Content uploaded by Walter W. Powell
Author content
All content in this area was uploaded by Walter W. Powell
Content may be subject to copyright.
Regional Studies, Vol. 36.3, pp. 291–305, 2002
The Spatial Clustering of Science and Capital:
Accounting for Biotech Firm–Venture Capital
Relationships
WA LT E R W. P OW E L L , * K E N N E T H W. K O P U T,† J A M E S I . B OW I E ‡ and
L AU R E L S M I T H - D O E R R §
*509 CER A S, S C A N CO R Building, Stanford University, Stanford, C A 94305, USA. Email: woodyp@ stanford.edu
†405 McClelland Hall, Department of Management and Policy, University of Arizona, Tucson, AZ 85721, US A.
Email: kkoput@ u.arizona.edu
‡Department of Sociology, University of Arizona, Tucson A Z 85721, U S A.
§Department of Sociology, Boston University, Cummington Street, Boston, MA 02215, U SA
(Received June 2001; in revised form October 2001)
POW E L L W. W., KO P U T K. W., BOW I E J. I. and S M I T H - DO ER R L. (2002) The spatial clustering of science and capital:
accounting for biotech rm–venture capital relationships, Reg. Studies 36, 291–305. This paper focuses on the spatial
concentration of two essential factors of production in the commercial eld of biotechnology: ideas and money. The location
of both research-intensive biotech rms and the venture capital rms that fund biotech is highly clustered in a handful of key
US regions. The commercialization of a new medicine and the nancing of a high-risk start-up rm are both activities that
have an identi able timeline, and often involve collaboration with multiple participants. The importance of tacit knowledge,
face-to-face contact, and the ability to learn and manage across multiple projects are critical reasons for the continuing
importance of geographic propinquity in biotech. Over the period 1988–99, more than half of the U S biotech rms received
locally-based venture funding. Those rms receiving non-local support were older, larger and had moved research projects
further along the commercialization process. Similarly, as venture capital rms grow older and bigger, they invest in more non-
local rms. But these patterns have a strong regional basis, with notable diVerences between Boston, New York and West Coast
money. Biotechnology is unusual in its dual dependence on basic science and venture nancing; other elds in which product
development is not as dependent on the underlying science may have diVerent spatial patterns.
Biotechnology Venture capital Networks Spatial agglomeration
POW E L L W. W., KO PUT K. W., B OW I E J. I. et SMI T H - POWE L L W. W., KO PU T K. W., B O W I E J. I. und SM I TH -
DO ER R L. (2002) Le regroupement ge
´ographique de la DO R R S L. (2002) Ra
¨umliche Konzentration von Wissen-
science et du capital: comment expliquer les rapports entre schaft und Kapital: Versuch einer Erkla
¨rung der Beziehung
les entreprises du secteur de la biotechnologie et les socie
´te
´s zwischen Biotechnologie und Risikokapital, Reg. Studies 36,
de capital risque, Reg. Studies 36, 291–305. Cet article porte 291–305. Dieser Aufsatz befaßt sich mit der ra
¨umlichen
sur la concentration spatiale de deux facteurs de production Konzentration zweier wesentlicher Faktoren bei der Produk-
cle
´s dans le domaine commercial de la biotechnologie: a
`tion auf dem kommerziellen Gebiet der Biotechnologie:
savoir, les ide
´es et l’argent. La localisation et des entreprises Ideen und Geldmittel. In den USA treten Standorte
a` fort e inten site´ de recherche-de´veloppement du secteur de forschungsint ensiver Biotechno logie rmen und der Ri siko-
la biotechnologie, et les socie
´te
´s de capital-risque qui n- kapitalunternehmen, die ihre nanzielle Grundlage bereit-
ancent la biotechnologie, s’ave
`re tre
`s concentre
´e dans une stellen, stark geha
¨uft in wenigen Schlu
¨sselregionen der USA
poigne
´e de re
´gions cle
´aux E-U. La commercialisation d’un auf. Die Kommerzialisierung eines neuen Arzneimittels und
nouveau me
´dicament et le nancement d’une cre
´ation d’en- die Fina nzierung eines hochriskanten ‘start-up’s’ stellen Akti -
treprise a
`haut risque sont, tous les deux, des activite
´s qui vita
¨ten in einer klaren zeitlichen Abfolge dar, die oft die
ont une date limite, et ne
´cessitent souvent une collaboration Zusammenarbeit mit mehreren Teilnehmern verlangt. ‘Tacit
avec de multiples partenaires. L’importance de la connaissance knowledge’, perso
