependent on access to a large, homogeneous market, something
that hardly existed in Europe at the time, which may help explain its slow diffusion
there. The second concept points to the various efforts and capabilities that develop
ing countries have to develop in order to catch-up, such as improving education,
infrastructure and, more generally, technological capabilities (R&D facilities etc.).
Abramovitz explained the successful catch up of Western Europe in relation to the
US in the first half of the post-war period as the result of both increasing techno
logical congruence and improved social capabilities. As an example of the former, he
mentioned the manner in which European economic integration led to the creation
of larger and more homogeneous markets in Europe, facilitating the transfer of
scale-intensive technologies initially developed for US conditions. Regarding the
latter, he pointed to, among other things, such factors as the general increases in
educational levels, the rise in the share of resources devoted to public and private
sector R&D, and the ability of the financial system in mobilizing resources for
change.
There have also been attempts to develop testable models of cross-country
differences in growth performance (or productivity) that includes the potential
for catch-up as one of the explanatory factors. Classical papers on the subject are
Nelson (1968) and Gomulka (1971). In a highly innovative contribution, Cornwall
(1977) analyzed economic growth in the first half of the post-World War II period as
driven by catching-up processes, the ability to mobilize resources for change (invest
ment), demand, and endogenous technological change (through the working of the
so-called “Verdoorn’s law” ). Baumol et al. (1989) presented and tested a model of
524 JAN F A G E R B E R G A N D M A N U E L M. G O D I N H O
cro ss-c o u n try grow th , including the potential for catch-up and social capability
(p ro xied b y e d u catio n ), fo r a large n u m b er o f co u n tries an d d iffe re n t tim e-span s,
and since then there has been a p leth ora o f su ch exercises c o n firm in g (o r q u estio n
ing) the im p o rta n ce o f such factors (see F agerb erg 1994 an d T em ple 1999 for
overviews)* H ow ever, m ost o f these studies have ig n o red Abramovitz’s em phasis
on the im p o rtan ce o f tech n o logical co n gru en ce as w ell as the role o f innovation*
M o re to the latter, Fagerb erg (19 8 7,19 8 8 ) has suggested an e m p irica l m o d e l b ased on
Sch u m p eterian lo gic that inclu des in n o va tio n , im itatio n , an d o th er effo rts related to
the co m m ercial exp lo itatio n o f tech n o lo g y as d riv in g forces o f growth* F o llo w in g
this ap p ro ach , catch -u p o r convergence is b y no m eans gu aran teed , as it dep en d s on
the balan ce o f in n o va tio n and im itatio n , h o w ch allen gin g these activities are, and the
extent to w h ich co u n tries are eq u ip p ed w ith the n ecessary cap ab ilities. A cco rd in g to
V erspagen (19 9 1), w h o im plem en ted sim ilar ideas into a n o n -lin e a r setting that
allow s fo r b o th catch -u p an d a “ lo w -g ro w th trap,” p o o r co u n tries w ith a lo w “ social
ca p a b ility ” are the ones at risk o f b ein g “ trap p ed ”
A b r a m o v itz s w o rk has been criticized b y Sh in (1996) fo r n o t b ein g sufficiently
h isto rica lly specific. In p articu lar, he argu es that the “ so cial c a p a b ility ” concept is
v e ry d ifficu lt to o p eratio n alize, a fact ad m itted b y A b ra m o v itz h im self. H ow ever,
A b ra m o v itz ’s em ph asis on tech n o logical co n gru en ce d e a r ly p o in ts to an awareness
o f the im p o rtan ce o f changes in tech n o logical d yn am ics o ver tim e, alth o u gh it is
clear that he h im se lf did little to su bstantiate it. In this he d e a r ly sided w ith
G ersch en kron , w h o also focused alm ost exclu sively o n ca tch -u p in scale-based
tech nologies. T h ere is, how ever, no scarcity o f c o n trib u tio n s that argu e that the
d yn am ics o f the scale-based system is w a n in g (N elso n an d W righ t 1992; Fagerb erg et
al. 1999)* I f so, this m a y have stron g im p lica tio n s fo r the c o n d itio n s fo r catch -u p . We
retu rn to this issue in the final section o f this chapter.
19.3 C a t c h i n g u p : A R e v i e w of
Recent Evidence
We w ill n o w take a closer lo o k at the glo bal catch in g -u p p ro cess (o r lack o f such)
d u rin g the fo u r last decades. W hile oth er studies have p ro v id e d h ig h ly aggregated
analyses o f differences in grow th across large sam p les o f co u n tries (fo r overview s, see
Fagerb erg 1994 an d Tem ple 1999), w e w ill in this section lim it the analysis to a
selection o f cou n tries that we find p a rticu la rly relevant fo r the stu d y o f such
processes, and fo r w hich g o o d data on relevant factors, such as R&D an d innovation,
are available. This includes the cou n tries discussed so far, such as the p re v io u s— and
IN N O V A T I O N A N D C A T C H IN G - UP 52.5
Table 19.1 Income groups, 1960- 1999 (GDP per capita, 10 3 $US, 1990 constant
PPPs)
Source: Calculations based on Angus Maddison/Groningen Growth and Development Centre and The
Conference Board, Total Economy Database, July 2003, http://www.ggdc.net.
So u th Korea
T a iw a n *
S in g a p o re *
Hong K ong*
M a la y s ia *
C h in a
Jap a n
Ireland
Po rtugal
S p a in
G reece
Finland
Italy
France
B ra zil*
United S ta te s
In d ia *
C h ile *
United K in g d o m
G e rm a n y**
M e x ic o
A rg e n tin a *
P h ilip p in e s*
1 2 3 4 5 6
Source: Calculations based on A n gus Maddison/Groningen Growth and Development Centre and The Conference Board,
Total Economy Database, July 2003, http://www.ggdc.net.
■ 1995
□ 1965
10 11 12
UK ~27j
Finland ---------- ------------ ----------- ------- ---- --------- ------------;______ ---------
—---- 30,01
Korea ■---------- 1-----------1----------- 1-----------,______ ! _ ____ L_ ~ T ,
36,4
1 1 1
Singapore 3 37,8
France 3 79,6
Japan 1 25,6
Taiwan 1 І3 .9
Germany 33,8
Ireland 26,7
Spain 3 19,
US 17,0
Italy 25 ,6
Portugal 17,2
Greece 23,5
Chile 1 26,3
Brazi
Thailand
Mexico
Argentina
China
India
Malaysia
Figure 19.3 Ratio of first university degrees in natural sciences and engineering to
24-year-olds in the population, 1999 (all values in °/o)
Notes:Alt figures for 1999 or most recent year. The numbers to the right of the horizontal bars refer to the percentage share
of "1st University Degrees in Natural Sciences and Engineering*' in "Total 1st University Degrees" (French and Greek
numbers refer only to "lo n g " degree courses, and are therefore not directly comparable to the numbers for other countries).
co u n tries d irect their ed u catio n al investm ent tow ard s typ es o f e d u c a tio n o f p a rtic u
lar im p o rta n ce fo r tech n o logical catch -u p (and in n o v a tio n ).
It sh o u ld be n oted , how ever, that there are so m e exam p les o f co u n tries that have
fallen b eh in d despite q u ite su bstan tial investm ents in h igh er e d u ca tio n , fo r exam ple,
in the presen t sam ple, A rg en tin a and the P h ilip p in es. A rg u a b ly , im p o rta n t as
ed u catio n is, w h at m atters fo r grow th in the lo n g ru n is h o w it is p u t in to use, and
the failu re to exp an d the em p lo ym en t o p p o rtu n ities fo r h ig h ly ed u cated la b o u r m ay
se rio u sly im p ed e the p o ten tial gro w th effects fro m in vestm en ts in h ig h e r education.
In fact, o n e o f the reasons the A sia n N IC s m an aged to e xp a n d h ig h er technical
ed u catio n so rap id ly w as the sim ilar rap id increase in e m p lo y m e n t o p p o rtu n itie s fo r
engineers (an d scientists). T h u s, fo r these co u n tries, in d u strial, tech n o logy, and
ed u catio n p o licies w ere co m p lem en ts, n ot substitutes, and the a b ility to c a rry out
these p o licies in a su stain ed and co o rd in ated fash io n p ro b a b ly exp lain s a go o d deal
o f th eir eco n o m ic success. Sim ilarly, attem pts to target h ig h -g ro w th , strategic in d u s
tries w ith o u t investing su fficien tly in c o m p le m e n tary assets, such as h ig h er ed u ca
tio n , o r w ith o u t p ro v id in g sufficient incentives fo r tech n o lo g ical u p g ra d in g (a
“ d y n a m ic ” co m p etitive en viro n m en t), are also b o u n d to fail, as the evidence o f
som e co u n tries in, fo r instance, Latin A m erica show s.
O ne im p o rta n t use o f h ig h ly com peten t la b o u r is, o f co u rse, in R & D . Figu re 19.4,
w h ich focuses o n R & D as a share o f G D P, show s that, in the early 1960s, o n ly a few o f
the co u n tries in o u r sam ple, w ith the U S A , the U K , and Fran ce in the lead, devoted a
sign ifican t share o f G D P to R & D activ ities.13 A p art fro m these three co u n tries, and
G e rm a n y and Ja p an , all co u n tries in o u r sam ple d evoted less than 1 p er cent o f G D P
to R & D . Today, the U S A has been replaced b y Ja p a n as the c o u n try that em p lo ys the
largest share o f its in com e o n R & D activities, and the clu b o f h ig h R & D p erfo rm ers
has been enlarged b y a n u m b er o f new m em b ers, w ith S o u th K o rea, Finland,
and T aiw an d eservin g o f p articu lar m en tion . H ow ever, S in g ap o re, Irelan d , and
Italy h ave also increased exp en d itu res o n R & D b ey o n d the 1 p er cent o f G D P level.
T h e rem a in in g cou n tries, in clu d in g those fro m L atin A m erica, m a n y A sian and
m o st o f the catch in g -u p econ o m ies in E u rop e, rem ain lo w R & D p erfo rm ers,
alth o u gh R & D investm ents h ave in several cases increased sig n ifica n tly co m p ared
to the situ ation a few decades ago. D ata on patents reveal a v e ry sim ilar pattern
(Figu re 19.5).
A n o th er in d icato r that is often in vok ed in analyses o f catch in g -u p an d tech n o logy
tran sfer is inw ard fo reign direct investm ents (F D I), on the gro u n d s that th ose w ho
do such investm ents are assu m ed to co n tro l, and are w illin g to share, su p erio r
tech nology. T h e available evidence, how ever, indicates that the d istrib u tio n o f FD I
is h ig h ly skew ed, w ith a d isp ro p o rtio n a tely h igh a m o u n t invested in tw o sm all
econ o m ies, H o n g K o n g and Sin gap ore, and, m o re recently, an d to a lesser extent,
In a n u m b er o f oth er lo w er in com e cou n tries, such as Irelan d , C h ile, M alaysia, and
C h in a. H ow ever, som e o f the m ost successful catch in g-u p eco n o m ies, such as Jap an ,
Taiwan, and Sou th K orea, have received very little in w ard F D L T h is d oes n o t im ply,
1 N N ОV А ТIО N a n d
С А Г с И I КГ G " U p
ofcourse, that these countries did not benefit from “ *e™ f* more efficient—wavs of
h just that they found other— and perhaps equ У
U n ite d S t a t e s
Japan
Taiwan*
G e rm a n y **
Finland
France
UK
South Korea
Hong Kong*
Italy
Singapore*
Ireland
*95-01
* 01*8-7
Spain
Greece
Malaysia*
Argentina*
Portugal
Mexico
Chile*
Brazil*
Philippines*
India*
China*
Source. U S P T O , P a te n t c o u n t s by c o u n t r y / s t a t e a n d y e a r —A ll p a t e n t s , a ll ty p e s,
h t tp :/ / w w w .u s p to .g o v / w e b / o f fic e s / a c / id o / o e ip / t a f / r e p o r ts .h tm
INNOVATION a n d c a t c h i n g "UP 533
Asian catch-up has benefited greatly from technology developed elsewhere, However,
the mechanisms used to tap foreign technology sources differ. One central mechanism!
used extensively by Singapore, is inward Foreign Direct Investment (FDI). By contrast!
Taiwan and especially South Korea relied mostly on a form o f subcontracting, “ Ori
ginal Equipment M anufacturing” (O EM ). As suggested by Hobday (2 0 0 0 ), OEM
might be seen as an organizational innovation, facilitating learning and technological
upgrading in latecomer firms.
Under an OEM contract, a product is produced according to a customer's specifica
tions, normally a transnational corporation (TN C ), that markets and sells the product
under its own brand-nam e (such as, for instance, “NIKE” or “ IBM ” ). From the 1 9 7 0 s
onwards, many US and Japanese firms, particularly in the ICT sector, used this
mechanism to contract out their production to Korean and Taiwanese firms. This
allowed the latter to acquire basic producing capabilities in electronics, since the TNCs
normally “ helped with the selection o f equipment; the training o f managers, engineers
and technicians; and advice on production, financing and management. . . . Local
learning was encouraged because the T N C depended on quality, delivery, and price o f
the final output” (H obday 2 0 0 0 ; 1 3 4 ).
As successful OEM arrangements evolved into closer long-term relationships, the
Korean and Taiwanese firm s gradually acquired more advanced capabilities, first in
process engineering and later in product design. This led OEM to evolve into a more
advanced stage, O DM (Own Design and Manufacturing), with a greater emphasis on
R&D. A next step, O BM (Own Brand Manufacturing), occurs if a firm uses the
acquired capabilities to produce and to market products under its own brand-name.
This requires new capabilities in marketing, and very substantial investments in
distribution. Hence, it is a difficult step, but potentially very rewarding, since a lot o f
the value added is generated at this stage. Several Korean and Taiwanese firms have
tried, with mixed success, although a few (e.g. Samsung) have managed quite well.
Others, such as Portugal, Spain, and Greece, have preferred to pursue catch-up
without sim ilarly am bitious goals for changing the industrial structure, and, argu
ably, with m ore modest results, both in terms o f econom ic perform ance, accum ula
tion o f skills, and technological capabilities. Still others, such as the Latin Am erican
countries considered here, have failed to invest sufficiently in skills and technological
capabilities, and have as a consequence fallen further behind.
19.4 C a t c h in g -u p a n d P o l ic y
A weakness o f m uch o f the existing discussion on catch-up and policy has been an
excessive focus on the policy level (government) at the expense o f the recipients o f
these policy initiatives, e.g. the firms o f the potential catching-up country. As
pointed out by Teece (2000:124): “ If firms are indeed the instrum ents o f develop
ment, the study o f econom ic development cannot take place separate from the study
o f the theory o f the growth o f the firm ." We suggest this as an im portant area for
further research, theoretical as well as applied. Although an extensive treatment o f
the role o f firm s in catching-up processes is beyond the scope o f this chapter, we will
nevertheless try to emphasize a few points that we believe m ay be useful for further
w ork.15 Research on the role o f firms in innovation and long-run econom ic change
com m only stress that, in most cases, firms only have im perfect knowledge o f the
relevant options in front o f them, and that they tend to be m yopic, searching in the
neighborhood o f their existing competence for relevant inform ation, suggestions,
and solutions (Nelson and Winter 1982; Dosi 1988; Fagerberg Ch. 1 and Lam Ch. 5,
both in this volum e). These characteristics are, o f course, com m on for developed
and developing country firms, but, being far from both the technology frontier and
the potential market, greatly accentuate these problems. Moreover, the developing
country firm m ay be, to a much larger extent than developed country firms,
constrained by its environment: it m ay have a wish (and perhaps even the capability)
to introduce a new product or process, but the possibility to do so m ay depend on
capabilities in other firms or skills that are sim ply not there (or require substantial
investments to occur). Arguably, to avoid being stuck along an inferior path and
never catch up, “ institutional instrum ents" m ay be needed to compensate for some
o f these “ latecomer disadvantages," to use a Gerschenkronian term. In particular,
what the developing country firm m ay need are “ institutional instrum ents" that
improve:
* links with the technology frontier,
* links with markets (and sophisticated users),
* supply o f needed skills, services and other inputs,
* the local innovation system/network.
Arguably, much o f what firms and governments in catching-up countries have done
can be understood from this perspective. For instance, the diversified business-
groups that developed in Japan and South Korea might be seen as “ institutional
instrum ents" fulfilling some o f those needs (Shin 1996). The O EM system (original
equipm ent manufacture, see Box 19.3) that has developed in the electronics indus
tries o f East Asia m ay also be seen as an “ institutional instrum ent"— or “ organiza
tional in n ovation ’ (H obday 2000)— geared towards sim ultaneously im proving
links with the technology frontier and the market. Similarly, attracting inward FDI
may be seen as a “ functional equivalent" to OEM , which, however, judged by the
empirical evidence, seems to be less favourable for indigenous innovation. Other,
more demanding, but perhaps also more rewarding ways— since it allows the
INNOVATION AND C ATC H IN G -U P 537
latecomer firm to reap a larger share of the profit generated— include technology
licensing, investments in own brands (OBM), etc. Improving the supply of needed
skills has, of course, been a central preoccupation of many latecomer governments,
as illustrated in Section 19.3 above. Moreover, we have witnessed sustained efforts by
several latecomer governments in accommodating the needs of firms for a high
quality RM ) infrastructure (innovation system).
What can the extraordinary success that some catching-up countries have had,
and the failure of others, teach present-day developing countries? One important
lesson is that there is no one unique way to successful catch-up that every country
has to emulate. Every country has to find its own way based on an understanding of
(a) the contemporary global technological, institutional, and economic dynamics,
(b) the behavior (and needs) of the relevant agents (of which the firm arguably is the
most important), and (c) the specific context in which the catch-up takes place and
the broader factors that influence it, being economic, technological, institutional,
political, or cultural (Freeman and Lou<;a 2001). There are, one may suggest, big
potential rewards in following the Gerschenkronian strategy of targeting techno
logically progressive sectors, as part of a broader attempt to transform the economy,
stimulate learning, and enhance the creation of new skills (or assets). However, not
every country is equipped with the necessary capabilities to pursue such a strategy.