¨nlicher Kontakt und die Fa
¨higkeit, zu
tacite, du contact direct et de la capacite
´d’apprendre et lernen und mit mehreren Projekten gleichzeitig zurecht zu
d’administrer de multiples projets sont des raisons essentielles kommen, sind dabei entscheidende Gru
¨nde fu
¨r die Bedeu-
pour l’importance continuelle de la proximite
´ge
´ographique tung ra
¨umlicher Na
¨he im biotechnologischen Bereich. Im
dans la biotechnologie. Sur la pe
´riode de 1988 a
`1999, plus Zeitraum 1988 – 1999 wurden mehr als der Ha
¨lfte der
de la moitie
´des entreprises ame
´ricaines du secteur de la Biotechnik rmen der USA Finanzmittel von lokalen Risiko-
biotechnolog ie ont pro te
´du capit al-risque l ocal. Les kapitalunter nehmen zur Verfu
¨gung gestellt. Firmen , die nicht
entreprises qui be
´ne
´ cient du soutien externe e
´taient plus lokale Unterstu
¨tzung genossen, waren a
¨lter, gro
¨ßer und
vieilles, plus grandes, et ont fait plus avancer davantage des hatten ihre Forschungsprojekte bereits weitgehend kommer-
0034-3404 print/1360-0 591 online/02/030291-15 ©2002 Reg ional Studies Association DOI: 10.108 0/00343400220122089
http://www.region al-studies-assoc.ac.uk
292 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
projets de recherche le long du processus de commercialis- zialisiert. Auch Risikokapitalunternehmen investieren mit
ation. De la me
ˆme manie
`re, les socie
´te
´s de capital-risque zunehmendem Alter und zunehmender Gro
¨ße mehr in
investissent pluto
ˆt dans des entreprises externes, au fur et a
`nicht–lokale Fir men. Doch diese Muster variieren regional
mesure qu’elles vieillissent et s’agrandissent. Les fondements deutlich und zeigen klare Unterschiede zwischen Bosten,
d’une telle distribution s’ave
`rent fortement re
´gionaux, avec New York und den Finanzinstituten der Westku
¨ste. Biotech-
de notables diVe
´rences pour ce qui est de l’argent provenant nologie ist ungewo
¨hnlich in ihrer Doppelabha
¨ngigkeit von
de Boston, de New York et de la Coˆ te de l’ouest. La Grundlagenwissenschaften und der Finanzierung durch
biotechnolog ie est hors du commun e
´tant donne
´sa double Risikokapital; andere Sektoren, in denen die Produktion-
de
´pendance de la science de base et du capital-risque; il se sentwicklung nicht so stark von Grundlagenforschung
peut que d’autres domaines ou
`le de
´veloppement des produits abha
¨ngt, mo
¨gen durchaus andere ra
¨umliche Muster
ne de
´pend pas de la science sous-jacente aient une distribu- aufweisen.
tion ge
´ographique diVe
´rente.
Biotechnolog ie Risikokapital Netzwerke
Biotechnologie Capital-risque Re
´seaux Ra
¨umliche Konzentration
Agglome
´ration ge
´ographique
I N T RO D U C T I O N on rms that provide venture capital to our sample of
biotech companies. Venture capital is also spatially
concentrated in the Bay Area, Boston and New York.Our focus is on the relationships between dedicated
biotechnology companies and the venture capital rms We use descriptive statistics to analyse whether the
linkages between biotech and venture capital arethat nance them. These are, in a sense, unusual
relationships in that they are designed with a termina- exclusively local, have a local component or are non-
local.tion point in mind, at which time the venture capitalist
exits and moves on. Nor are they exclusive relation-
ships. A venture capitalist is likely to invest in many T H E C O - L O C AT I O N O F S C I E N C E
diVerent biotech rms, including some which are likely A N D C A P I TA L
to be competitors in a particular therapeutic area, such
We take as our starting point the spatial concentration
as cardiology, or with a particular technology, such as
of two key factors of production in the commercial
genomics. Biotech rms may well have backing from
eld of biotechnology: ideas and money. Casual obser-
multiple venture capitalists, either as part of a collective,
vers might wonder why these two endowments, which
such as a group or syndicate, or separately as a means
are highly fungible, easily transportable, in short,
to nance discrete projects, such as a specialized use of
weightless (LE A D B EA T E R, 2000), are so strongly con-
a more general-purpose technology. Biotech r ms also
centrated regionally. Abundant evidence points to the
garner nancial support from multiple sources, through
clustering of both knowledge and capital.