For instance, when, in the mid-nineteenth century, Japan initiated its efforts to catch
up with the West, the technology gap vis-a-vis the more advanced countries was
much smaller (compared to what developing countries face today), and its popula
tion already had a level o f education that compared favourably with most other
countries at the time (Odagiro and Goto 1996). Given that educational standards
have been rising ever since, investing in education may be a good place to start for
countries that have not succeeded in catering to those needs already. For those that
have, there is a wider set of options, and for them some of the experiences outlined in
this chapter may be highly relevant (see Box 194).
Ireland and Portugal are two small countries on the western periphery of Europe. Both
failed, for various reasons, to exploit fully the possibilities for catch-up and industri
alization during the first half of the twentieth century, and were, therefore, at the
beginning of our period of study, among the poorest in Europe, their economies
Characterized by an industrial structure dominated by traditional, low-skill activities.
However, during the second half of the century, both countries took steps to
integrate their economies with the more dynamic economies of Western Europe, first
through membership of EFTA (1961), and later by joining the EU (1981 and 1986 ,
respectively). As members of the EU, Ireland and Portugal received substantial
economic support through the so-called “ Structural Funds, designed to induce
53$ JAN F A G E R B E R G A N D M A N U E L M. G O D I N H O
N o tes
1. We thank Fulvio Castellacci, Sandro Mendonca and the authors and editors of this
volume for helpful comments and suggestions, retaining sole responsibility for remaining
errors and omissions.
2. More recently, economists have formulated theories (so called “ new growth theory” ) that
do not necessarily predict convergence. See Fagerberg (1994, 2000) and Verspagen, Ch. 18
in this volume, for an extended account.
3* I w instance, although there is a sizeable literature on US economic growth, most of it is
not written from a catch-up perspective, e.g. focusing on what the US experience may tell
us about what other countries should do (or not do) to succeed in catching up. See,
IN NOVATION AND CATCH IN G -U P 539
however, the recent book by Chang (2002) and the discussion in Section 19.4 of this
chapter,
4. Veblen mentions factors such as the “funds available for investment” ! Veblen 1915:186), a
sufficient supply of “educated men” (ibid. 194 ), as well as a “sufficiently well-instructed
force of operative workmen” (192). The latter, he noted, did not have to be particularly
well educated or trained (188).
5* To the extent that industrialization took place, it was largely by the application o f the
most modern and efficient techniques that backward countries could hope to achieve
success, particularly if their industrialization proceeded in the face of competition from
the advanced country” (Gerschenkron 1962; 9).
6. Surprisingly, perhaps, he did not (nor did Veblen) put much emphasis on the achieve
ments made by Germany in other areas, such as the educational sector, and in pioneering
the development of an R&D infrastructure.
7. See Gerschenkron (1962:16-20),
8. For instance, as a result of these efforts, illiteracy—which was relatively low by inter
national standards even before the Meiji-restoration—was almost eliminated among
Japanese youth by the turn of the century, and by 1920, more than half of the children
graduating from elementary school proceeded to secondary schools (Odagiri and Goto
1996).
9. Productivity (and productivity growth) remained low, however, in sheltered industries
(agriculture and services), and this imposed growing burdens on the economy during
the 1990s and the early twenty-first century.
10. See Johnson (1982) on Japan, Amsden (1989) on Korea and Wade (1990) on Taiwan.
11. Most countries in our sample have between 80 per cent and 100 per cent of their youth
attending secondary schooling. However, some (but not all, Malaysia, for instance, has
93 per cent) of the poorer economies have less. For instance, Argentina and Chile were
reported to be in the 70-75 per cent range, and China and Mexico lower still, between
50 per cent and 60 per cent, Brazil is reportedly very low, only 15 per cent. Note,
however, that the data for Argentina and Brazil quoted here were from the 1985-7 period,
while for the other countries it was 1998 (source UNDP, Human Development Report
2002).
12. UNESCO defines the “gross enrolment ratio in tertiary education,” which is the indica
tor used in Figure 19.2, as “total enrolment in tertiary education, regardless of age,
expressed as a percentage of the population of the five year age group following on from
the secondary-school leaving age.” If many tertiary students are older than this, as may be
the case in several advanced countries (with a high emphasis on “long” degree courses),
this indicator may give a too optimistic view on the actual share of an age group enrolled
in tertiary education. Comparisons with other available indicators of tertiary education,
based on educational attainment, indicates that such a bias may be present, and more so
for the USA than other countries. Hence, the figure probably exaggerates the difference
between the USA and other countries in the emphasis on tertiary education.
U. Note that the data reported here refer to total R&D, including the public part. If we had
focused on only the part undertaken (and/or financed) by the business sector, the
ranking of the countries would have been approximately the same, but the differences
between the top and the bottom would have increased (in general, the more a country
invests in R&D, the higher the share financed by business).
40 JAN F A G E R B E R G AND M A N U E L M . GODINHO
Ie f e r e n c e s
Foundation Press, 299-ЗЗО, 350-4 (repr. as ch. 6 in J. Fagerberg (2002), Technology, Growth
and Com petitiveness: Selected Essays, Cheltenham: Edward Elgar).
----- G uerrieri, P., and V erspagen , B. (eds.) (1999), The Econom ic C hallenge fo r Europe:
Adapting to In n ovation B ased G row th , Cheltenham: Edward Elgar.
and V erspagen , B. (2002), lechnology-Gaps, Innovation-Diffusion and Transform
ation: An Evolutionary Interpretation” Research P o lic y 31:1291-304.
* F reem an , C. (1987)» Technology Policy a n d Econom ic Perform ance : Lessons fro m Ja p a n ,
London: Pinter.
------ and Lou<pA, F. (2001), As Tim es Goes By. From the Industrial Revolutions to the Infor
mation Revolution, Oxford: Oxford University Press.
----- and Soete , L. (1997). The Econom ics o f In du strial Inn ovation , 3rd edn., London: Pinter.
*Gerschenkron , A. (1962), Econom ic Backw ardness in H istorical Perspective, Cambridge,
Mass.: Belknap Press.
Godin но, M. (2000), “ Desen.volvimento competitive do sector de cal^ado em Portugal:
Li^oes de um caso de clu sterin g? in I. Salavisa Lan^a (ed.), Trajectories com petitivas na
industria portuguesa, Oeiras: Celta.
Gomulka, $. (1971), “ Inventive Activity, Diffusion and the Stages of Economic Growth,”
Report No. 24, Institute of Economics, Aarhus University, Aarhus.
Granstrand, O. (1999), T he Econom ics o f M an agem en t a n d o f Intellectual Property, Chelten
ham: Edward Elgar.
Hobday, M, (2000), “ East versus Southeast Asian Innovation Systems: Comparing OEM-
and TNC-led Growth in Electronics,” in L. Kim and R. Nelson (eds.), Technology Learning
& Innovation: Experiences o f N ew ly Industrializing Econom ies , Cambridge: Cambridge
University Press, 129-69.
J ohnson, C. A, (1982), MITI a n d the Japa nese M iracle: The Growth o f Industrial Policy ;
1925-1975, Stanford: Stanford University Press.
*Lall, S. (2000), “Technological Change and Industrialization in the Asian Newly Industri
alizing Economies: Achievements and Challenges,” in L. Kim and R. Nelson (eds.),
Technology, L ea rn in g & In n ovation : Experiences o f N ew ly Industrializing Economies, Cam
bridge: Cambridge University Press, 13-68.
Landes, D. (1998), T he W ealth a n d Poverty o f Nations, London: Abacus.
List , F. (1885), The N a tio n a l System o f Political Economy, trans. from the original German edn.
(1841), London: Longmans, Green and Company.
Nelson, R. R. (1968), “A ‘Diffusion Model of International Productivity Differences in the
Manufacturing Industry,” A m erican Econom ic R ev iew 58:1219-48.
-----and W inter , S. G. (1982), A n E vo lu tio n a ry Theory o f Econom ic Change, Cambridge,
Mass,: Harvard University Press,
* --- and W right , G. (1992), “The Rise and Fall of American Technological Leadership:
The Postwar Era in Historical Perspective,” Jo u rn a l o f Econom ic Literature 30:1931-64,
Odagiro, H„ and G oto, A. (1996), Technology a n d In du strial D evelopm ent in Ja p a n , Oxford:
Clarendon Press.
O'Sullivan , M. (2002), “Industrial Development: A New Beginning?” in John O'Hagan
(ed.), The E con om y o f Irelan d : Policy a n d Perform ance o f a European Region , Gill and
Macmillan, 260-85.
Rodrik, D, (1994), “ King Kong Meets Godzilla: The World Bank and the East Asian Miracle,”
in A. Fishlow et al. (ed.), Miracle or Design? Lessons from the East Asian Experience, Policy
Essay No. 11, Washington, DC: Overseas Development Council, 13- 53-
>4* JA N F A G E R B E R G A N D M A N U E L M. G O D I N H O
*Sh in , J ang -S up (1996)5 The Economics o f the Latecomers: Catching-up, Technology Transfer
and Institutions in Germany, Japan and South Korea, London: Routledge.
----- and Ha -J oon C hang (2003), Restructuring Korea In c , London: Routledge.
>o l o w >R. M. (1956), “A Contribution to the Theory ofEconom ic Growth,” Q uarterly Journal
o f Economics 70: 65-94.
Femple , J. (1999)5 “ The New Growth Evidence,” Journal o f Economic Literature 37:112-56 .
Feece , D. J. (2000), “ Firm Capabilities and Economic Development: Implications for the
Newly Industrializing Economies,” in L. Kim and R. Nelson (eds.), Technology, Learning &
Innovation: Experiences o f Newly Industrializing Economies, Cambridge: Cambridge Uni
versity Press, 105-28.
^eblen , T. (1915), Im perial Germ any and the Industrial Revolution, New York: Macmillan.
/erspagen , B. (1991), “A New Empirical Approach to Catching Up or Falling Behind,”
Structural Change and Economic Dynamics 2:359-80.
W ade , R. (1990), Governing the M arket: Economic Theory and the Role o f Governm ent in East
Asian Industrialization, Princeton: Princeton University Press.
Vomack , J., J ones , D. T., and R oos, D. (1991), The M achine that Changed the World,
Cambridge, Mass.: M IT Press.
Vorld B ank (1993), The East Asian M iracle: Economic Growth and Public Policy, New York:
Oxford University Press.
C H A P T E R 20
INNOVATION AND
COMPETITIVENESS
20.1 In tr o d u c t io n
Traditionally , econom ists and econom ic historians since Adam Smith have
discussed econom ic growth principally in the context o f the national level— why
some countries grow faster (in modern terms, acquire the capabilities for sustained
growth that make them more competitive) and so become wealthier than others.
While in neoclassical economics questions o f national competitiveness came to
assume a lesser degree o f importance, as attention was shifted away from issues of
growth towards those o f static resource allocation and efficiency, there was even less
concern with the notion o f competitiveness at the firm level. The theory of the
(comparative) growth o f the firm was a minority interest o f those such as Downie
(1958), Penrose (1959), and Marris (1964), typically treated as a rather esoteric sub
branch of industrial economics, that was to be accorded a lesser status in the
discipline than the conventional theory of the firm (which was really a theory of
the relationship between the firm and markets). In recent years two related changes
in economics and allied areas o f research have been under way. One is a revival of a
more widespread interest in the classical issues of competitiveness at a national level,
and the other is the growing attention now paid to competitiveness at the level o f
industries, regions, and firms, in which fields o f research a substantial new literature
544 JOHN C A N T W E L L
has emerged. In Section 20.2 the contribution is assessed o f the new literature on
com petitiveness across countries. Section 20.3 exam ines innovation and com peti
tiveness at the industry level that connects together firm s and their environm ent,
and Section 20.4 looks at the regional and firm level. Section 20.5 draws some
conclusions with respect to the interaction between innovative actors, between the
different levels o f analysis o f competitiveness, and opportunities for future research.
C o m p etitiven ess is here taken to mean the possession o f the capabilities needed
for sustained econom ic growth in an internationally com petitive selection environ
ment, in w h ich e n viro n m en t there are oth ers (cou n tries, clu sters, o r in d iv id u a l
firms, depending upon the level o f analysis) that have an equivalent but differen
tiated set o f capabilities o f their own. The term competitiveness is also sometimes
taken to necessarily im ply as a result a continuing rise in the living standards o f the
individuals that are members o f a social group with the required capabilities
(notably in this context, to im ply a sustained increase in the living standards o f
the citizens o f the country that is suitably competitive in world markets— see Tyson
1992). W hile it is indeed necessarily true that productivity growth increases incomes
on average (i.e. per capita income), it m ay well be that the process o f capability
generation and growth also has a disruptive effect on the distribution o f incomes.
This issue is not addressed directly here, since the way in which innovation affects the
em ploym ent opportunities o f individuals, which is a m ajor influence upon their
respective earning capacities, is the subject o f Pianta (Ch. 21 in this volum e).
The winners from innovation are those that construct appropriate capabilities,
but capabilities are localized and nationally differentiated (as explained by Edquist,
Ch. 7 this volum e), and so there can be m any successful players in the competitive
game, each to some extent learning from and interacting with the somewhat
alternative paths to capability creation being taken by others. Put in these terms
few could object that the pursuit o f competitiveness through innovation is a
laudable objective o f national policy, and indeed an increasingly im portant objective
as the role o f innovation has risen in the modern knowledge-driven economy, even
for (actually especially for) countries that start behind and wish to catch up (Fager-
berg and Godinho, Ch. 19 in this volume).
To be meaningful, competitiveness must be thought o f as entailing a relative
com parison o f growth rates or benchm arking o f perform ance to assess how well
each participant has done in developing the capabilities for innovation and growth,
and not be about the mutual potential for dam aging one another (a misleading
interpretation o f competitiveness criticized by Krugm an 1994a, 1996). It is reason
able to expect that, at least on average, the spillover benefits for others o f a good
perform ance in one location or by one agent tend to outweigh the costs o f that good
performance for others. This argument is largely applicable whether the unit o f
analysis is countries in the world econom y or firms in an industry. At a country level
the efforts o f each national system o f innovation to prom ote the competitiveness o f
businesses sited locally are increasingly com plem entary as scientific and engineering
__ INNOVATION a n d c o m p e t i t i v e n e s s 545
20.2 C o m p e t it iv e n e ss at the
N a tio n a l L evel
When looking at the country level, competitiveness is about the way in which the
pattern of international trade evolves over time to reflect changing patterns of
capabilities and hence competitive advantage (what might be thought of as the
evolution in the comparative advantage of countries), rather than about the estab
lished pattern o f comparative advantage which is the usual focus of trade theory.
While the earliest theories o f trade and growth can be traced back to the classical
economists, such dynamic accounts of the paths of international trade and invest
ment were revived in recent times by the technology gap approach (Posner 1961) and
the product cycle model (Vernon 1966). However, a major shortcoming of the
product cycle model was its reliance upon an overly simplistic demand-driven
theory o f innovation (which reflected the spirit o f the 1960s, when it was devised),
through which the firm was assimilated to the product, and innovation was sup
posed to be concentrated in just one leading country— the US (see Cantwell 1989, for
a further discussion o f the m odel). Sadly, when the product cycle model broke down
in the 1970s, in large part ow ing to the reemergence o f multiple centers for innov
ation in a num ber o f international industries, the amended versions o f the model
(Vernon 1974, 1979) focused upon considerations o f oligopolistic strategy rather
than revisiting the underlying theory o f innovation and competitiveness. It was only
in the 1980s that scholars based at Sussex once again wedded an analysis o f structural
shifts over time in the pattern o f international trade to a more realistic approach to
innovation: see Soete 1981; Dosi and Soete 1988; Dosi, Pavitt, and Soete 1990; and
Fagerbergs 1987 paper on structural changes in international trade (repr. as ch. 7 in
Fagerberg 2002).
Part o f the inspiration for Fagerberg’s research had been that economists some
times use the term “ com petitiveness” in various different ways, and especially in
macroeconomic policy discussions not always in the way that has been defined here.
T h is ch ap ter is co n cern ed w ith in n o va tio n an d co m p etitiven ess, an d th is is so m e
tim es d istin gu ish ed as b ein g ab o u t lo n g er-term tech n o lo g ical co m p etitiven ess, as
o p p o sed to sh o rter-term p rice com petitiven ess. T h ere are tw o d ifferen t w ays o f
d iscu ssin g co m p etitiven ess in the latter sense o f sh o rter-te rm p rice co m p etitiven ess.
In the context o f co n ven tio n al d em an d m an agem en t p o lic y d iscu ssio n s, i f (say)
lo w er g o vern m en t b o rro w in g m eans a fall in interest rates an d so a rise in net
o u tw ard investm ent, and i f this leads to a decline in the v alu e o f the d o m estic
currency, then the price "'competitiveness” o f dom estically produced goods and
services can be said to have increased, as export prices fall in foreign currency terms
while im port prices rise in domestic currency terms. However, this type o f com peti
tiveness is unlikely to be sustainable, especially if (for example) the rise in im port
prices sparks o ff domestic inflation, or if lower net inward investment has adverse
consequences for domestic productivity growth. The second and for our purposes
m ore substantive context is the conventional cost-based account o f competitiveness,
in which a fall in relative unit labor costs means lower prices (or a lower rate of
inflation), which in turn leads to a rise in exports and fall in im ports, and so an
increase in the value o f the domestic currency.
Longer-term technological competitiveness is more akin to the second o f these two
versions o f price competitiveness, in supposing that a faster growth o f (output and)
exports drives up the domestic currency, rather than it being a falling currency that
promotes net exports. In the context o f what is sometimes termed “ non-price”
competitiveness to distinguish it more clearly from the kind o f cost-based competi
tiveness just referred to, innovation and new lines o f value creation m ay mean higher
average prices as an indicator o f higher quality, but in any event they lead to a faster
growth o f productivity and trade, and thus an upward trend in the value o f the
domestic currency. Given what has been said already, it is worth stressing here that
in this perspective the rise in the value o f the currency is simply the reflection o f
competitiveness, defined as a relatively rapid growth in productivity and the value o f
(output and) exports. The rise in the value o f the domestic currency is not itself the
achievement o f competitiveness (an improvement in the terms o f trade that is
essentially a potential side effect resulting from competitiveness). It is also worth
making explicit that the departures from comparative advantage associated with trade
imbalances are merely a temporary result o f competitiveness in this framework, and
again not in themselves the objective o f competitiveness. What is implicitly supposed
here is that faster productivity growth is associated with a rising share o f w o rld trade,
and that in this process the growth o f exports leads the growth o f imports. So net
exports rise until imports catch up, and this catching up is facilitated by the conse
quent rise in the value o f the domestic currency and in domestic wage rates.
Now neo-Schumpeterian approaches to international competitiveness focus on
this kind o f process o f forging technological competitiveness, which for those whose
innovative efforts are most successful implies a sustainable increase in the share o f
world trade (or at the firm level, a sustainable increase in the share o f the relevant
______ ______ in n o v a t io n and c o m pe t it iv en e ss 547
world market). However, as has been discussed at length already, in the Schumpe
terian perspective competition entails the positive sum game of establishing new
spheres of value creation, so innovations expand the overall magnitude of world
trade and the world market. Those that contribute most to this process of expansion
see their shares rise as they are responsible for more of the new element of value
creation, and not because o f a substitution effect within some fixed total level of
world trade or some fixed and given world market (or even within some steadily
exogenously growing world market). In this context, the neo-Schumpeterian analy
sis of innovation and competitiveness is unlike equilibrium growth accounts, even
when those accounts incorporate an acknowledgement of research activity, if invest
ment in innovation is treated as being inherently like investment in any other
economic activity, and if the only difference between activities is treated as lying in
their wider impact on other activities through externalities. Instead, in the neo-
Schumpeterian story the very nature and purpose of innovative activity is to disturb
and add to the existing circular flow of income generation, in an experimental and
non-equilibrium fashion.
Such neo-Schumpeterian models of innovation and growth might be specified in
at least two alternative ways. The first of these leans heavily on the distinction just
drawn between shorter-term price competitiveness and longer-term non-price
technological competitiveness. In Fagerbergs (1987,1988) technology gap formula
tion of international competitiveness across countries, the impact on growth of
national rates of innovation and distance behind the technology leader are treated
primarily as additive elements, to be added on to the more traditional determinants
of economic growth in the form o f capital accumulation (the share of investment in
national output) and relative unit labor costs. The origins of viewing cross-country
growth in this kind o f additive framework can be traced to Abramowitz (1956),
Solow (1957), and Denison (1967), for whom technological improvements (and the
productivity growth to which they led) were an obvious means of accounting for the
substantial “residual” in variations in growth that remained after allowing for the
effect of the increase in factor inputs. So in this context capital accumulation proxies
for the extension o f the scale o f established activities, relative unit labor costs capture
cost-based “price” competitiveness, while the contribution of corporate research
and the capacity to catch up through imitating the achievements of a leader
represent “non-price” technological competitiveness.
Setting the p ro b le m u p in th is w a y is con ven ien t, as the em p irical evidence then
generally suggests th at te ch n o lo g ic a l com p etitiven ess is m ore im p o rtan t than the
more co m m o n ly c o n sid ered tra d itio n a l in flu en ces u p o n com petitiveness. Techno
logical com p etitiven ess is ju d g e d to b e m o re sign ifican t than relative u n it lab o r costs,
although Krugm an (1994b) an d Y o u n g (1995) p o in t to the co n tin u in g im p o rtan ce o f
capital accu m u latio n w ith in th is k in d o f fram ew o rk . T h e evidence for three conn
tries— Japan , the U K , an d th e U S — o ver the p erio d 19 6 0 -7 3 illustrated in
Table 20.1. B ased o n th e e stim atio n o f his em p irical m o d el o f intern ational
ТШ еЛІ The decomposition of the predicted growth in national market shares
from an estimated empirical model of cross-country competitiveness,
for 1 9 6 1 - 7 3 ^
Japan UK USA
Table 20.2 Actual growth rates achieved by countries, 1960-89, over and above
that predicted by ( in te r a lio ) their rates of capital accumulation
Portugal 23.7 -0 .0 0 2
Jamaica 25,0 -0 .0 3 7
out as m anaging to achieve growth rates well in excess o f what m ight have been
predicted from their favorable rates o f capital accum ulation alone. What was
different in these economies was their greater ability to innovate» to upgrade and
restructure their indigenous industries» and to learn and absorb m ore effectively
from foreign technologies. Capital accum ulation can em body innovation to the
extent that it is linked to the transform ation o f the productive activities being
conducted.
So an alternative approach also in the Schumpeterian tradition is to treat techno
logical accum ulation and capital accum ulation as sim ply aspects o f a common
process, rather than as independent (even if com plem entary) contributions to
growth. In this case innovation can be seen as driving up profitability and hence
lowering the share o f wages in output (even though wages are rising faster, and so
m ay be unit labor costs), which leads to a higher share o f investment in output, and
so higher capital accum ulation and growth as a result o f higher technological
accum ulation (Cantwell 1989, 1992). The basic idea here is that in fast-growing
countries just as an increase in imports tends to follow an increase in exports with
a lag, so wages tend to follow productivity increases with a lag, enabling innovation
to create a fresh source o f profitability and growth. Yet this also suggests that
technological competitiveness is in part cost-based. It should be noted, though,
that labor productivity is defined here sim ply as the value o f output per worker
employed, which implies that productivity growth is as m uch attributable to
product quality improvements (that raise the value or unit price o f output, as
stressed in Fagerbergs approach), as it is to the cost reductions associated with
process improvements. In this alternative neo-Schum peterian form ulation we need
w orry less about the distinction between embodied and disem bodied technological
change, or the distinction between improvements in product quality and delivery
times as opposed to improvements in processes that are reflected in costs and prices.
The renewed interest in international competitiveness and variations in growth
rates has spawned a substantial new literature on cross-country convergence or
catching-up versus divergence or falling-behind (see e.g. Baum ol, Nelson, and Wolff
1994). The evidence suggests that whether one observes convergence or divergence
depends upon the period studied and the countries selected. In any case, the overall
trend in cross-country variance at a world level m ay not be the m ost im portant issue.
Rather than trying to work out whether East Asian convergence statistically out
weighed the effect o f African divergence in aggregate, the issue is m ore why and how
firms in East Asia had the capabilities to catch up in the period since i960» while
those in Africa did not. The concepts o f a techno-socio-econom ic paradigm (Free
man and Perez 1988), or o f an evolution in the institutional characteristics of
capitalism (Lazonick 1991» 1992), can be useful in this respect as a means o f explain
ing occasional shifts in technological leadership or longer term competitiveness, and
in the direction o f those shifts. Emphasizing again the role o f structural change in
growth, and in particular during periods o f paradigm change, when the prevailing
I N N O V A T I О N A N D C OMP E T I T I V E N E S S 551
20.3 C o m p e t it iv e n e s s at th e In d u st r y
Lev el: th e N exu s of R ela tio n sh ips
b e t w e en F ir m s and t h e i r E n v iro n m en t
o f the debate has in mind a rather simple model o f that relationship (or lack o f a
relationship) which may apply at some places and at certain times, but cannot be
universally applied in the way that some advocates seem to imagine.
The relationships that exist between the development o f the technological cap-
abilities in firms that are responsible for competitiveness and the institutions o f the
wider society vary from one country to another, but in particular they tend to be
different in countries that belong to an already leading industrialized group and
those that are attempting to catch up with them (see also Fagerberg and Godinho,
Ch. 19 in this volum e). It is noticeable that there have been a greater num ber o f cases
in which governments in catching-up economies, partly through measures o f
domestic protection, have contributed m ore actively to the fostering o f capabilities
in local infant industries and in indigenous companies. This was true o f the US and
Germ any when they were catching up with Britain in the nineteenth century (Landes
1969), it was true o f Japan when it was catching up with the West during the
twentieth century (Ozawa 1974), and it was true o f Korea when it was catching
up after i960 (Enos and Park 1988). It is true that there are occasionally other cases
o f catching-up economies, like those o f Singapore or M exico in recent years, that
have taken advantage instead o f various aspects o f trade liberalization. However,
what is m ost noticeable in all these instances o f successful catching-up is that the
trade policies o f governments were merely part o f a much wider package o f support
for the longer term nourishment o f capabilities in indigenous firms. Since the
emergence o f science-based industries towards the end o f the nineteenth century
this has meant especially investing in science and higher education, in the training o f
engineers, and in the learning o f skills more widely (Freeman and Lou^a 2001).
Equally important, where there were measures o f trade protection local firms
accepted their part o f the bargain to invest very substantially in capability creation
in an outward-looking and export-oriented fashion, rather than sim ply remaining
an inefficient enclave as in so many other cases o f protectionism or so-called import-
substituting industrialization.
O f course, the institutional structures o f catching-up economies changed mark
edly (and any protectionist measures were largely reversed) as their firm s caught up
and themselves sometimes forged ahead and became innovative leaders in their own
right. This is perhaps the most vivid illustration o f the general observation that the
development o f technological capabilities in firms and the character o f the insti
tutions that support these competitive efforts in the wider society tend to coevolve
with one another (Nelson 1995), through a process of continual interaction. Another
perspective on these interrelated national systems for the construction o f competi
tiveness is offered by Porter (1990), as represented through the four com ers o f a
diam ond o f factor conditions; demand conditions; related and supporting indus
tries; and firm strategy, structure, and rivalry, In Porter s view the capacity o f firms to
innovate depends ciitically on having sufficient domestic rivalry in their own home
country o f origin, but also on the presence o f spillovers between firms associated
I N N O V ATION AND C O M P E T I T I V E N E S S 55^
with localized clusters (to which issue we return below). In other words, innovation
requires an appropriate mix o f interfirm rivalry and cooperation or exchange
(Richardson 1972)- Lazonick (1993) has argued that when confronted with a major
new competitive challenge from some new source of innovation from outside,
domestic industries may need to shift this balance away from rivalry and towards
cooperation in order to respond effectively. To express this another way in the light
of the trend towards globalization mentioned earlier, it may be that firms in some
national industry may need to increasingly collectively focus their efforts in what
they do locally (as opposed to activities they may locate abroad) to be better
mutually aligned with whatever may be the fields of particular local excellence or
of specialization in innovation. This would have the effect of tending to reinforce
national patterns of comparative advantage in innovation.
As has been mentioned already, with the emergence of science-based industries
over a century ago, the need for an infrastructure that suitably supports relevant
education, skill formation, and training became critical to the competitiveness of
industries, and is widely believed to have become more important still in the modem
techno-socio-economic paradigm associated with computerization and informa
tion processing. For firms to be able to create capabilities requires costly and difficult
internal learning processes, but these in their turn depend upon having suitable
organizational and technical skills in the management and workforce on which they
rely. The composition o f skills in the workforce of the home base of firms is therefore
critical to the success or failure of countries that are trying to catch up, but it also
becomes a central influence upon the fields in which any national group of firms has
its specific pattern of comparative advantage in innovation and capability creation.
Of course, this is not just a one-way street, since the types of investments and
commitments to training that are made by firms themselves in the course of
learning, the professional associations they help to form, and the pressures they
place upon governments and others imply again a process of coevolution between
firms and their environment in this respect too.
Table 20.3 helps to illustrate the significance o f education and skills in the
catching-up of the four East Asian tiger economies (see also Fagerberg and Godinho,
Ch. 19 in this volume). Korea stands out as having surpassed even the commitment
of the traditional industrialized countries to higher education in the natural sciences
and mathematics. Yet a key to the success of these countries as a group lies more in
the investments they have undertaken in support of engineering graduates while
Hong Kong lies behind the industrialized group (and this may help to account for
wby iis local learning and upgrading has been more limited than in the other three,
as discussed by Tail 2001), Singapore is above the industrialized country average, and
Korea and Taiwan are way ahead o f that average for tertiary level engineering
enrolments as a proportion of the population. Other developing countries are
generally well behind the achievements in engineering education of the tiger econ-
omies, although the Philippines, Argentina, and Mexico have at least matched the
554 J O H N C A NT WE L L
position o f Hong Kong. Considering the enorm ous size o f its population, it is also
clear that China has been catching up fast in this area.
T h e Japan ese and G e rm an system s are k n o w n to fo cu s o n a b ro a d an d deep skill
base by em p h asizin g the acq u isitio n o f general en gin eerin g skills a n d the good
stand ard o f basic edu cation o f the p o p u latio n as a w h ole, w h ile the U S and U K
system s tend to be m o re elitist an d fo cu s o n the d evelo p m en t o f h igh er-en d skills
o ver a n arro w er range o f p eople (Prais 1995; L azo n ick and O ’S u llivan 1997; Lazonick
1998). I his helps to explain w hy the Jap an ese an d G e rm a n fields o f co m parative
advantage in in n o vatio n (and th u s, o f co m p etitive ad van tage) in clu d e m o to r
vehicles and engineerin g that rely (and have in creasin g ly co m e to rely) o n a broad
innovation and competitiveness 555
skill base, while the US and UK advantages include aerospace, software, pharma
ceuticals, biotechnology, and medical equipment that rely on very highly skilled
individuals and an intensive R&D effort.
When examining the extent of path-dependency in the specific technological
traditions of national groups of large firms, and in their patterns of technologi
cal specialization as a measure of their relative contributions to each of the major
international industries (the cross-sectoral pattern of their technological competi
tiveness), some cross-border interdependencies appear to emerge. This raises again
the issue of whether or not there have been any elements of convergence across
countries, but in this context in the specific mix of strengths and weaknesses in
international industries, rather than in aggregate productivity or performance.
Examining patterns of technological specialization among national groups of the
largest firms from six countries (the US, Germany, the UK, France, Switzerland, and
Sweden) based on their patterns of corporate patenting, it has been observed that
these profiles are path-dependent and tend to persist to some extent even over
periods of sixty years, from the interwar period to the present day (Cantwell
2000). This may be taken to imply that the positive effect on the continuity of
collective technological trajectories of intercompany technological cooperation and
spillovers within national groups has tended to outweigh the negative effect of
mobility in cross-company distributions of activity. There is some evidence that
through the evolution in these patterns of technological competence that has
occurred, certain national groups have come somewhat closer to one another than
they were in the past, or they have changed in similar ways. Indeed, it might be
argued that the six national groups examined can now be divided into three clusters
of two countries each.
The first cluster comprises the largest US and UK firms, in which the profile of
technological competence can be characterized as being resource-based, oil-related,
and defence-related. It is increasingly also health-related. As shown in Table 20.4, in
the US case since the interwar period a continuing comparative advantage in
innovative activity in the largest industrial firms has been sustained in the oil,
food products, rubber products, aerospace (defence and larger-scale transport
systems) and building materials industries. The greatest continuing strengths for
the largest British companies over the same historical period has been in textiles,
other transport (defense) and oil since the 1930s. Thus, it can be argued that there has
been some convergence in the profiles of the US and UK innovation systems
(Vertova 1998). UK firms have also seen a post-war shift into technological compe
tence in the pharmaceutical industry, although it can be claimed that this too
represents the revival of a much earlier nineteenth-century tradition in biological
&wdmedical technologies. In any event, consistent with the overall UK or US pattern
of technological development, the British pharmaceutical industry had links with
the food industry, unlike in Germany where it derived purely from the chemicals
industry (Cantwell and Bachmann 1998). In the US there has been a related post-war
556 JOHN CANTWELL
continuation and strengthening o f the medical instrum ent industry (grouped under
professional and scientific instruments in Table 20.4), and a m ore recent m ove into
biotechnology although this has not yet been reflected in a com parative advantage
in the pharmaceutical industry as a whole.
The second cluster is that o f the Germ an-and Swiss-owned corporate grou p s, in
which technological development since the end o f the nineteenth century has been
largely science-based, and revolved around the dom inance o f the chemicals indus
try In the post-war period this has been increasingly com plem ented b y engineering
excellence, although some recent comm entators have seen this direction o f change
(as opposed to a move Into the other science-based area o f electronics) as a weakness
o f the m odern Germ an innovation system (Albach 1996; Audretsch 1996). The
leading German firms have held a consistent focus on developm ent in the chemicals
and metal product industries, with some recent shift towards industries m ore reliant
on engineering-based technologies, linked in part to the emergence o f a wider range
o f innovative smaller specialist supplier companies. The Swiss concentration histor-
Table 20.4 The industries in which the largest nationally owned firms have
persistently held comparative advantage in innovation, 1 9 2 0 *3 9 and
1 9 7 8 *9 5
US-owned UK-owned
Food a n d d rin k T e x tile s
O ffic e e q u ip m e n t a n d com puting O th e r tra n s p o rt e q u ip m e n t
O th e r tra n s p o rt e q u ip m e n t C oal a n d p e tro le u m p ro d u cts
(o th e r th a n m o to r ve h ic le s )
French-owned Swedish-owned
M e ta l p ro d u cts
M e c h a n ic a l e n g in e e rin g
R u b b er an d p la stic pro d u cts
20.4 C o m p e t i t i v e n e s s a t t h e R e g i o n a l
a n d F irm L ev els
Table 20.5 The shares of patenting of the largest industrial firms attributable to
research facilities located in the biggest single region of selected
European countries, in 1 9 6 9 -9 5
Country/Region Percentage
Belgium(Flanders-Brussels) 78.6
France(liede France) 57.9
G erm any (NordrheinW estfalen) 27.0
Italy(Lombardia) 52.3
Netherlands (South Netherlands) 62.7
Sweden(S to ckh olm -O stra Mellansverige) 49.5
Switzerland (Basel) 57.5
UK(SouthEast England) 46.9
Source. Cantwell and lammarino (2001), and (for Germany) Cantwell, lam m arino, and Noonan (2001).
___________ i n n o v a t i o n a n d competitiveness 559
level On the one hand, as seen above, there are general external economies and
spillover effects which attract all kinds of economic activities in certain regions and
determine, in the case of corporate integration, the localization of new research units.
These centripetal forces strengthen the interborder intrafirm integration and the
feedback of knowledge, expertise, and information which occurs within networks of
affiliates. On the other hand, sector-specific localization economies intensify intra
border sectoral integration, implying local external networks between affiliates,
indigenous firms and local non-market institutions. By tapping into local knowledge
and expertise, foreign affiliates gain a competitive advantage which can not only be
exploited locally but may also be transferred back to the parent company, enhancing
its global technological competence. Thus, Narula and Zanfei (Ch. 12 in this volume)
refer to the recent shift away from asset-exploiting and towards asset-augmenting
investments by MNCs, which is typically associated with a greater dispersion of
innovative activity in the international network of an MNC. However, the nature
of the relationship between MNC international innovation systems and local
systems varies across regions (Cantwell and Iammarino 2000, 2001). This entails
different types of regional strategy for technological competitiveness.