government research grants, R &D alliances with major
Ideas, especially knowledge from the frontiers of
corporations and selling minority equity stakes. For a
cutting-edge science, have a strong tacit dimension
biotech rm to become nancially successful, it needs
(NE L SO N and WI N T ER , 1982). When knowledge is
to develop a promising pipeline with numerous new
more tacit in character, face-to-face communication
medicines. Each potential product is, in some respects,
and interaction are important (V O N HI PP E L , 1994).
a separate project that involves diVerent internal staV
Consequently, to understand the science, one has to
and disparate external collaborators. At a venture rm,
participate in its development. Hence new scienti c
a portfolio of investments is developed with divergent
advances have a form of natural excludability (Z UCK E R
levels of risk, diVerent timelines and varied expected
et al., 1998). In the early years of the biotechnology
payoVs. For both biotechs and venture rms, learning
industry, rms were founded in close proximity to
across partners and projects, and developing experience
research institutes and universities where the advances
working with diverse parties, is critical to success
in basic science were being made (K E N N E Y, 1986;
(POW E L L et al., 1996).
AU DR ET S C H and S TE P H A N , 1996; PR E V EZ E R , 1996;
We analyse the spatial aspects of these relationships,
ZUC K E R et al, 1998). There are two key elements to
examining how the role of location shifts over time as
this clustering process. One aspect is captured by
projects, r ms and regions mature. Our data are drawn
research on knowledge spillovers, where geographic
from the commercial eld of human biotechnology,
proximity facilitates the spread of innovative ideas
speci cally the wave of founding of new biotech rms
( JA F F E et al., 1993; AU D R ET S C H and FE L DM A N ,
in the US over the period 1988–99. This eld is
1996). But while intellectual capital is necessary, it may
remarkably clustered spatially, with over 48% of all US
not be suYcient. A supportive institutional infrastruc-
rms located in either Northern California, the Boston
ture that fosters knowledge transfer and the formationMetropolitan area or San Diego County. We map the
of technology-based companies is also critical (POW-industry’s growth, showing a pattern of cluster-based
proliferation. We match our biotech data to a data set EL L , 1996).
The Spatial Clustering of Science and Capital 293
Consider the case of Atlanta, Georgia, where there AU D R ET S C H and F EL D M A N , 1996, p. 634, put the
question aptly: ‘even after accounting for the geo-is a major research centre, the Center for Disease
Control, a technology-based university, Georgia Tech, graphic concentration of the production location, why
does the propensity for innovative activity to clusterand one of the top medical schools in the country at
Emory University. The metropolitan area is reasonably vary across industries?’ The relevant scienti c expertise
in biotech is, by now, broadly distributed throughoutwell-to-do and well-educated, and a number of Fortune
500 rms are headquartered there. But there is little in the industrial world, with major centres of scienti c
excellence in the U S, the UK, Sweden, France, Ger-the way of commercial biotechnology, despite abundant
intellectual resources. One biomedical entrepreneur at many and Switzerland. But the science is commercial-
ized by rms in a signi cant manner (by which weGeorg ia Tech told us that he has had numerous over-
tures from nanciers and angel investors for his techno- mean the ability to bring novel medicines to a global
marketplace) in only a handful of locations worldwide.logies, but they have all made leaving Atlanta and
moving to California a requirement of obtaining the To understand this phenomenon, we have to explain
why some regions are hubs for organizational creation, nancing.
Or consider the often-cited list of founders of some that is, populated, by organizations, that are in the
business of creating other organizations (S T I N C H-of the key rms created in the late 1970s and 1980s:
Genentech (Herbert Boyer, University of California, CO M B E, 1965). Put diVerently, some regions are incu-
bators and constitute an ecology for organizationalSan Francisco); Biogen (Walter Gilbert, Harvard
University); Hybritech (Ivar Royston, University of formation (BR OW N , 2000). These regions have a rich
mix of diverse kinds of organizations (e.g. universities,California, San Diego); Genetics Institute (Mark
Ptashne, Harvard University); Systemix (David Balti- law rms specializing in intellectual property, public
research institutes, consultants and venture capitalists)more, Massachusetts Institute of Technology and
Whitehead Institute); and Immulogic (Malcolm Gefter, that contribute in varying ways to founding techno-
logy-based companies. The advantages of location,Massachusetts Institute of Technology).1All of these
eminent scientists retained their university aYliations, then, are very much based on access and information.