Evidence has now emerged that the choice of foreign location for technological
development in support of what is done in the home base of the MNC depends upon
whether host regions within countries are either major centers for innovation or not
(termed “higher-order” or “ lower-order” regions by Cantwell and Iammarino
2000). Whereas most regions are not major centers and tend to be highly specialized
in their profile of technological development, and hence attract foreign-owned
activity in the same narrow range of fields, in the major centers much of the locally
sited innovation of foreign-owned MNCs does not match very well the specific fields
of local specialization, but is rather geared towards the development of general
purpose technologies (GPTs) that are core to cross-industry innovation today
(notably information and communication technologies, ICT) or in the past
(notably mechanical technologies). The need to develop such GPTs is shared by
the firms of all industries, and the knowledge spillovers between MNCs and local
firms in this case may be inter-industry in character. Thus, ICT development in
centres of excellence is not the prerequisite of firms of the ICT industries, but instead
involves the efforts of the MNCs o f other industries in these common locations.
Turning to competitiveness at the level of an individual firm, the determinants of
cross-company growth summarized in Table 20.6 derive from a cross-sectional
regression analysis o f 143 o f the world's largest firms between 1972 and 1982 (Cantwell
and Sanna-Randaccio 1993). As has been remarked earlier, a key aspect of innovation
and growth in the firm has to do with the largely industry-specific environment that
firms have in common and which regulates their individual behavior and partially
reflects their mutual interactions (Levin, Cohen, and Mowery 1985). Thus, the
growth of demand and o f technological opportunities in their own industry are
key influences on corporate performance.
5бО JOHN C A NT WE L L
It is curious that although the w ork o f Penrose and Downie m entioned at the start
o f this chapter emphasized issues o f the creation o f firm -specific capabilities and
intra-industry competitive rivalry, until quite recently even that m inority o f econo
mists that did work on firm growth paid relatively little attention to these issues.
However, now the notion o f corporate competence has m oved center stage in the
strategic management literature (see Lam, Ch. 5 in this volum e). Penrose had argued
that the competitive advantages o f a firm derive essentially from the cumulative and
incremental learning experience o f its management team, which experience differ
entiates it from other firms. The distinctiveness o f the firm's accum ulated experience
and knowledge determines the set o f opportunities for growth which it perceives
ahead o f its rivals when screening the external environm ent (the growth o f demand
and technological opportunities in its industry). Corporate technological competi
tiveness is the principal advantage o f this kind associated with differentiated learn
ing (Cantwell 1989; Teece, Pisano, and Shuen 1997). A higher technological capability
lowers the unit costs and raises the demand curve o f the firm at a given rate of
growth, and it facilitates new entry Into related product lines. A m ore technologic
ally competent firm is able to utilize its existing experience to lower the costs of
expanding its managerial and technical team in related areas. Technological com
petitiveness comes out as one o f the most statistically significant influences upon
firm growth o f the variables listed in Table 20.6. In that large firm study, techno
logical competitiveness was measured by the firm ’s share o f patenting in its industry
relative to its market share (o f industry output), in terms o f the relevant world
industry
I N N O V ATION AND COM PETITIVENESS 561
20.5 C o n c l u s i o n s
To return to where this chapter began, competitiveness derives from the creation of
the locally differentiated capabilities needed to sustain growth in an internationally
competitive selection environment. Such capabilities are created through innov
ation, and because capabilities are varied and differentiated, and since the creative
learning processes for generating capabilities are open-ended and generally allow for
multiple potential avenues to success, a range of different actors may improve their
competitiveness together. Innovation is a positive sum game that consists of the
efforts often of many to develop new fields of value creation, in which on average the
complementarities or spillovers between innovators tend to outweigh negative
feedback or substitution effects, even if there are generally at least some actors that
lose ground or fail. The basic conclusion is that efforts to promote competitiveness
through innovation can rarely be understood in isolation from what others are
achieving at the same time. This applies whether we are speaking of countries, of
national groups of firms in an industry, of subnational regions, or of individual
companies. Indeed, it is worth emphasizing that the degree of interaction between
innovators in search o f competitiveness has tended to rise substantially historically,
and has attained new heights in recent years.
Firms are less independent than they were, and they nowall float in a much deeper
sea of background knowledge, which Nelson (1992) refers to as the “public” element
of technology. There are at least four aspects to this: intercompany knowledge flows
have increased, there is a growing role for governments and other non-corporate
institutions in knowledge development and transfer, the importance of science for
technology has risen and diversified in its impact, and there has been a tendency
towards more rapid codification and the formation and spreading of professional
and scientific communities. We can now think of firms and the individuals aboard
them like ships floating in a sea o f public knowledge which connects them, or more
accurately potentially public knowledge since the extent that they can draw upon it
depends upon their own absorptive capacity and on their membership of the
appropriate clubs (whether intercompany alliances or professional associations
and the like). Over time, especially since 1945, firms have been designed to float
deeper down in the water, but they still always leave a critical part comprising their
own tacit capabilities above the surface, which does not sink down or fall into the
general mass. Indeed, holding stronger capabilities above the surface is positively
related to the depth to which one can reach below the surface, both for the absorptive
capacity to extract complementary knowledge and for the extent to which one
contributes oneself to the public knowledge pool. Universities and governments
have increasingly contributed to the sea of public knowledge as well. Additionally,
am°og firms that deliberately cooperate through technology-based alliances per
sonnel can be exchanged so as to coordinate learning efforts.
562 JOHN C A N T W E L L
The sharing o f knowledge between firms implies not just that tech n o lo g y must be
d evelo p ed th ro u g h an interactive so cial an d cultural e v o lu tio n rather than through a
biological evolutionary process involving competition between g en etically inde
pendent entities, but also that followers and innovative adapters may stand to make
greater gains than the original leaders in some new field o f technological endeavor.
For example, knowledge developed in on e co n text may ultimately p ro ve to h ave a
b ig g er im p a ct in another, w h ich was n o t foreseen b y the o rig in a to r o r even perh ap s
in itia lly b y the m o st in n o vatively su ccessfu l recipien ts. Firms also n o w d evo te m uch
greater effort to attem p ts to u n d erstan d th eir o w n tech n o lo g ica l p ra ctice a n d that o f
others. C o d ific a tio n o f k n o w led ge is the o u tco m e o f a co n sc io u s effo rt, sh iftin g back
the d iv id in g lin e betw een w h at is p o ten tially p u b lic an d w h at is tacit (Cohendet and
S tein m u eller 2000; C o w a n , D a v id , a n d F o ra y 2 0 0 0 ). So firm s th at b e c o m e especially
adept at co d ificatio n m a y fin d that this is a so u rce o f co m p e titive ad van tage since
th ey can then m o re rea d ily d raw o n the p u b lic p o o l.
To engage in this in terco m p an y in teractio n fru itfu lly, firm s m u st m ain tain an
adequate d iversificatio n o f th eir in -h o u se tech n o lo g ical effo rts, since the clo ser that
k n o w led ge is to the p ro p rie ta ry interests o f a firm the m o re lik ely th at it w ill o n ly be
sh ared in return fo r so m eth in g else that is p ro b ab ly tech n o lo g ica lly co m p lem en tary,
w h ich is w h at each firm needs to jo in the relevant co rp o ra te clu b (C an tw ell and
B arrera 1998). T h e en trep ren eu rial fu n ctio n has n o t been e lim in a ted b u t it is more
in stitu tio n a lly em b ed ded in an ab ility to n etw o rk an d m ak e n e w co n n ectio n s (see
P ow ell an d G ro d a l, C h . 3 in this vo lu m e ).
T h e in teractio n effects betw een in n o va to rs h as b een fu rth er c o m p o u n d e d b y the
role o f IC T as a m eans o f co m b in in g fields o f k n o w led g e creatio n that w ere p revi
o u sly kept largely ap art (o r w h at K o d a m a 1992, term s te c h n o lo g y fu sio n , an d has led
to the creation o f new fields such as bioinformatics). IC T thus broadens the field for
potential innovation by linking formerly separate areas o f innovative activity, ICT
p o ten tially co m b in es the v a rie ty o f tech n o logical fields th em selves and so increases
the sco p e fo r w id e r in n o vatio n .
So, in the light o f these recent changes in em p h asis in the co n text fo r in n o vation
and competitiveness, how can we reevaluate some o f the earlier literature in the field
that has been su m m arized here? W h ich lines o f research h ave p erh ap s run o u t o f
steam fo r the tim e being, an d w h ich offer the m o st p ro m isin g n ew avenues o r
o p p o rtu n ities fo r fu tu re research? T h e revival o f the classical issues o f co m p etitive-
ness at a n ation al level (Sectio n 20.2) has been u sefu l in that it represents a return o f
interest into the m a jo r q u estion s o f the w ealth o f n ation s, w h ich are o f p rim ary
im p o rtan ce fro m a social v iew p o in t. H ow ever, this n ew literatu re h as also exposed
the lim itatio n s o f tryin g to tackle w h at are really issues o f stru ctu ral ch an ge at an
aggregate kveL It might perhap s be ju d g ed that w e h ave g o n e a b o u t as fa r as w e can
for n o w w ith pu rely cro ss-co u n try m odels. It may w ell be fo r th is reaso n that the
fo cu s o f the latest research has tended to shift to the in d u stry an d firm levels,
allo w in g fo r (indeed n o w em p h asizin g m o re) the sco p e fo r tech nological
interactions betw een firm s an d in d u stries, so as not to lose sight o f the w ider context
or the aggregate effects.
When recasting the an alysis at an in d u stry level (Section 20.3), we now know a
good deal about h isto rical sh ifts in p attern s o f in d u strial leadership betw een cou n
tries, about the role o f e d u c a tio n an d skills in catch in g up econom ies, and how
particular kinds o f skills h elp to a cco u n t fo r in ter-in d u stry discrepancies in in n o va
tive potential A rg u a b ly w e still k n o w too little ab out the interaction between
governments, n o n -b u sin ess in stitu tio n s, an d firm s (especially large firm s) in the
process o f establishing co m p etitiven ess. In particu lar, w e w o u ld like to kn o w m ore
about u n iv e rsity -in d u stry (sc ie n c e -te c h n o lo g y ) in teraction over a w id er range o f
countries, beyond the rela tiv ely clear p ictu re that w e have fo r the U S (see M ow ery
and Sampat, C h. 8 in th is v o lu m e ). T h e con text here is w hat seem s to be the grow ing
significance o f a local scien ce b ase fo r the co n stru ctio n o f corporate capabilities and
hence com petitiveness, in c lu d in g an d p erh ap s especially in latecom er econom ies.
Note that this n ew ly e m e rg in g v ie w reverses the “ trad itio n al” perspective that
developing cou n tries sh o u ld co n cen trate o n (o rgan izatio n al in n ovation in) low er-
skill activities, and leave scien ce to the largest m o st developed econom ies.
Coming to com p etitiven ess at the firm o r clu ster level (Section 20.4), the latest
research has also h ad a ren ew ed fo cu s o n the role o f in tercom pan y interaction in
knowledge creation an d in n o v a tio n , esp ecially in su b n atio n al regional areas, and
through alliances o r c o o p e ra tiv e agreem en ts. H ere w e n o w kn o w m ore o f the details
of the localized ch aracter o f in n o v a tio n , an d o f the steady grow th in technology-
based alliances as a m ean s o f facilititatin g com p etitiven ess through know ledge
exchange and spillovers. W o rk o n firm size an d in n o vatio n or grow th seems to
have rather run o u t o f steam fo r n o w , at least in so far as it had regarded in dividual
firms as quite in d ep en d en t en tities. W e need to k n o w m o re ab o u t the specificities o f
knowledge flow s b etw een reg io n s an d b etw een firm s, o f h o w and w here tech no
logical knowledge is so u rce d b y firm s, an d then h o w such know ledge is effectively
combined in n etw o rks o f in terrelated in n o v a tio n w ith in and betw een firm s. This is
surely an exciting agen d a fo r fu rth e r research.
Re fe r e n c e s
Abramowitz, M. (1956), “ Resources and Output Trends in the United States since 1870,”
46(1): 5-23.
American Econom ic R e v ie w
Albach, H. (1996), “ Global Competitive Strategies for Scienceware Products, in G. Koop-
mann and H.-E. Scharrer (eds.), The Economics of High-Technology Competition and
Cooperation in Global Markets, Baden-Baden: Momos Verlagsgesellschaft, 203-17.
Audretsch, D. B. (1996), “ International Diffusion of Technological Knowledge, in
G- Koopmann and H.-E. Scharrer (eds.), The Economics of High-Technology Competition
and Cooperation in Global Markets, Baden-Baden: Momos Verlagsgesellschaft, 107-35.
564 JOHN C A NT WE L L
Fagerberg , J. (1987), “A Technology Gap Approach to Why Growth Rates Differ,” Research
Policy 16(1): 87-99*
*(1988), “International Competitiveness/5Econom ic Jo u rn a l 98: 355-74.
*(2002), Technology, G row th a n d C om petitiveness: Selected Essays, Cheltenham: Edward
Elgar.
■---- Verspagen, B., and C aniels , M. (1997)» “Technology, Growth and Unemployment
across European Regions/5 R eg io n al Studies 31(5): 457-66.
Freeman, C , and LougA, F. (2001), As T im e Goes B y: From the Industrial Revolutions to the
Information R evolution , O xford and New York: Oxford University Press.
---- and Perez, C (1988), “ Structural Crises of Adjustment, Business Cycles and Investment
Behaviour/5in G. Dosi, C. Freeman, R. R. Nelson, G. Silverberg, and L. L. G. Soete (eds.),
Technical Change a n d E con om ic Theory ; London: Pinter 38-66.
Hagerstrand, T. (1967), In n ovatio n D iffusion as a Sp atial Process , Chicago: University of
Chicago Press.
Kline, G . a n d R osenberg , N. (1986), “An Overview of Innovation/5in R. Landau and N.
Rosenberg (eds.), The Positive Su m Strategy: H arnessing Technology fo r Econom ic G row th ,
Washington, DC: National Academy Press, 275-306.
Kodama, F. (1992), “ Technology Fusion and the New R&D,” H arva rd Business R eview
(July-August): 70-8.
*Krugman, P. R. (і 994«)> “ Competitiveness: A Dangerous Obsession/5Foreign Affairs 73(2):
28-44.
---- (1994b), “The Myth of Asia’s Miracle/5Foreign A ffairs 73(6): 62-78.
---- (1996), “Making Sense of the Competitiveness Debate/5 O xford R eview o f Economic
Policy 12(3): 17-25,
*Lall , S. (2001), C om petitiveness, Technology a n d Skills , Cheltenham: Edward Elgar,
Landes , D. S. (1969), The U n b o u n d Prom etheus: Technological a n d Industrial D evelopm ent in
Western Europe fro m i j s o to the Present , Cambridge and New York: Cambridge University
Press.
Lazonick ,W . (1991), Business O rgan ization a n d the M yth o f the M arket Economy, Cambridge
and New York: Cambridge University Press,
*(1992), “Business Organization and Competitive Advantage: Capitalist Tranformations in
the Twentieth Century/5in G. Dosi, R. Giannetti, and P. A, Toninelli (eds.), Technology and
Enterprise in a H istorical Perspective , Oxford and New York: Oxford University Press,
119-63,
~~— (1993)) “Industry Clusters versus Global Webs: Organizational Capabilities in the
American Economy/5In d u stria l a n d C orporate C hange 2(1): 1-24.
■"— (1998), “Organizational Learning and International Competition/5 in J. Michie and
I* Grieve-Smith (eds.), G lo ba lizatio n , G row th a n d G overnance , Oxford and New York.
Oxford University Press, 204-38.
---- and O’Suleivan, M. (1997), “ Big Business and Skill Formation in the Wealthiest
Nations: The Organizational Revolution in the Twentieth Century, in A. D, Chandler,
E Amatori, and T. Hikino (eds.), B ig Business a n d the W ealth o f Nations, Cambridge and
New York: Cambridge University Press, 497-521.
Levin, R. C , C ohen, W. M., and M owery , D. C. (1985), “ R&D Appropriability, Opportunity
and Market Structure: New Evidence on Some Schumpeterian Hypotheses/5 Am erican
Economic R ev iew 75: 20-4,
$66 JOHN C A NT WE L L
INNOVATION AND
EMPLOYMENT
MARIO PI A NTA
21 . 1 I n t r o d u c t i o n
Main research questions Genera! approach Major streams of literature and key findings Key assumptions and methodology Main level of
Does technology create or destroy jo b s ? analysis
question as to where and to what extent jobs are created or destroyed depends upon
the highly dynamic process that shapes the content o f specific technological innov
ations and the speed o f their adoption, with an often blurry distinction between the
two (see below). Moreover, jobs lost and new jobs offered may take place in different
areas or require different skills, leading to mismatches. The speed o f adjustment is
therefore crucial and makes the difference between frictional unemployment (easily
absorbed by well-functioning labor markets) and technological unemployment.
Sustained and sustainable growth can be expected only once the mismatches
between the new technologies and the old economic and social structures and
institutions are overcome, with a two-way adjustment. Innovation has to be adapted
to social needs and econom ic demands; economic and social structures evolve under
pressure from new technologies. New technologies need to be matched by organiza
tional changes, new institutions and rules, learning processes, the emergence o f new
industries and markets, and the expansion o f new demand. Several studies on the
emergence o f technological paradigm s and key technologies in the past have pointed
to the long time required before the impact o f these elements (positive or negative)
on economic growth and em ploym ent become evident (Freeman, Clark, and Soete
1982; Freeman and Soete 1987, 1994; Freeman and Louya 2001). While extremely
powerful in its explanation o f long term economic changes and historical evolution,
this approach has yet to be “ operationalized” with more specific questions on the
type of innovation and on the interaction with economic and employment variables.
Since the emergence of the Industrial Revolution, the extensive substitution of labor by
tnachinery incorporating the new technology of the time has led economists and policy
makers to debate the economic and social consequences. At the end of the eighteenth
century, James Steuart drew attention to the difficulty of reabsorbing the unemploy
ment caused by sudden mechanization, in spite of the positive effects from the
construction o f new machines and price reductions, and already envisioned a role
for the government. Adam Smith linked the invention of machines to the division of
labor and emphasized its labor-saving effects. Jean-Baptiste Say had less doubts about
the ability of markets to adjust, while Thomas Malthas emphasized the positive effects
resulting from the strong demand dynamics experienced by England at the time. The
optimism of classical economists in the early nineteenth century contrasted with the
dramatic impoverishment of the English working classes—industrial workers, small
artisans, and displaced peasants— who had started to organize trade unions and to
s launch Luddite straggles against the job losses and deskilling brought about by
, :fccbamzation. David Ricardo was convinced that the economy could compensate
the negative employment effects, but in a famous passage in the chapter On machin
e r y added in the third edition of his Principles of Political Economy and Tamttm%
572 MARIO P I A N T A
argued that “ The opinion, entertained by the labouring class, that the employment of
machinery is frequently detrimental to their interests, is not founded on prejudice and
error, but is conformable to the correct principles of political economy* (Sicardo
1951:392), '
The most articulate criticism of compensation theory was developed by Karl M as,
who emphasized the losses for workers in terms o f jobs, skills, wages, and control over
their work resulting from the way mechanization was proceeding at the time. Arguing
that unemployment grows as technical change displaces labor more rapidly that the
accumulation o f capital demands new workers, M a s developed important insights on
the functioning o f capitalism. The drive to capital accumulation leads to a constant
search for new production techniques and new products (a key starting point in Joseph
Schumpeter’s theory o f innovation). High unemployment assures lower wages and
greater control over workers, but capital accumulation ultimately encounters the
problems o f finding adequate markets and demand, and making adequate profits
(for a reconstruction o f the debate see Heertje 1973; Vivarelli 1995).
The distinction between process and product innovations should not be carried
too fan Most innovative firm s introduce both at the same time, but in most firms and
industries (see Ch. 14 by M alerba in this volume) it is possible to identify the
dominant orientation o f innovative efforts, associated with strategies o f either
p ric e competitiveness (and m ainly process innovations) or technological competi
tiveness (and m ainly product innovations). In addition to product and process
innovation, organizational innovation also can affect the quantity and quality o f
employment, and is usually closely linked to the introduction o f new technologies
(Caroli 2001). Sections 21.22 and 21.3 below will consider the impact on job creation
and loss o f each o f these types o f innovation.
Conceptual and empirical difficulties have long led economists to adopt a rather
homogeneous view of innovation, described either by R&D expenditure (one of its
inputs), or proxied by patenting activity (one of its output). In the last decade, the
Spread of innovation surveys in Europe and of surveys on panels of firms in the US has
provided important new evidence on the variety of innovative activities.
The key results for Europe in the period 1994-6 are that innovation is present in 51 per
centof manufacturing firms and in 40 per cent of services firms (see Chapters 6 and 16 by
Smith and M ies in this volume). Close to 40 per cent of manufectoring firms
Aange their production processes, while a fifth develop minor product improvements
Industry cades
01 Food products
02 Textiles
03 Apparel
Percentage change in employment, 1994-2000
Countries
FR France
IT Italy
NL the Netherlands
UK United Kingdom
Share of new products in sales,1994-1996
yaffil another fifth introduce products new to their market, whose impact on sales is
limited to 12 per cent. In all indicators, important differences are found across
countries, associated with the nature of their industrial structure and national innov
ation system (European Commission -Eurostat 2 0 0 1 ).
The latter variable offers the most accurate description o f the economic relevance o f
innovations, and is related, in Figure 2 1 . 1 , to the employment performance (average
annual rate of change) of 20 manufacturing industries in four EU countries (France,
Italy, the Netherlands, and the UK). Looking at industry data for the share of new
products in sales (drawn from the CIS 2 -SIEPI database, which provides data at the
two-digit industry level for major countries), a very strong variability is evident, from
more than 20 per cent in office computing and telecommunications, to close to zero in
more traditional industries (again, with important cross-country differences due to
national specializations). Looking at overall employment change, we first see that,
between 1994 and 2 0 0 0 , the majority o f industries lost jobs. The relationship between
product innovation and employment experienced in European industries looks like a
positively oriented curve, but a closer look at the distribution of cases is important.
ІСГ industries (computing, telecommunications, precision instruments, and other
transport, including aerospace) are generally in the top right quadrant, where new
products in sales have a share higher than the average value for the EU and where no
dramatic job losses are found. As expected, industries characterized by the new technol
ogy (and by a high concentration of product innovations) show the highest impact of
new products in their turnover and better employment performances. In a few cases,
however, moderate job losses are found even in this group, as strong international
competition may lead to the decline of some ICT industries in some countries.
Traditional industries (textiles, wearing apparel, leather, and a few others) tend to
concentrate in the bottom left quadrant, where a below average innovativeness (and a
strong dominance o f process innovations) is matched by dramatic job losses.
The remaining sectors, in the top left quadrant, combine a low or intermediate impact
of innovation with modest job losses or substantial gains, showing again a positive
association. While several factors alongside innovation affect employment change
(macroecomic conditions, competitiveness, etc., see Pianta 2000), the distribution of
Figure 21Л highlights, on the one hand, the generally positive link we expect between
product innovation and jobs; on the other hand, it shows the presence of winners and
losets maU industries, reflecting the importance o f national specializations, of economic
АпШіге$, and o f the intensity o f international competition in open economies.
21.3 T h e E f f e c t s o n t h e Q u a n t i t y
of Employment
ЧЛ
VI
VI
Vl
Table 21.2 (cont.)
MARIO PIA N T A
S tu d y C o u n trie s Y ears Level o f an alysis In n o v a tio n d a ta sources Results o n employment
M a c r o e c o n o m ic -le v e l studies
Layard an d Nickel! 1 9 8 5 UK 1954-83 M a c ro model Labor p ro d u c tiv ity Neutral
Viva relit 1995 US an d Ita ly 1966-86 M a c ro m odel REtD lin ked to p ro d u c t and Differentiated by compensation
process in n o v a tio n s m e c h a n is m and country
S im o n e tti, Taylo r, an d US, Italy, 1965-93 M a c ro m o d e l REtD lin ked to p ro d u c t and Differentiated by compensation
Vivarelli 2 0 0 0 France, Jap an process in n o v a tio n s mechanism
S im o n e tti a n d T a n c io n i UK an d Ita ly 1970-98 M a c ro m o d e l q u a rte rly REtD lin ked to p ro d u c t an d Differentiated by compensation
2002 data process in n o v a tio n s mechanism
Simulation studies
Leontiefand Duchin 1986 US 1980-2000 In p u t o u tp u t m o d e l ail A s s u m p tio n s on Negative
econom y p e rfo rm a n c e
K a lm b a c h an d K urz 1 9 9 0 G e rm a n y 2000 In p u t o u tp u t m o d e l all A s s u m p tio n s Negative
econom y
IPT5~E5T0 2001 Europe 2000-2020 Gen, e q u il. m o d e l all Assumptions on Positive, differentiated by
econom y p ro d u c tiv ity g ro w th innovation policy
IN N O V AT IO N AND EM PL OYM EN T 579
thejobs created or lost may be accounted for by the entry or exit of firms left outside
thepanel. In order to address these issues, we need to turn to industry-level studies
Vivarelli (i 995)> has sum m arized the compensation mechanisms and the way
they may (or may not) operate in the economy.
The compensation mechanism v ia d e c re a s e in p r ic e s is one o f the most important
ones.* new technologies m ay make lower prices possible, increasing international
competitiveness and output, offsetting job losses due to the original innovation
This outcome, however, is contingent on the lack o f demand constraints, on the
decision o f firms to transfer in lower prices the productivity gains due to the
innovation, and on the lack o f oligopolistic power in the relevant markets (Sylos
Labini 1969).
The compensation m echanism v ia n e w m a c h in e s may create jobs in the industries
in which the new means o f production are made, responding to the increased
demand for equipment by users. However, at the aggregate as well as at the firm
level, the rationale for mechanization is by definition saving on the overall use o f
labor, putting a limit on the relevance o f this mechanism.
The compensation m echanism v ia n e w in v e s tm e n t argues that the temporary
extra profits available to the innovator may be turned into new investment if profit
expectations are favorable (and assum ing that Say's law operates); this, however,
may expand production capacity and jobs, or m ay introduce additional labor saving
effects.
The compensation m echanism v ia d e c re a s e in w a g es is typical o f the neoclassical
view of the labor market . As technological unemployment appears, wages should fall
and firms should hire m ore workers. This mechanism, however, is based on strong
assumptions as to the feasibility o f any com bination o f labor and capital, competi
tive markets, flexibility o f wages, and labor markets.
The compensation m echanism v ia in c re a s e in in c o m es operates in the opposite
way, through the increased dem and associated with the distribution o f part o f the
gains from innovation through higher wages, as has happened in large, oligopolistic
firms in mass production industries. However, any wage increases can hardly be
large enough to sustain additional aggregate demand.
Finally n e w p r o d u c t s m ay lead— as discussed above— to new economic activities
and new markets ( w e lfa r e effects) or, on the other hand, they may simply replace
existing goods ( s u b s titu tio n e ffe c t , see Katsoulacos 1986),
Aggregate empirical patterns for the US are examined in a descriptive way by
Baumol and W olff (1998). Considering five innovation indicators for the whole
economy and their link to the structure and changes in unemployment in the US in
the 1950-95 period, they conclude that faster innovation leads to a higher natural
rate of unem ploym ent" and to longer spells o f frictional unemployment.10 Layard
and Nickell (1985), on the other hand, have argued that the working o f various
compensation mechanism s ruled out the possibility o f technological unemploy
ment in the UK.
Building on Boyer (1988a and b) and on the Regulationist approach, Vivarelli
(1995) has developed a sim ultaneous equations model for testing the compensation
5І2 MARIO PIANTA
2 1 .4 T h e E ffe c t s on th e Q u ality
of E m p lo ym en t
M A R I O PI A N T A
S tu d y C o u n t r ie s Y e a rs L e v e l o f a n a ly s is I n n o v a t io n d a t a s o u r c e s R esults
T e c h n o lo g y a n d w a g e d is p e r s io n
F ir m - le v e l s tu d ie s
C asavo la, G avosto , a n d I t a ly 1986-90 C ro s s f i r m s °/o o f in t a n g i b l e c a p it a l P o s it iv e e f f e c t o n s k ills
Sestito 1998 IM P S w a g e d a t a L im it e d w a g e d is p e r s io n
i n d u s t r y l e v e l s t u d ie s
O r g a n iz a tio n a l in n o v a t io n
F i r m - l e v e l s t u d ie s
skilled white collars have declined more rapidly, and little change in the skill
structure o f services em ploym ent is apparent. ICT investment since the mid-1980s
therefore appears to have negatively affected the low skilled white collars (mainly
female) much more than the lowest blue collar skills.
affect the supply o f labor (see Acem oglu 2002). Slower population growth, higher
immigration flows, an ageing population and the forms o f women s presence in the
labor force all influence labor supply. In addition, the quality o f labor suppiy~~and
the potential for innovation— are affected by education and training, learning
processes, and the accum ulation o f competences by workers and firms.
21.5 C o n c l u s i o n : S t y l i z e d F a c t s and
Di r e c t i o n s for F u t u r e Re s ea rc h
This chapter has examined the relationship between innovation and employment,
reviewing a large em pirical literature. The com plexity o f the issues is such that no
single approach can account for the direct and indirect consequences o f techno
logical change, or for its effects on the quantity and quality o f labor employed in the
economy.15 Theories and em pirical research have to proceed in parallel, with a close
interaction between concepts and measures, hypotheses and tests, building links to
related research areas. This final section summarizes some key stylized facts
emerging from the em pirical evidence and highlights possible directions for future
research.
N o t e s
1. Fo r their com m en ts and suggestions, I w ish to th an k the editors, C h arles E dquist, Bengt-
о
A ke Lu n d vall, M arco V ivarelli, and several con trib u to rs to this vo lu m e,
2. T he em ergence o f n ew fo rm s o f o rganization , the o p en in g o f n ew m arkets and o f new
sources o f m aterials w ere other types o f in n o vatio n con sidered by Schum peter.
3. G en erally po sitive effects o f in dicators o f in n o vatio n o n the n u m b er o f jo b s (once the
characteristics o f firm s have been controlled for) have been fo u n d in the studies o f U K
firm s by Van Reenen (1997), w ho related a large panel o f m an u factu rin g firm s to the SP R U
database o f B ritish in n ovation s, and b y M ach in an d W adhw ani (1991), u sin g the British
w orkplace in dustrial relations su rvey on the ad o p tio n o f ICTs. In a pan el o f G erm an firm s,
Sm o ln y (1998) has fo u n d a positive effect o f p ro d u ct in n o vatio n and no effect o f process
in n ovation s. D utch firm s investigated b y B rouw er, K leinknecht and R eijnen (1993)
show ed an overall negative link betw een in n o vatio n and job s, but w here p ro d u ct in n o v
ations were d o m in an t, better em ploym ent outcom es have been fo u n d .
4. T his view is pro po sed b y A n ton elli (2 0 0 1:17 3 ) in a stu dy on the Italian car m aker Fiat over
the tim e p erio d 19 0 0 -19 7 0 , w hich show s that em ploym en t grow th G ran ger-cau sed patent
grow th w ith different tim e lags. In turn, patent grow th led to p ro d u ctivity grow th,
5. In France, G reen an and G uellec (2000) fo u n d a po sitive relation ship betw een both
p rod u ct and process in n o vatio n and em ploym ent at the firm level, but at the in dustry
level, o n ly the fo rm er played a positive role. U sing data fro m the French h o u seh old survey,
Entorf, G ollac, and K ram arz (1999) fo u n d that co m p u ter use reduces the risk o f u n
em ploym ent in the short term , but not in the lon g term .
6. H igh d em an d is a necessary, but not sufficient, con d itio n fo r em p loym en t grow th. In
order to expand jobs, dem and and outpu t have to gro w faster than prod u ctivity, a
con d ition that has been often fou nd in the U S, but not so m u ch in Europe.
__ _________________________________________ 1 N N O V A T I O N a n d E M P L O Y M E N T 593
7 - Typical of the former approach is the OECD Jobs Study (OECD i994), which invest!
gated the high and persisting unemployment in advanced countries and downplayed the
role of technological change. The report emphasized the positive role o f new technolo
gies associated to structural change and showed employment decline in low-technolaav
sectors and also in some R&D intensive industries.
8. Meyer Kramer (1992) has used a model of the whole German economy to assess the
impact o f ICTs, proxied by direct and indirect R&D in 51 sectors. The findings suggest a
generally negative em ploym ent impact, with some positive effects in higher technology
industries.
9. An early use o f these data for Italian manufacturing industry is in Vivarelli, Evangelista,
and Pianta (1996), who found a generally negative employment impact of technological
change and the expected contrasting consequences o f product and process innovations.
Pianta (2000,2001) investigated five European countries (Denmark, Germany, Italy, the
Netherlands, and Norway) in 1989-93 across twenty-one manufacturing industries,
finding a positive employm ent effect o f changes in demand and product innovations,
and an overall negative impact o f innovation intensity.
10. They argue that “ The evidence supports the conclusion that an increase in the pace of
innovation (all else equal) will raise both the natural rate o f unemployment and the
average length o f time during which an unemployed worker is 'between jo b s'" (p.10).
11. Their assumptions on productivity growth were based on the improvements in the
engineering performance o f robots. The long time-lags in their adoption and their poor
initial performance have meant that their actual economic effect on productivity has
been much lower.
12. Petit (1995) suggests that techniques and human labor had a complementary relation
ship in pre-industrial times, following strong social norms. Industrialization was char
acterized by the substitution o f machines for labor on the basis o f profit seeking; and the
post-war rapid growth was associated with technologies designed to overcome labor
shortages and increase productivity, while the current emergence o f a new technol ogical
system based on ICTs is marked by uncertainties.
13. Computer-related investment per employee rose from 150 $US in 1982 to about 1000 $US
in 1992 for both m anufacturing and services (Howell 1996: 292). The share o f low-skilled
blue collar US workers in manufacturing declined from 45*1 per cent in 1978 to 39*7 per
cent in 1982 and has since then remained stable, ending in 1990 with a 41 per cent share. In
the late 1980s, when ICTs became important, the most serious reduction in the employ
ment shares concerned the high-skilled blue collars (which since 1978, had a stable share
around 21.7 per cent, a share which declined after 1985 to 19.7 per cent in 1990) and the
low-skilled white collars (a share which stayed stable at around 12 per cent until 1985 and
then declined to 10.6 per cent in 1990). At the opposite end, the share o f high-skilled
white collars increased in the 1978—83 period (from 19.5 to 24.1 per cent), then was stable
until a rise in 1989-1990 to 25.8 per cent (Howell 1996: 299, tables 1 and 2).
H* This may amount to a shift in “ wage norm s": “ in the face o f mounting competition,
employers reduced unit labour costs and increased flexibility in the production process
by following the Tow road'— lower wages, little training, and fewer permanent employ
ees" (Howell 1996:301), Com bined with a large use o f part-time and temporary workers,
anti-union practices, relocation to low-wage sites and inflows o f low-wage foreign
workers, these developments have led to an increase in the supply o f labor competing
594 MARIO PIANTA
for low-skill jobs, leading to a major fall in their wages, accompanied by a 25 per cent
reduction in the real minimum wage over the 1980s (ibid. 292).
15. There is also a “ national specificity” to most o f the different approaches. The “ skills”
research is largely US-focused, while most industry-level studies focus on Europe. There
is little English-language empirical work on the em ploym ent-innovation relationship in
Japan or other Asian high-growth economies.
R efe r e n c e s*
F r e y , L. (1997), “ II lavoro nei servizi verso il secolo XXI,” Quaderni diEconom ia delLavoro 57,
M ilan : A ngeli.
G r e b n a n , N . (2003), “ O rgan isation al C hange, Technology, E m p lo ym en t an d Skills: A n
E m p irical Stu d y o f French Manufacturing,” C a m b r id g e J o u r n a l o f E c o n o m ic s 27: 287-316.
-------an d G u e l l e c , D. (2000), “ Technological In n o vatio n an d E m p lo ym en t R eallocation,”
L a b o u r 14 (4 ): 54 7-90.
* G r e e n a n , N ., UH o r t y , and M a ir e s s e , J. (eds.) (2002), P r o d u c tiv ity , In e q u a lit y a n d the
D ig it a l E c o n o m y ; C am b rid ge: M ass.: M IT Press.
H a n s e n , A. H, (1964), B u s in es s C ycles a n d N a t io n a l In c o m e , L o n d o n : A llen and U nw in
(expanded edn.).
H e e r t ie , A. (1973), E c o n o m ic s a n d T ech n ica l C h a n g e >L o n d o n : W eidenfeld an d N icolson .