Increasing retur ns are present in the form of over-often full-time. They were able, so to speak, to have
their cake and eat it too, precisely because their univer- lapping networks, recombinant projects, personal and
professional relationships, and interpersonal trust andsities had created rules and routines that enabled
technology transfer and faculty entrepreneurship. There reputation, all of which are thickened over time. In
such a milieu, access to reliable information about neware many regions where there is scienti c excellence
but not the requisite infrastructure to capture the rents opportunities occurs through personal and professional
networks, and these ties are critical in reducing uncer-from knowledge spillovers.
Our emphasis on this infrastructure of university tainty about projects that are not well understood by
non-experts, exceedingly risky in terms of their payoVtechnology transfer, venture capital, law rms, consul-
tants and the like is somewhat diVerent from treatments and unclear in terms of their eventual market impact.
Venture capital (V C), de ned as ‘independent,of industrial districts, in the tradition of M ARSHALL,
1920. Ec onomists and geog raphers have lon g recognized professionally man aged, dedicated pools of capital that
focus on equity or equity-linked investments inthe tendency for production to cohere geographically,
whether it is cars in Detroit, steel in the Ruhr, silk in privately held, high growth companies’ (G O M P E R S
and LE R NE R , 2001, p. 146), is one of the key elementsLyon or lmmaking in Hollywood. Spatial concentra-
tion confers advantages in terms of transportation costs, of the infrastructure of innovation. The private equity
market has become a major source of nancing foraccess to skilled labour markets, communication net-
works, sophisticated customers and access to technology start-up rms, and has grown at an explosive rate; in
1979 venture rms dispersed US$500 million in funds,(S CO TT and S TOR P E R , 1987; FL O RI D A and KE N NE Y,
1988; A N G E L , 1991; S A X E N IA N , 1994; S T O R P E R and that amount climbing to well over $67 billion by
2000 (W R I G H T and RO B B I E, 1998; GO M P E RS andSA LA I S , 1997). Once these agglomeration economies
are established, a dynamic process of increasing returns LE R N E R, 2001). Both venture capital rms and ven-
ture capital investing are highly concentrated regionally.attracts new entrants, further fuelling the pace of
innovation (AR T H U R, 1990; KR UG MA N , 1991). For example, in the third quarter of 2000, as the global
slowdown in technology companies became moreConsequently, the geographic clustering of production
is a global phenomenon. (PO R TE R , 1998, provides pronounced, V Cs still poured $8·7 billion into new
companies located in Northern California. This sumnumerous examples.)
Our emphasis is less on the process of economizing represented 33·7% of the total US venture capital pie
for that period for all industries, according to Ventureon the transaction costs of founding a new rm, or the
many attractions that draw entrepreneurs to a region. Economics, a rm that tracks V C investing (SI N T ON ,
2000). In 1999, a little more than one-third of allWe are interested in understanding why rms – based
on a fast-moving science that is continually creating venture capital disbursements went to California
(GO M P ER S and LER N E R , 2001).new opportunities – are for med in particular locales.
294 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
A venture capital rm raises money from wealthy unplanned encounters at restaurants or coVee shops,
opportunities to confer in the grandstands during Littleindividuals, pension funds, nancial institutions, insur-
ance companies and other sources that are interested League baseball games or at soccer matches, or news
about a seminar or presentation all happen routinely inin investing in technology-based start-ups, but lack the
ability to do so. These investors become limited part- such settings. The combined impact of access to ‘news’
and more eVective monitoring help explain the patternners in the VC fund, while the partners in the VC
rm manage the money by investing in and advising of V C clustering.