H o w ell , D. (1996), “ In fo rm atio n Technology, Skill M ism atch an d the W age Collapse:
A Perspective o n the U S Experience,” in O E C D 1996: 29 1-30 6 .
-------and W o lff , E. (1992), “ Technical C h an ge and the D em an d fo r Skills b y U S Industries,”
C a m b r id g e J o u r n a l o f E c o n o m ic s 16 :12 8 -4 6 .
IP T S -E S T O (Institute fo r Prospective T echn ology Studies, E u ro p ean Science an d Technology
O b servatory) (20 0 1), Im p a c t o f T ech n o lo g ica l a n d S t r u c tu ra l C h a n g e o n E m p lo y m e n t .
P ro s p e c tiv e A n a ly s is 2 0 2 0 , S y n th e s is R e p o r t a n d A n a ly t ic a l R e p o r t , Seville: E u ropean C o m
m ission Jo in t Research Centre.
K a l m b a c h , P., and K u r z , H. D. (1990), “ M icro electro n ics an d E m p lo ym en t: A D ynam ic
In p u t-O u tp u t Stu d y o f the West G erm an Econom y,” S tr u c tu ra l C h a n g e a n d E co n o m ic
D y n a m ic s 1: 317-8 6 .
K a r a o m e r lio g lu , A ., an d A n s a l , T. (2000), “ C o m p en satio n M ech an ism s in D eveloping
C ou n tries,” in V ivarelli and Pianta 2 0 0 0 :16 5 -8 1.
K a tso u la co s , Y. S. (1986), T h e E m p lo y m e n t E ffe c t o f T e c h n ic a l C h a n g e , B righ to n : W heat-
sheaf.
* K r u e g e r , A . (1993), “ H o w C o m p u ters have C h an ged the W age Structure: Evidence from
M icro D ata 1984-1989,” Q u a r te r ly Jo u r n a l o f E c o n o m ic s 108: 3 3-6 0 .
L a y a r d , R ., an d N ic k e l l , S. (1985), “ The C auses o f B ritish U n em p loym en t,” N a tio n a l
In s titu te E c o n o m ic R e v ie w 111: 62-85.
L e o n t ie f ,W., and D u c h in , E (1986), T h e F u t u r e Im p a c t o f A u t o m a tio n on W orkers , O xford:
O xfo rd U n iversity Press.
о
Nelson , R*, and P h elps , E, (1966), Investment in Humans, Technological Diffusion and
Economic Growth,” A L A Papers a n d Proceedings 56: 69-75,
----- and Winter , S. (1982), A n E vo lu tio n a ry Theory o f Econom ic Change, Cambridge, Mass,:
Belknap Press.
Noble, D. (1984)» Forces o f P roduction : A Social H istory o f Industrial Autom ation , New York:
Knopf.
OECD (1994 )* The O E C D Jo b S tu d y E vid en ce a n d Explanations. Part f Labour M arket Trends
and Underlying Forces o f C han ge , Paris: OECD.
----- (i99h)» E m ploym en t a n d G row th in the K now ledge-B ased Econom y Paris: OECD.
Padalino, S., and V ivarelli , M. (1997), "The Employment Intensity of Economic Growth
in the G-7 Countries,” In tern a tio n al L a b o u r R ev iew 136:191-213.
Pasinetti, L. (1981), Structu ral C han ge a n d Econom ic G ro w th Cambridge: Cambridge
University Press.
Perez, C. (1983)» “ Structural Change and the Assimilation o f New Technologies in the
Economic and Social Systems,” Futures 15(5): 357-75.
*Petit, P. (1995) у “ Em ploym ent and Technological Change,” in P. Stoneman (ed), Handbook
of the Economics o f In n ovation a n d Technological Change, Amsterdam: North Holland,
366-408.
*----- and Soete , L. (eds) (2001л), Technology a n d the Future o f European Employment,
Cheltenham: Edward Elgar.
----------- (2001b), "Technical Change and Employment Growth in Services: Analytical and
Policy Challenges,” in Petit and Soete 2001a: 166-203.
Pianta, M. (2000), "T h e Em ploym ent Impact o f Product and Process Innovation” in
Vivarelli and Pianta 2000: 77-95.
----- (2001), "Innovation, Dem and and Employment,” in Petit and Soete 2001a: 142-65.
----- Evangelista , R., and Perani, G. (1996), "The Dynamics o f Innovation and Employ
ment: An International Com parison,” Science , Technology Industry R eview 18: 67-93.
Рші, P. (1995), "Econom ic Growth, Technological Change and Employment: Empirical
Evidence for a Cum ulative Growth Model with External Causation for Nine OECD
Countries, 1960-1990,” Stru ctu ral C han ge a n d Econom ic D ynam ics 6: 185-213.
----- (1996), "An Integrated Cum ulative Growth Model: Empirical Evidence for Nine OECD
Countries, 1960-1990,” L a b o u r 10: 93-150.
Piva, M., and V ivarelli , M. (2002), "T h e Skill Bias: Comparative Evidence and an Econo
metric Test,” In tern a tio n al R e v ie w o f A p p lie d Econom ics 16(3): 347-58.
Ricardo, D. (1951), P rin ciples o f P olitical E conom y a n d Taxation, in P. Sraffa (ed.), The Works
and Correspondence o f D a v id R icardo , vol. 1, Cambridge: Cambridge University Press (3rd
edn. 1821).
Sanders, M., and ter W eel , B. (2000), "Skill Biased Technical Change: Theoretical Con
cepts, Empirical Problems and a Survey o f the Evidence,” Druid working paper, Copen
hagen Business School and Aalborg University.
Schumpeter, J. A. (1934), T h eory o f E conom ic D evelopm ent , Cambridge, Mass.: Harvard
University Press (1st edn. 1911).
Shaiken , H. (1984), Work Transform ed: A utom ation a n d L abor in the C om puter Age, New
York: Holt, Rinehart and Winston.
Simonetti, R., T aylor , K., and V ivarelli , M. (2000), "M odelling the Employment Impact
of Innovation: Do Com pensation Mechanisms Work?” in Vivarelli and Pianta 2000.
598 M A R I O PI A N T A
SCIENCE,
TECHNOLOGY,
AND INNOVATION
PO LICY
BENGT-AKE LUNDVALL
SU SAN A BORRAS
22.1 In tr o d u c tio n
2 2 Л Л From G u n s. . .
O n e im p o rta n t early exam p le o f successful tech n o lo g y p o lic y w as th e in itiative by
H e n ry V I II to d evelop the co m p etitive p ro d u c tio n o f can n o n s m ad e o u t o f iro n in
E n g la n d in the first h a lf o f the sixteenth century. T h e reaso n fo r estab lish in g this
tech n o lo g y p o lic y p ro g ra m w as that E n g la n d d esp erately n eed ed m o re can n o n s to
w in the o n g o in g w a r w ith France, and that the b ron ze ca n n o n s at use w ere to o scarce
an d to o exp en sive (Y aku sh iji 1986).
O ne facto r m a k in g the success o f the p ro g ra m p o ssib le w a s th at E n g la n d had
access to iro n ore as w ell as forests fo r fuel. A n o th e r im p o rta n t facto r w as that skilled
p eo p le w h o w ere exp ert forgers h ad m o ved to E n g la n d fro m the C o n tin e n t, esp e
cially fro m France. S o m e cam e because th ey h ad been exp elled becau se o f their
religio u s faith, w h ile oth ers w ere b ro u g h t to E n g la n d becau se o f th eir u n iq u e skills.
T h e p ro g ra m w as realized th ro u g h estab lish in g a c o n so rtiu m u n d er the lead ersh ip o f
W illiam L eavitt, h ead o f the royal iro n w o rk s in N ew b rid g e, and b rin g in g together
exp ertise co m in g fro m d ifferen t co u n tries, especially fo u n d ers w ith th eir ro o ts in
France.
M a n y o f the in stru m en ts o f tech n o lo g y an d in n o v a tio n p o lic y that w ere used at
that tim e are still in use today. T ech n ological d iffu sio n th ro u g h im m ig ra tio n was
im p o rtan t. C o m p e titio n p o lic y an d p ro p e rty rig h t regim es w ere m an ip u lated , and
p u b lic p ro cu re m e n t used. A fter a p erio d w h en several co m p e tin g co m p an ies w ere
allow ed to exp o rt can n o n s (pirates w ere a m o n g the m o re ad van ced users!) m o n o p
olist righ ts w ere gu aran teed to one p ro d u c e r (R alp h H o g g e). B a rrie rs to the m o v e
m en t o f skilled people to the C o n tin en t and e xp o rt licen sin g w ere established to
m ake sure that o n ly P rotestant co u n tries got access to the tech n o logy. Public
p ro cu rem en t w as certain ly a p o lic y in stru m en t that p ro m o te d the refin em en t o f
the tech n o logy, in this case w ith a n ation al secu rity objective.
2 2 Л .2 . . . to Butter
In the m id d le o f the nineteenth ce n tu ry the D an ish econ o m y w as h ig h ly dependent
o n the exp o rt o f co rn to E n glan d . D an ish rye served as the “ fu e l” d riv in g the English
h orse-b ased tran sp o rt system . In the second h a lf o f the ce n tu ry an d especially in the
1870s a d ram atic red u ction in overseas tran sp o rt costs gave R u ssia an d the U S easier
access to this an d oth er m arkets. Prices w ere b ro u g h t d o w n to a level w h ere D anish
fa rm in g co u ld not com pete an ym ore. A m a jo r ag ricu ltu ral crisis b ro k e out.
A t that tim e there w as already som e e xp o rt o f an im al p ro d u cts in clu d in g butter
esp ecially from the m a n o r houses. B u t the m o st im p o rta n t b a rrier w as the uneven
q u ality o f the p ro d u cts. Stan d ard ization w as a key elem ent in the tra n sfo rm a tio n o f
D an ish p ro d u ctio n and exp o rts aw ay fro m co rn to w ard b u tter an d oth er anim al
S C I E N C E , T E C H N O L O G Y , AND I N N O V A T I O N POLICY 6Q1
A s can be seen fro m the oth er chapters in this b o o k — an d fro m the tw o stories told
a b o v e — in n o v a tio n p o lic y covers a w id e range o f in itiatives an d it is n ecessary to give
so m e stru ctu re to the co m p le x reality. O ne w ay to d o so is to in tro d u ce “ ideal types”
that b rin g o u t in a m o re d istin ct fo rm classes o f p h e n o m en a that are m u d d led and
m ixed in the real w o rld .
T h e d istin ctio n to be used here is betw een science p o licy, te c h n o lo g y policy, and
in n o vatio n policy. F o r each w e w ill d iscu ss h o w it b ecam e an exp licit p o lic y field,
w hat the m a jo r issues at stake are, w hat p art o f the in n o v a tio n system is in focus,
w h ich are the m ajo r actors in volved an d w h at are the in stru m en ts used. We w ill
p o in t to the critical sets o f data su p p o rtin g the p o licies. A n d since the O E C D has
played a u n iq u e role a m o n g in tern atio n al o rg an izatio n s in the d iffu sio n o f ideas
ab o u t in n o va tio n p o licy w e w ill use O E C D d o cu m en ts to illu strate so m e o f the m ore
im p o rta n t shifts in the p o st-w a r debate.
It w o u ld be m islead in g to argu e that w e pass fro m science p o lic y to tech n o lo gy
p o lic y an d then to in n o vatio n p o licy as w e pass fro m on e h isto rical stage to another.
For instance som e o f the classical issues in science p o lic y are very h ig h o n the current
p o licy agen da (Pavitt 1995; M artin an d Salter 1996). C u rren t d evelo p m en ts in
b io tech n o lo g y and p h arm aceu ticals have m ad e the d istan ce fro m b asic research to
co m m ercial ap p licatio n m u ch sh orter and therefore the o rg a n iz a tio n o f un i versities
has becom e a m a jo r issue (see C h . 8 b y M o w e ry an d S am p at). G en etic en gin eerin g
S C I E N C E , T E C H N O L O G Y , A N D I N N O V A T I O N P O L I CY
603
The OECD has played a key role in the evolution of the understanding of the policy
fields discussed in this chapter. It is certainly one o f the best sources for internationally
comparable data on science technology and innovation. Data are accessible through
regular publications in the form o f periodical policy reviews and through data bases
that are regularly updated. But it is also interesting to follow the policy discourse
organized at the OECD secretariat. What has been said at OECD meetings and
recommended by its expert groups might not always be transformed into practical
use in member countries but it reflects the new ideas. Therefore a brief summary of the
most important reports (in fact they all tend to arrive at the beginning of a new decade)
will be made here.
OECD 1963: Rationalizing science policy and linking it to economic growth. In an
OECD document from the beginning o f the sixties (OECD 1963a) а shift in attention
toward economic objectives was signaled for science policy. This document with
Christopher Freeman, Raym ond Poignant, and Ingvar Svennilsson as major contribu
tors is quite remarkable in the emphasis it gives to national and rational planning. It
came the same year as the Frascati meeting on a new manual for gathering R&D
statistics took place and it argues for a strong link between better data on R&D and
more systematic policy. This report obviously aimed at giving science policy legitimacy
outside the narrow circles o f ministries o f education and science.
OECD 1970: Bringing in human and social considerations on technology policy.
This report (the “ Brooks Report” ) introduced a broader social and ecological perspec
tive to science and technology policy. It also gave strong emphasis to the need to
involve citizens in assessing the consequences o f developing and using new technolo
gies. The new focus reflected a combination o f growth satiation and dissatisfaction
with the social consequences o f technical change. Therefore it was also assumed that
mom should be given for wider concerns and the uncritical optimism was challenged.
OECD 1980; Innovation policy as an aspect of economic policy. The OECD expert
report “ Technical Change and Econom ic Policy” (OECD 1980) redefined the agenda
for innovation policy in the light “ o f a new economic and social context/ Its message
was that the slow-down o f growth and the increase in unemployment could not be seen
as something that m acroeconom ic expansionary policies could solve. Neither did the
report restrict itself to recommending an increase in investments in science and R&D.
As compared to earlier reports on science and technology, the focus was moved toward
die capacity o f society to absorb new technology. The experts included, among others,
Freeman, Nelson, and Pavitt. It is interesting to note that this report was published
simultaneously with a more traditional document on innovation policy (1980). This
was obviously a period when innovation policy began to be regarded as a legitimate
i k y field.
OECD 1990: Innovation defined as an interactive process. The TEP project, edited
id given its flavor by Francois Chesnais and with Luc Soete as one of the leading
Jemal experts brought together in one coherent framework a broad set of new
search results on innovation and used those to point in new policy directions
604 BE N G T-A K E L U N D V A L L AND S U S A N A BORRAS
contributing to national security, health and economic growth. Like Bernal, the
Bush report gave strong emphasis to the potential economic impact of investments
in science.
The real reason for the breakthrough o f science policy and the increased public
investment in research was probably the way World War II ended and the Cold War
started. The success o f the Los Alam os project made plausible the idea that a massive
investment in science (especially physics but also chemistry and biology), applied
science and technological development could produce solutions to almost any
difficult problem, and underscored the importance o f science and technology for
national security. This pressure to invest in the promotion o f science was reinforced
in the arms and the space races between the US and Soviet Union. In 1957 the launch
of Sputnik put extra pressure on the West, and especially the US, to invest in defense
and space-related research.
The major issues in science policy are about allocating sufficient resources to
science, to distribute them wisely between activities, to make sure that resources are
used efficiently and contribute to social welfare. Therefore, the quantity and quality
of students and researchers receives special attention. The objectives for science
policy that are actually pursued by governments are mixed and include national
prestige and cultural values besides social, national security, and economic
objectives.
The elements o f the innovation system that are focused upon are universities,
research institutions, technological institutes, and R&D laboratories. Science policy
is both about the internal regulation o f these parts o f the innovation system and
about how they link up to the environment— not least to government and industry.
However, strengthening this linkage becomes even more crucial in technology and
innovation policy.
There are two m ore or less standing debates within the science policy community.
The first debate has to do with how far scientific progress is identical with progress in
general. Critical scholars would point to how science is abused in the control o f
people and nature— including genetic manipulation and undermining ecological
sustainability. Those m ore positive to science would respond that none o f this can be
seen as emanating from science: rather, it should be seen as the result o f unwise use of
science. Scholars from hum anities and sociology might more frequently join the first
camp while scientists and technologists have a tendency to join the second.
The second debate is about to what degree science should be made the obedient
servant o f the state and/or capital and to what degree it should be autonomous.
Sociologists o f knowledge have devoted considerable attention to this question, also
in relation to different national research policy styles— with more or less govern
mental steering and with different levels o f aggregation— large/small research activ
ities (Rip and van der M eulen 1997; Jasanoff 1997). Others have argued that there is a
shift in the mode o f research and knowledge production, with different effects upon
the autonomy o f science (Cozzens et al. 1990; Gibbons et al. 1994; Jasanoff 2002).
6o6 BENG T-A KE LU N D V A LL AND SU SA N A BORRAS
Policy makers m ay also see internal evaluation as being too slack. This seems to
have been the case in the U K where a very ambitious and detailed reporting system
has been imposed on scholars doing research and teaching. So far the effect seems to
be a lot o f time used to report on research and that in England university scholars
have become the professional group most unsatisfied with working conditions.
While this kind o f draconian reform linking quantitative indicators o f performance
to access to public m oney might be useful in shaking up conservative institutions for
a brief period they tend to become a nuisance for the innovation system as a whole if
not loosened up again.
One o f the most fundam ental questions in science policy is whether it is true that
“good research is always useful research” or, more demanding, “ the higher the
scientific quality the m ore useful is the research.” I f it were true it would be a strong
argument for leaving at least some o f the allocation to the academic communities.
However, the evidence on this question is contradictory. While Zucker and Darby
(1998) demonstrate that star scientists are important for the success o f biotechnol
ogy firms, Gittelman and Kogut (2003) show that at least within biotech there is no
clear relationship between prestigious publishing and high impact innovations— if
anything the relationship is negative (M owery and Sampat 2001).
It might be expected that the requirements between doing outstanding research
and creating well-designed technology are even more distinct in other fields where
the distance from scientific discovery to innovation is longer. The truth may be that
also science has a lot to learn from interacting with its users and that the best system
is one with several layers and with career shifts— some going only for “ excellence”
within a scientific discipline, others focusing only on user-needs and some operating
in a mode between the two.
Since the early 1990s, basic science has been constantly pushed by politicians to
demonstrate its social and econom ic usefulness. This has been termed the new
social contract o f basic science” in the age o f budgetary retrenchments, particularly
in Europe and the U S (M artin and Salter 1996). Several authors argue that this “ value
for money” attitude disregards some essential aspects o f science policy, such as the
training o f scientists and technicians, and the development o f knowledge capabil
ities in areas where uncertainty about exploitation is so high that private investors
lack incentives (Sharp 2003) and that lessons from the US experience should be
examined critically, and along with for instance the Scandinavian and Swiss experi
ences (Pavitt 2001).
n u clear pow er, space tech n o logy, co m p u te rs, d ru gs an d gen etic en g in eerin g are seen
as b ein g at the v e ry core o f eco n o m ic grow th . T h ese tech n o lo g ies get in to fo cu s fo r
several reasons. O n the o n e h an d th ey stim u late im a g in a tio n b ecau se th e y m ake it
p o ssib le to do su rp risin g things— th ey co m b in e science w ith fiction . O n the o th er
h an d they o p en up n ew co m m ercial o p p o rtu n ities. T h e y are ch aracterized b y a h igh
rate o f in n o va tio n an d they add ress ra p id ly g ro w in g m arkets.
T ech n o lo gy p o lic y m ean s d ifferen t th in gs fo r c a tch in g -u p co u n tries th an it does
fo r h ig h -in c o m e co u n tries an d it m ig h t also m ean d ifferen t th in gs fo r sm all an d b ig
co u n tries. In b ig h ig h -in c o m e co u n tries the fo cu s w ill be o n estab lish in g a ca p a city
in p ro d u cin g the m o st recent scien ce-b ased tech n o lo gies, as w ell as a p p ly in g these
in n o va tio n s. In sm aller co u n tries it m ig h t be a q u estio n ab o u t b ein g able to ab sorb
an d use these tech n o logies as th ey co m e o n the m arket. C a tc h in g -u p co u n tries m ay
m ake efforts to enter into sp ecific p ro m isin g estab lish ed in d u stries u sin g n ew
tech n o lo gies in the p ro cess o f d o in g so.
C o m m o n fo r these strategies is that th ey ten d to d efin e “ strategic tech n o lo gies''
an d so m etim es the sectors p ro d u c in g th em are also d efin ed as strategic sectors. T h e
idea o f strategic sectors m a y be related to P e rro u x and to H irsch m a n , b oth students
o f Sch um p eter. P e rro u x used concep ts such as “ in d u strializin g in d u strie s" and
grow th poles w h ile H irsch m an in tro d u ced u n b alan ced g ro w th as a p o ssib le strategy
fo r less-d evelo p ed co u n tries (P e rro u x 1969; H irsch m a n 1969).
In the lead cou n tries, go vern m en t in itiatives o f the te c h n o lo g y p o lic y k in d w ere
triggered w h en n ation al p o litical o r e co n o m ic interests w ere th reaten ed and the
threats co u ld be lin ked to the co m m a n d o f sp ecific tech nologies. S p u tn ik gave extra
im p etu s to a fo cu s o n space tech n o lo g y an d the C o ld W ar m o tivated the m ost
am b itio u s tech n o lo g y p o lic y effo rt ever in the U S. In E u ro p e, S ervan Schreiber's
b o o k L e Deft a m ericain (Servan Sch reib er 1967) gave a p ictu re o f a g ro w in g d o m in
ance o f the U S m u ltin atio n al firm s especially in h ig h te c h n o lo g y sectors. It gave the
b ig E u ro p ea n co u n tries such as France, the U K , an d G e rm a n y in cen tives to d evelop a
p o lic y o f p ro m o tin g n ation al ch am p io n s in sp ecific sectors. A sp ecific im p o rta n t
event triggerin g French and, later, E u ro p ean efforts w as the e x p o rt em b argo o f
co m p u te r tech n o lo g y that w as seen in France as b lo ck in g its p ro g ress in the d evel
o p m e n t o f nu clear technology.
T h e m o tiv a tio n b eh in d the tech n o lo g y p o licy in Ja p a n — an d later on in co u n tries
such as Taiw an an d K orea— is different. It is d riven b y a n atio n a l strategy a im in g at
catch in g up and in the Jap an ese case it has roo ts b ack to the M eiji rev o lu tio n w hen
the first ideas o f m o d ern izatio n based on im itatin g the te c h n o lo g y o f the W est w ere
fo rm ed .
A t this p o in t w e need to be aw are o f a n u m b er o f fu n d am en tal q u estio n s regard in g
tech n o lo g y policies.
* Is it at all legitimate and effective for the state to intervene for commercial reasons
in promoting specific sectors or technologies? Or is the only legitimate technology
S C I E N C E , T E C HNO LO GY, AND IN NOVATION POLICY 609
policy one where national societal issues are at stake, including establishing
national military power? It is a paradox that in the country having the most
massive public intervention in terms o f technology policy (the US), most o f the
policy has been motivated by non-com m ercial arguments and the discourse has
been anti-state. Japan is the country with the most explicitly commercially driven
technology policy with a recognition o f a role for the state, and here the interven
tion has been much m ore modest, at least in terms o f the amount o f public money
involved.
t What technologies should be supported? Is it always the case that high-tech and
science-based sectors should be given first priority? Again the Japanese govern
ment as well as governments in smaller countries has been more apt to think about
the modernization o f old industries than the US and the big European countries.
* At what stage should the support be given? Should it be given only to “ pre-
competitive” stages or should it also be helpful in bringing the new products to
the market? In the second case there might be a combination o f government
support o f new technology and more or less open protectionism.
* What limits should be set for public sector competence? Technology policy may be
pursued with competence where government operates as a major user but when it
comes to developing new technologies for the market, the role o f governments
must be more modest. To be m ore specific, there are several historical examples o f
how government am bitions to make technological choices that reduce diversity
have ended in failure, for example, the “ minitel” experience in France, and the
high definition T V policy in the EU, both in the early 1990s.
* How can prom oting a technology or a sector best be combined with competition?
The period in the 1980s o f prom oting single firms as national champions in the
bigger European countries was not a great success while the Japanese public
strategy to prom ote “ controlled competition” among a handful o f firms was
more successful.
The objectives o f technology policy are not very different from those o f science
policy but— at least to begin with— it represented a shift from broader philosophical
considerations to a m ore instrumental focus on national prestige and economic
objectives. Technology policies were developed in an era o f technology optimism.
But later on— in the wake o f the 1968 student revolt— more critical and broader
concerns relating to technology assessment and citizen participation came onto the
agenda (OECD 1971).
The elements o f the innovation system in focus remain universities, research
institutions, technological institutes, and R & D laboratories. But the attention
moves from universities toward engineering and from the internal organization o f
universities toward how they link to industry. Technology policy may go even
further and include the com m ercialization o f technologies, but then we approach
what we will call innovation policy.
6 lO B E N G T -A K E L U N D V A L L AND S U S A N A BORRAS
B o th o f these ap p ro ach es co ver all aspects o f the in n o v a tio n pro cess— in clu d in g
d iffu sio n , use an d m ark etin g o f n ew tech n o lo gies— an d in a sense th e y m a y be seen
as an im p o rta n t fo rm o f “ e co n o m ic p o lic y ” w h ere the fo cu s is m o re o n in n o v a tio n
th an on allo catio n . B o th ten d to p u t stro n g er em p h asis o n “ in stitu tio n s” and
“ o rg a n iz a tio n s” th an d o science and tec h n o lo g y policy. In the laissez-faire version ,
the p re d o m in a n ce o f the m ark et an d o f co m p e titio n b eco m es the m o st im p o rta n t
p rereq u isite fo r in n o v a tio n — there is in p rin cip le o n e sin gle re c o m m e n d a tio n fo r
in stitu tio n al d esign v a lid fo r all co u n tries.
In the system ic a p p ro ach the im p o rta n ce o f co m p e titio n is reco gn ized b u t so is
the need fo r clo ser co o p era tio n v ertically betw een users an d p ro d u c e rs an d so m e
tim es even h o rizo n ta lly a m o n g co m p etito rs w h en it co m es to d evelo p generic
tech n ologies. In the system s a p p ro a ch it is reco gn ized that the in stitu tio n a l set-up
d iffers acro ss n ation al eco n o m ies an d that this has im p lic a tio n s fo r w h at types o f
tech n o lo gies an d sectors thrive in the n atio n al con text. To d esign a su itab le in n o v
atio n p o lic y req u ires specific in sigh ts in the in stitu tio n a l ch aracteristics o f the
n atio n a l system .
In n o v a tio n p o lic y does n o t im p ly an y a p rio ri preferen ce fo r h igh versu s lo w
tech nology. T h e system s a p p ro ach in tro d u ces a vertical p ersp ective o n the in d u strial
system , seeing it as a n etw ork an d as valu e ch ains w h ere certain stages m ig h t be m ore
su itable fo r firm s in a sp ecific c o u n try
T h e respective theoretical fo u n d atio n s o f the tw o d ifferen t v ersio n s o f in n o vatio n
p o lic y are (1) an a p p lica tio n o f stan d ard n eoclassical e co n o m ics on in n o vatio n ,
and (2) a lo n g -te rm o u tco m e o f research o n in n o v a tio n an d e c o n o m ic evo lu tio n
(M etcalfe 1995; M etcalfe an d G e o rg h io u 1998). T h e in n o v a tio n system a p p ro a ch m ay
be seen as b rin g in g together the m o st im p o rta n t stylized facts o f in n o va tio n . It
m akes use o f em p irical m aterial an d an alytical m o d els d evelo p ed in in n o vatio n
research, as w ell as in in stitu tion al and e v o lu tio n a ry econ o m ics.
T h e m a jo r reason fo r in n o va tio n p o lic y b eco m in g m o re b ro a d ly used as a concept
w as the slo w -d o w n in e co n o m ic grow th a ro u n d 1970 and the persisten ce o f sluggish
grow th as co m p a red to the first p o st-w a r decades. T h e reason s fo r the slo w -d o w n in
the grow th in “ total facto r p ro d u c tiv ity ” w ere, an d still are, n o t w ell u n d ersto o d but
there w as a feelin g that it h ad to do w ith a lack o f c a p a b ility to e xp lo it tech n ological
o p p o rtu n ities. A t the sam e tim e, the restriction s im p o sed o n gen eral econ o m ic
p o lic y b y fear o f in flatio n m ad e it im p o rta n t to u n d erstan d the po ssib ilities to
p ro m o te grow th fro m the su p p ly side.
T h is im plies that the m ajo r o bjectives o f in n o va tio n p o licy are e c o n o m ic grow th
and in tern atio n al com petitiven ess. In the E u ro p ean U n io n d isco u rse these o b je ct
ives are co m b in ed w ith “ social co h esio n ” an d equality. In n o v a tio n m ig h t also be
seen as a w ay to solve im p o rta n t p ro b lem s relating to p o llu tio n , energy, u rb an ism ,
an d poverty. B u t the m ain focus is o n the creatio n o f e co n o m ic w ealth.
A m o n g the in stru m en ts to be used are the reg u latio n o f intellectu al p ro p erty
rights and access to ven tu re capital. One fu n d am en tal d istin ctio n in in n o vatio n
S C I E N C E , T E C H N O L O G Y , AND INNOVATION POLICY
613
policy lies between initiatives aim ing at promoting innovation within the insti
tutional context and those aim ing at changing the institutional context in order to
promote innovation. The first category overlaps with instruments used in science
and technology policy. The second m ay include reforms o f universities, education,
labor markets, capital markets, regulated industries, and competition laws.
One fundamental question is how far these other policy fields can and should be
adapted to the needs to prom ote innovation. Universities and schools obviously
have other missions than to support innovation and economic growth. And the
614 BENGT-AKE LUNDVALL AND SUSANA BOREAS
Science policy
Focus: P ro d u c tio n o f s c ie n tific know ledge
In s tru m e n ts :
*P ubiic research funds granted in com petition
• (Semi-*) Public researchinstitutions (i,e:
laboratories, universities, Research centers..,)
• Tax incentives to firm s
• H igher education
• In te lle c tu a l Property Rights
Technology policy
Focus: A dvancem ent and co m m ercialization o f sectorial
te ch n ica l know ledge
In s tru m e n ts :
•P u b lic procurem ent
• P ublic aid to s tra te g ic sectors
• Bridging in s titu tio n s (between research w orld and industry)
• Labor force tra in in g and im provem ent o f technical skills
• S tandardization
• Technology forecasting
•B enchm arking in d u stria l sectors
Innovation policy
Instruments:
• Im proving individual skills and learning a b ilitie s (through general education system and labour training)
• Im proving o rg a nizational perform ance and learning (i,e. ISO 9000 standards, quality control, etc.)
•Im proving access to in fo rm a tio n : in fo rm a tio n society
■E nvironm ental regulation
• B ioethical regulation
• Corporate law
•C o m p e titio n re g ulation
• Consumer p ro te ctio n
• Im proving social ca p ita l fo r regional developm ent: Clusters and industrial districts
• In te llig e n t bench m arking
• In te llig e n t, reflexive and dem ocratic forecasting
It is characteristic for the evolution o f new policy fields that measurement and
quantitative guidance gives more legitim acy to the field. Within the field what can be
біб B E N G T - АКБ L U N D V A L L A N D S U S A N A B O R R A S
Community innovation surveys have been collected several times in most o f the
member states (see Ch. 6 by Smith in this volume).
Among the m ore interesting inform ation that can be obtained through these
surveys is the share o f new products in total sales in firms in different sectors and
countries. This is a measure o f diffusion o f product inno vations in the economy and
may be seen as an im portant intermediate performance variable. Another field
where the diffusion o f technology has been mapped quite thoroughly is in relation
to information and com m unication technologies. These indicators are important
since for innovation policy one m ajor performance indicator should be the diffusion
and effective use o f new technologies.
The most im portant rem aining tasks for building indicators to support innov
ation policy relate to the diffusion o f process innovations, innovation in services,
organizational innovations— and their diffusion— and, finally, to experience-based
learning. Even with better indicators in these fields, we cannot expect to get very
simple and clear conclusions from quantitative evaluation exercises.
Therefore case studies bringing together qualitative and quantitative information
and dialogue with policy practitioners will remain important sources o f insight
when designing policy. Richard Nelson has more than any other scholar developed
this approach (Nelson, Peck, and Kalachek 1967; Nelson 1982, 1984, 1988, 1993).
N o t least, in the era o f in n o v a tio n p olicy, w h ere in stitu tio n s m atter m o re th an ever, it
is d ifficu lt to see h o w q u an titative an alysis co u ld stan d alon e as the b asis fo r policy.
Robustness: Decisions and social structures should withstand the occurrence o f differ
ent future scenarios.
Flexibility. In the occurrence o f sudden socio-economic change institutions must be
able to change direction rapidly
Internal diversity Structurally dissimilar characteristics must be built in to allow
survival if the selection environment changes.
External diversity Variety o f links to different kinds o f agents will help adaptation when
change in the environment arises.
Window of opportunity: Attention to timing and sequence in face o f path-dependent
systemic context.
Incremental approach: The whole can be changed only through the cumulative impact
o f small steps.
Experimentation and prudence, new policy ideas should be submitted to trial in
localized contexts before full deployment.
Source Sandro Mendon^a
S C I E N C E , T E C H N O L O G Y , A ND I N N O V A T I O N P O L I C Y 6l9
have helped the creation o f standards, have induced more risk-taking attitudes o f
innovators, have fostered long-term rather than short-term strategies in firms’
research, and have enhanced the acquisition o f new skills and knowledge (Peterson
and Sharp 1998). This has also been framed as the problem o f attribution. Closely
related is the problem o f synchronizing the evaluators’ time horizon with the
political time horizon o f the political consumers o f their work. Big programs’ effects
may not be realized for years if not decades.
p rivate sector d riv in g an d d efin in g the policy. T h ese co m p lexes w ere o p e ra ted w ith
little c o o rd in a tio n betw een them and they w en t all the w a y fro m p ro c u re m e n t o f
sp ecific tech nical system s an d c o m p o n e n ts to the su p p o rt o f b asic research and
research train in g . In particu lar, the defense b u d g et w as u sed fo r p ro m o tin g activities
w ith little d irect co n n ectio n to sh o rt-te rm m ilita ry needs. R esearch in c o m p u te r
science an d softw are got su b stan tial su p p o rt fro m this b u d g et an d p ro d u c e d quite
gen eric kn o w led ge, in a d d itio n to p ro d u c in g sh o rt-te rm so lu tio n s to m ilita ry needs
(L an glo is and M o w e ry 1996).
O n e m a jo r issue in relatio n to the p u b lic m issio n te c h n o lo g y p o lic y is the lack o f
c o o rd in a tio n . O n e negative aspect is the resu ltin g b ia s in the d ire c tio n o f m ilita ry
an d space exp en d itu re. It seem s as i f it is m u ch easier to m o b ilize ta xp ayers' m o n ey
an d d evelo p m en t efforts to go to the m o o n than it is to solve the p ro b le m o f the
gh etto (N elso n 1977). In p art, this reflects the presence o f w e ll-o rg an iz e d lo b b ies and
interest g ro u p s a m o n g p rivate sector b en eficiaries; it also reflects the sh eer difficulty,
n o ted b y N elson , o f "'so lvin g5' such in tractab le p ro b lem s as u rb an p o v e rty o r p rim a ry
ed u catio n al ach ievem en t. A n o th e r p ro b lem is that the calcu latio n o f exp ected costs
an d benefits o f sp ecific p ro jects m ay either be co m p letely n eglected becau se n ation al
p rid e is at stake o r be system atically biased d o w n w ard s to m ak e a p ro ject lo o k m ore
attractive. A th ird w eakness m a y be a o n e-sid ed fo cu s o n the d ev elo p m en t o f science-
based in d u stries an d tech n o logies an d a general a ssu m p tio n that science and
tech n o lo g y is a q u ick fix fo r all k in d s o f p ro b lem s. M o re m u n d a n e in d u stries aim in g
at co n su m ers, an d p ro b lem s w h ere the so lu tio n s are less g la m o ro u s in tech n o logical
term s, m a y be neglected.