With all these advantages of geographic propinquity,entrepreneurial start-ups. Venture capitalists nance
new rms with the potential for high growth in return it might seem unlikely that more distant relations occur
at all. There are, to be sure, several ways that V Csfor partial ownership. When the young company is
suYciently developed, the rm goes public through an overcome some of the liabilities of distance. Both the
creation of branch oYces and involvement in VCinitial public oVering (I P O) or is acquired by another
company. At this point the V C cashes in its ownership syndicates are means to counter the challenges of more
distant relations (S O R E N S E N and ST UA RT , 2001).stake, and reaps its rewards. Venture capital obviates the
need to grow slowly via self- nancing, and fuels more Increased size and greater experience could also provide
VC rms with the capability to support more distantrapid growth. As F R E E M A N, 1999, puts it, venture
capitalists buy time. The success of a V C rm in rms. V C rms may follow diVerent approaches when
they are investing their own money versus that ofattracting money is contingent on its past track record
of spotting winners and generating rewards for its limited partners, or when they join another V C’s fund
as a member of a syndicate. In addition, the pace oflimited partners. The business of identifying opportun-
ities is highly uncertain and diYcult. Of course, V Cs advancement of new industries and the mix of rms
within them may oVer new opportunities for invest-receive innumerable proposals for new businesses. But
the rejection rate for these proposals is extremely high ment. For example, VCs may perform a diVerent role
with an early-stage company than in a rm that has(estimated by S A H L M A N , 1990, to be at 99%). As in
many other walks of life, many call but few are already undergone its rst round of nancing and
shown evidence that its technology can be brought toanswered. More opportunities are identi ed through
active search by V Cs. In part, this is because the market. We turn now to a discussion of the factors that
shape the proclivity of biotech–V C relations to occurexpected pay-oVdemanded from V C backing is very
high and the ratio of success to failures about two in on a local or more distant basis.
ten (BYG R AV E and T IM M O N S , 1992; GO M PE R S and
LER N E R , 1999). E X P LA I N I N G C E N T R E A N D
In the life sciences and other technology-based elds, P E R I P H E RY
venture rms provide more than money. Because many
of the founders of biotech rms are research scientists, The literature on knowledge spillovers provides useful
leads on both how and when geographic localizationventure capitalists often do much more than monitor
or advise; they may even play a hands-on role in the matters.2One insight is that the importance of propin-
quity can decline over time. JA F F E et al., 1993, reportrunning of the young company. Keeping scientists
focused on key commercial milestones is no small feat. that patent citations to other patents (excluding within-
organization citations) are ve to ten times more likelyA powerful tool for focusing their attention is the
‘staging’ of V C nancing, thus the commitment of to occur within the same city. This pattern of localiza-
tion is most pronounced in the rst year following acapital is contingent upon ‘progress’ (G OM P E R S ,
1995). V Cs also routinely help in recr uiting key staVpatent’s issue, and subsequently declines. In a parallel
vein, they also found that patents in such fast-and important collaborator s, and provide referrals to
law and accounting rms, and eventually to investment developing elds as optics and nuclear technology have
high initial citation rates that fade rapidly. A L M E ID Abanks (F L O R ID A and K E N N E Y, 1988). Many V Cs
serve on the boards of directors of young rms they and KO G U T , 1997, report similar results for patenting
activity in the semiconductor industry, with high ratesfund. As G I L S O N and BLA C K , 1998, put it, ‘by provid-
ing both money and advice, the venture capitalist puts of local citations that subside over time.
The joint eVects of technological evolution and theits money where its mouth is’. Obviously, the roles of
monitoring, advising and managing are much more stages in a r m’s life cycle are not easily disentangled,
however. Two excellent studies of biotechnology pointeasily accomplished when the young rm is located
nearby. Experienced VCs have abundant contacts and out this diYculty. ZU CK E R et al., 1998, show that
the founding of new biotechnology rms in the 1970sdeep knowledge of particular industries; thus, referrals
to relevant sources of expertise are another important and 1980s occurred in those regions rich in the
relevant intellectual capital, and that ‘star’ scientists hadresource they provide. This social network is also
more readily tapped when rms are geographically a direct role in this process as founders and advisors.
AU DR ET S C H and S T E P H AN , 1996, examine a sampleproximate. Finally, there are real advantages that accrue
to r ms and venture capitalists to being ‘on the scene’ – of biotech rms at the time of their initial public
The Spatial Clustering of Science and Capital 295
oVerings in the early 1990s and analyse the geographic new and unproven technologies, previous aYliations
can serve as a proxy for quality (P O D O L N Y, 1994).location of founders and members of scienti c advisory
boards. They nd considerable geog raphic reach in the Not surprisingly, start-up companies go to considerable
lengths to advertise the backing of elite venture rmscomposition of advisory boards, but somewhat closer
linkages when scientists are involved as founders. This to attract employees and collaborators. In short, social
relationships are essential to the process of garneringcomparison raises two questions: (1) is the contrast
between the studies a consequence of diVerences in resources to found new organizations.