B u t the U S -ty p e n atio n al in n o v a tio n system certain ly h as its stro n g p o in ts b eyon d
its sheer scale advantages. T h e fact that several o f the agen cies are prep ared to finance
m o re o r less generic research an d research tra in in g w h ile still h a v in g the use o f
research in m in d ten ds to o verco m e the bias tow ard the su p p ly side. It tends to
establish a "ch a in -lin k e d m o d el o f in n o v a tio n 5w ith stro n g feed b ack elem ents (K line
an d R o sen b erg 1986). O ne stro n g elem ent is d iversity. T h e fact that d ifferen t agencies
"c o m p e te 55 in fu n d in g g o o d research m a y b e b en eficial in su p p o rtin g the d iv ersity o f
research efforts. T h e coexisten ce o f b ig p rivate fo u n d a tio n s that c o m m it resources
to research team s o n the basis o f their track record also co n trib u tes to diversity.
In these respects cu rren t E u ro p ean centralistic in itiatives in research p o lic y in
co n n ectio n w ith the sixth fra m e w o rk p ro g ra m and the E u ro p ea n R esearch Era
co u ld learn fro m the U S, w h ich ben efits not o n ly fro m scale b u t p erh a p s even more
fro m diversity.
objective. This is different when we look at technology policy in Japan. More than
perhaps any other market econom y Japan has made use o f an explicit national policy
to promote specific sectors and industries with the ultimate aim o f stimulating
economic growth and competitiveness. According to Freeman (1987), the radical
character of the policy had to do with the fact that since the mid-1950s policy ended
up being designed by experts with an engineering background in the Ministry of
Technology (MITI) who were much less concerned with “comparative advantage”
than the economists in the Bank o f Japan.
But the M ITI has not acted alone. In the area o f telecommunications, NTT__a
public company in m onopolistic control o f telecommunication services— has
played an important role in coordinating technology development efforts in major
electronic firms such as Hitachi and N EC. The policies pursued have not involved
massive subsidies and, actually, the public sector has been much less involved in
financing R&D in the private sector than has been the case in the US (Nelson 1984).
The strategic prom otion o f the car industry, consumer electronics, and “mega-
tronics” typically com bined different policy instruments such as subsidies to re
search and development o f generic technologies with elements o f “ infant-industry”
protection. Bringing together competing firms in consortia aiming at solving
common problems has been an im portant role for M ITI. This has been done on
the basis o f attempts to map new trends in technology and markets through, for
instance, technology forecasting.
One interesting aspect o f M IT Is technology policy has been that it did not
focus exclusively on high technology sectors. For instance, consortia initiated by
MITI aiming at prom oting the modernization o f textile and clothing industries
brought together firm s producing textiles and textile machinery with electronics
firms.
the 1950s, w h ereas the m a in th ru st cam e in the 1970s. T h is fig u re d oes n o t in clu d e the
E u ro p ean stan d ard izatio n b o d ies (E T S I, C E N , an d C E N E L E C 2) an d the E u ro p ean
agen cies gran tin g in tellectu al p ro p e rty righ ts (like E P O , O H IM , an d C o m m u n ity
P lan t V ariety O ffice3), w h ich h ave also played im p o rta n t roles in p ro m o tin g scien
tific, tech n o lo gical, an d in n o vative synergies.
C E R N is the oldest an d o n e o f the m o st successful o rg a n iz a tio n s, an d is dedicated
to n u clear research. Its u n d en iab le scien tific success co n trasts sh a rp ly w ith the
tro u b les o f its T w in ,” the JR C (Jo in t R esearch C e n te r), u n d er the E u ro p e a n C o m
m u n ities an d also d ed icated to n u clear e n erg y research. D isp u te s b etw een France
and G e rm a n y ab o u t n u clear reacto r d esigns in the 1960s w eak en ed this in stitu tion ,
u n til the lau n ch o f the Jo in t E u ro p ea n T orus (JE T ) co n cern ed w ith n u clear fu sio n
(G u zzetti 1995). O th er E u ro p ea n b u t n o n -E U scien tific an d tech n o lo g ica l o rg an iza
tio n s em erged in the 1960s an d 1970s, a m o n g th em the E u ro p ea n S o u th ern L a b o ra
to r y (A stro n o m y ), the E u ro p ean M o lecu la r B io lo g y L a b o ra to ry (E M B L ), the
E u ro p ea n Space A g en cy (E S A ), the E u ro p ea n Science F o u n d a tio n (E S F ), gathering
n atio n al research co u n cils, and A irb u s, a p u b lic -p riv a te en terp rise w ith heavy
in vestm en t in a v ia tio n tech n ology. A ll these b o d ies are in te rn a tio n a lly con stituted
an d d irected tow ard o n e sp ecific scie n tific -te c h n o lo g ic a l field.
In 1971, C O S T (C o o p e ra tio n in Science an d T ech n o logy) w as estab lish ed as an
in tergo vern m en tal p ro g ra m . T h e n o velty w ith C O S T w as that it covered several
AIRBUS
ESA (1970)
European Union (1975) EADS +
25 states 15 states British
aerospace
RTD Framework
Program
EUREKA
JET JRC (1985)
33 states + EU
ESF
(1974)
76 institutions
29 states
Europe and Japan are prepared to compete „ ,,h the US in becomtng centers to,
gtobal interaction wtth the res,. The more positive scenario » „ uld he one" h ”
dte three poles used therr capaatres promote ,h , drfhtsion and use of k n o„led «
,0 less pnvtleged parts o f the world. In anp case, the eaten, „ hich there i, 'ro
globalization o f knowledge production and diffusion, and the different form i
assumes, ts a topic that has started be analyzed (Arch,bug, and Michie , J
Georghiou 1998), but deserves m ore attention. 997>
22*5 C o n c lu sio n s
It should be clear from what has been said that the most pressing issues on the policy
agenda are specific to each national system. Even so, there are some issues that are
common for all countries, and these represent new research challenges for scholars
in this field. We have presented a stylized sequence, beginning with science and
ending with innovation policy. In the most recent debates about the learning
economy and the knowledge-based society we can see the contours o f a new policy
that we might call “ knowledge policy.” 4 It recognizes that innovation and compe
tence building involve m any different sources o f knowledge and that innovation
itself is a learning process. This raises the need for new analytical efforts and for
rethinking the organization and implementation o f policy in several respects.
It is increasingly im portant to understand better the connection between science
and technology on the one hand and economic performance on the other. The rise
and fall o f the new econom y demonstrates that assumptions about simple and direct
connections are problematic. Between the new technologies and the performance o f
the economy the organizational characteristics o f innovation systems and firms
including “ slippery” elements such as “ social capital” affect the impact. This is an
issue that remains understudied. In terms o f public policy there is a need for
innovative thinking about how governments can support the diffusion o f good
and sustainable practices in cooperation with management and employees. In
terms of research opportunities, this links with the importance and need to devote
more analytical efforts exam ining how technical innovation interacts with organiza
tional change. The academic traditions o f business organization and o f innovation
systems’ research have to come closer to each other in order to answer questions
regarding how organizational change affects innovation processes in the economy.
A second issue is about aggregate demand. In a period o f growing fears for
deflation and with little room left for expansionary monetary policy it might be
useful to reconsider what Keynesian policy could mean in a knowledge-based
6 l6 B ENGT - AKE L UNDVALL AND S US ANA BORRAS
N otes
l Ergas has suggested the distin ction betw een “ m issio n -o rien ted " and “ d iffu sio n -o rie n te d "
policy designs, based respecti vely on the m assive su p p o rt to a sm all n u m b er o f scien ce-
tech nology fields, and on su p p o rtin g the scientific-technological in frastru ctu re (Ergas
S C I E N C E , T E C H N O L O G Y , AND I NN O V A T I O N POLICY 627
1987), This analytical distinction has also been used when comparing policy styles in
science and technology policies across countries. (It has been argued that the Japanese and
French policies are mission-oriented, whereas the German policy has been depicted as an
example o f a diffusion-oriented policy style.) Again it is important to note that what is
referred to is ideal types and that simple groupings o f this kind might miss some o f the
most important complexities in the set-up o f national systems o f innovation.
2. ETSI: European Telecommunications Standards Institute; CEN: Comite Europeen de
normalisation; and CEN ELEC : Com ite Europeen de Normalisation Electrotechnique.
These three bodies have divided their standardization activities by technological sectors.
3. EPO: European Patent Office; OHIM ; Office for the Harmonization o f the Internal
Market; dedicated respectively to patents, and to trademarks and other intellectual
property rights.
4. The concept o f knowledge policy has evolved in connection with European policy making
and as a follow-up to the Lisbon ministerial meeting 2000. One o f the architects behind
the lisbon-strategy, M aria Rodrigues, defines knowledge policies as 'policies aimed at
fostering and shaping the transition to a knowledge-based society” (Conce^aio, Heitor,
and Lundvall 2003: p. xx).
R e f e r e n c e s
Andreasen, L., C oriat , B,, D en H artog , R, and K aplinsky , R. (eds.) (1995), Europe's N ext
Step— O rganisational In n ovatio n , Com petition a n d Em ploym ent , London: Frank Cass.
Archibugi, D., and M ich ie , J. (1997), Technology Globalisation a n d Econom ic Performance,
Cambridge: Cambridge University Press.
----- and L u n d v a ll , B.-A. (eds.) (2000), The G lobalizing Learning Econom y: M ajo r Socio
economic Trends a n d E uropean In n ovation Policy >Oxford: Oxford University Press.
A ro cen a , R., and S u t z , J. (20 0 0 ), “ Interactive Learning Places and Development Policies in
Latin America,” D RU ID Working Paper no. 13.
Arrow, K. J. (1971), “ Political and Economic Evaluation o f Social Effects and Externalities,
in M. Intrilligator (ed.), Frontiers o f Q uantitative Econom ics , North Holland, 3-23.
Arundel, A., and Soete , L. (eds.) (1993), A n Integrated Approach to European Innovation and
Technology D iffusion Policy: A M aastrich t M em o ran d u m , Brussels. CEC: Sprint.
Be r n a l , J. D. (1939), The Social Function o f Science , London: Routledge & Kegan Paul
*Biegelbauer, P., and B orras, S. (eds.) (2003), Innovation Policies in Europe a n d the US. The
New Agenda , Aldershot: Ashgate,
Borras , S. (2003), The In n ovatio n Policy o f the EU, Cheltenham: Edward Elgar.
C a n t n e r , U., and P y k a , A. (2001), “ Classifying Technology Policy from an Evolutionary
Perspective” Research P olicy 30: 759-75**
C o lem a n , J. (1988) “ Social Capital in the Creation o f Human Capital,” A m erican Jou rn al o f
Sociology 94 (Suppl.): 95-120.
C onceicao , R , H e i t o r , M ., and L u n d v a ll , В.-A. (eds.) (2003), Innovation , Competence
Building a n d Social Cohesion in E urope: Towards a L ea rn in g Society >Cheltenham. Edward
Elgar.
Hirschman , A. О (1969) >The Strategy of Economic Development, New Heaven; Yale Univer
sity Press.
Jasanoff, S. (ed.) (i997)> Comparative Science and Technology Polity Cheltenham; Edward
Elgar.
*■---- (ed-) (2002), Handbook of Science and Technology Studies, 2nd edn., London: Sage.
Katzenstein, P, J. (1985), Small States in World Markets. Industrial Policy in Europe, New
York: Cornell University Press.
K u n e , S. J., and R o s e n b e r g , N. (1986), “An Overview of Innovation,” in R. Landau and
N. Rosenberg (eds.), The Positive Sum Game, Washington, DC: National Academy Press,
275-305.
Kristensen , P. H., and S t a n k ie w ic z , R. (eds.) (1982), Technology Policy and Industrial
Development in Scandinavia, Lund: Research Policy Unit.
K rull , W., and M e y e r - K r a m e r , E (eds.) (1996), Science and Technology in Germany
London; Cartemills.
K u enh , T. J., and P o r t e r , A. L. (eds.) (1981), Science, Technology and National Policy
London: Cornell University Press.
K u t zn e t s , S, (1960), “Economic Growth of Small Nations,” in E.A.G. Robinson (ed.),
Economic Consequences of the Size of Nations, Proceedings of a Conference held by the
International Economic Association, London: Macmillan.
La n g lo is , R., and M o w e r y , D. C. (1996), “The Federal Role in the Development of
American Computer Software Industry: An Assessment,” in D. C. Mowery (ed.), The
International Computer Software Industry, New York: Oxford University Press.
L ist , F. (1841), Das Nationale System der Politischen Okonomie, Basel: Kyklos (trans. and pub.
as The National о
System of Political Economy by Longmans, Green and Co., London, 1841).
Lu n d v a ll , B.-A. (1985), Product Innovation and User-Producer Interaction^ Aalborg: Aalborg
University Press.
---- (1988), “Innovation as an Interactive Process: From User-Producer Interaction to the
National Innovation Systems,” in G. Dosi, C. Freeman, R. R, Nelson, G. Silverberg, and
L, Soete (eds.), Technology and Economic Theory London: Pinter, 349-69.
-----(ed.) (1992), National Innovation Systems: Towards a Theory of Innovation and Inter-
active Learning, London: Pinter.
---- (1996), “The Social Dimension of the Learning Economy,” DRUID Working Paper, No.
1.
—— (2002), Innovation, Growth and Social Cohesion: The Danish Modeh Cheltenham:
Edward Elgar.
---- and J o h n s o n , B. (1994), “The Learning Economy,” Journal of Industry Studies 1(2):
23-42.
*— —and B o r r a s , S. (1998), The Globalising Learning Economy Implications for Innovation
Policy., Brussels: European Commission.
-----and T o m l in s o n , M. (2001), “Policy Learning by Benchmarking National Systems of
Competence Building and Innovation,” in G. P. Sweeney (ed.) Innovation, Economic
Progress and Quality of Life, Cheltenham: Edward Elgar.
----- Mendon^a, S. (2003), ‘Politica de Ciencia, Technologia e Inova^ao: Prindpios,
tendencias e questoes,’ mimeo, ISCTE University.
Luukkonen, T. (1998), “The Difficulties in Assessing the Impact of EU Framework
Programmes,” Research Policy 27(6): 599-610.
630 BEN G T-A KE LUND VA LL AND SUSA NA BORRAS
*-— (2001), “ Public Policies to Support Basic Research: What Can the Rest o f the World
Learn from the US Theory and Practice (And What they Should not Learn);’ Industrial
and Corporate C han ge 10(3): 761-79.
Per ez, C. (2002), Technological R evolutions a n d Fin a n cia l C apital, Cheltenham: Edward
Elgar.
Per ko u x, F. (1969)» L ” Econom ic du zoem e siecle , Paris; Presses universitaires de France.
Peter so n , J. (i99i)> “Technology Policy in Europe: Explaining the Framework Programme
and Eureka in Theory and Practice,” Jo u rn a l o f Com m on M arket Studies 29(3): 269-90,
,
------ and S h a r p , M. (1998), Technology Policy in the European Union London: Macmillan.
Pu tn am , R. D. (i993)> M a k in g D em ocracy Work: C ivic Traditions in M odern Italy., Princeton:
Princeton University Press.
R ip , A., and v a n d e r M e u l e n , B. J. R. (1997), “The Post-Modern Research System” in
R. Barre, M. Gibbons, ]. Maddox, B. Martin, and R Papon (eds), Science in Tomorrow's
Europe, Paris: Economica International, 51-67.
R omer , P. M, (1990), “Endogenous Technological Change,” Jou rn al o f Political Econom y 98:
71-102.
R o se n b e r g , N. (1972), Technology a n d A m erican G row th , New York: M. E. Sharpe.
Ro th w ell , R., and Z e g v e l d , W. (1980), Innovation a n d Technology Policy, London: Pinter.
S h a r p , M. (2003), “The UK-Experiment: Science, Technology and Industrial Policy in
and technological specialization 554-7 Court of Appeals for the Federal Circuit
and innovation 87,122, 545, 546, 561, 583 (CAFC) 228-9, 274, 277
and innovation policy 612, 618 craft labor, and innovation 35-7
and intellectual property rights 266» 274» 275, Cray Systems 368
276 creative destruction 32,105,115,126,359,382,
and Japanese policy 621 503
measurement o f 544 credit:
as national goal 544 and innovation 242, 243, 244
at national level: rationing of 521
and capital accumulation 547, 548-50 Crevoisier, О 311
and currency values 546 critical technology 349-50
and innovation 546-50 cross-shareholding, and Japanese industrial
and international trade 545-7 organization 42
and technology gap 547 currency, and competitiveness 546
and prices 546,581
and product and process innovation 573 da Vinci, Leonardo 5
at regional level 294,301, 303, 557-9 Darby, M 297, 607
research opportunities in 562-3 David, Paul 213,463,473-4
revival of interest in 543-4 Dawkins, Richard 495
and science parks 226, 303 decision-making:
sources o f 561 and adoption o f new technology 465-6
and speed to market 425 and finance 246
and United States 620-г in firms 39,42
Computer Associates International 48 and innovative investment 257-8
computer industry: organizational 123,124
and coevolutionary processes 397 and radical innovation 108
post-war development 368-71 and scientific knowledge 606
sectoral structure 393 DeFillippi, R 127,130
and standard setting 68 Dell Computer 48,49
Computer Research Associates 368 demand, and sectoral innovation systems 391
Conlisk, J 500-1 Denmark 428
Constant, E, and knowledge production 97 and collaborative networks 197
consultancy, and systems o f innovation 191 and comparative advantage 409-10
contingency theory, and organizational and cooperative movement 601
structure 117-21 and innovation policy 614
contract manufacturing 99 and national system o f innovation 199
control, and ownership 38 and patent laws 189
convergence: and research and development 194
and catch-up 514-15, 523 and vocational and technical education
conditional 506 194
cross-country 554-5, 557 developing countries:
and growth rates 505 and catch-up 520,523* 525.535»53^, 537
and intellectual property rights regimes and diffusion 460
273 and education 553-4
technological 91-2, 99,337 and employment 583
Cooke, P, and regional innovation systems 184, and foreign direct investment 332,338
300-5 and intellectual property rights 271, 273, 275,
Cooperation in science and technology 284
(COST) 622-3 and low- and medium-tech industries 410,
cooperative movement, Denmark 601 428—9
core competence 93,126 ,134 and science, technology, and innovation
Cornwall,! 4 9 h 523 policy 601, 626
corporate change 418-21 development blocks 389» 429
corporate finance 250-5 developmental state, and catch-up 51b, 534
636 INDEX