But can aYliations compensate for less expertiseroles, i.e. an advisory role involves less direct engage-
ment and can be accomplished from a distance, while or capability? Alternatively, can organizations that are
pursuing excellent science, but located away from keya founder’s role entails more hands-on involvement,
requiring the proximity of a scientist’s rm and labora- centres of activity and lacking access to well-connected
parties, nd much-needed support? Clearly, centralitytory; and (2) do the diVerent ndings re ect distinct
stages in the development of a company, with founding in networks and expertise are self-reinforcing (S TUART,
1998). But at what point are there diminishing returnsa time when new ideas are being explored among a
select few, and the I PO stage a point when patent to network centrality or local connectivity? We examine
these issues about the dynamics of centre and peripheryrights for these ideas have been secured and the rm is
ready to reveal to the public a good deal of information by addressing the following empirical questions:
about itself in order to obtain funds? An additional 1. To what extent are biotech rms and VC rms
complication is that not only are the rms under study co-located?
at diVerent stages in their life cycle, the industry and 2. How extensive is the phenomenon of regional
the nature of technological progress were at diVerent co-location, such that biotechs receive support from
points in their development. local V Cs and V Cs nance local biotechs?
To pursue the latter issue, regarding distinctive stages 3. What is the relationship between location of funding
in organizational, industry and technolog ical life cycles, and characteristics of both biotechs and VCs in
we explore whether biotech rms and venture capital ter ms of age, size, and centrality in the network?
funders are more likely to be co-located when the 4. How do the above patterns and relationships change
biotechs are younger and/or smaller. If biotech rms over time?
are able to wait until they are older and/or larger before
securing venture support, they may well be able to
choose from a broader set of nancial backers. We also DAT A S O U R C E S
explore the other side of this coin, recognizing that
just as biotech rms search for private equity, venture Our starting point in gathering data on biotech com-
panies is BioScan, an independent industry directorycapitalists look for new technologies to bankroll. Thus,
we ask, under what circumstances do venture rms founded in 1988 and published six times a year, which
covers a wide range of organizations in the life scienceslook outside their local environments?
There is an unexplored nding in the AU D RE T S C H eld.3We sample companies that are independently
operated, pro t-seeking entities involved in humanand S TE P HA N , 1996, study that intrigues us, suggesting
that the relevant actors in diVerent locales have diVerent therapeutic and diagnostic applications of biotechno-
logy. Our focus is on dedicated human biotech rms.‘propensities’ to either search locally or at a distance.
University scientists in Boston, the Bay Area and San Both privately-held and publicly-traded rms are
included in the sample. Companies involved in veterin-Diego that served on biotech advisory boards were
very likely to do so locally, while scientists in New ary and agricultural biotech, both of which draw on
diVerent scienti c capabilities and operate in a muchYork, Los Angeles, Maryland and Houston served on
the boards of more distant companies. Such variation diVerent regulatory climate, are omitted. We do not
include large pharmaceutical corporations, health carein search behaviour may re ect diVerences in access to
contacts or diVerent resource endowments. These are companies, hospitals, universities or research institutes
in our primary database; these participants enter theissues at the heart of research on inter-organizational
exchange. One strand of analysis emphasizes that inter- database as partners that collaborate with dedicated
biotech rms. Companies that are wholly-owned subsi-organizational ties are strongly in uenced by social
structure, with previous exchanges shaping subsequent diaries of other rms are excluded. We do, however,
include publicly-held biotech rms that have minorityties (G R A N OV E T T E R , 1985; G U L AT I , 1995). Organi-
zations privileged by prior access obtain better rates of or majority investments in them by other rms, as long
as the company’s stock continues to be independently nancing (U Z Z I , 1999) and overcome liabilities of
newness more easily (BAU M and OL I VE R , 1991). traded on the market. Our rationale for excluding
both small subsidiaries and large, diversi ed chemical,When organizations share a common prior partner,
they nd it easier to engage in exchange (G ULA T I and medical or pharmaceutical corporations in the primary
database is that the for mer do not make decisionsGA R G I U LO, 1999). And, when there is uncertainty
about the merits of an activity, as is often the case with autonomously, while biotechnology may represent only
296 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
a minority of the activities of the latter. Both circum- oYce and $1 million under management, to much
larger rms like Boston’s Advent International, withstances generate serious data ambiguities.
The sample covers 482 rms over the 12-year period, 16 worldwide oYces managing $4 billion. The sample
of VCs includes the Silicon Valley household name1988–99. In 1988, there were 253 rms meeting our
sample criteria. During the next 12 years, 229 rms Kleiner, Perkins, Cau eld, and Byers, as well as smaller,
less-known rms such as Hook Partners of Dallas,were founded and entered the database; 91 (of the 482)
exited due to failure, departure from the industry or Texas. In addition, we include the venture capital
arms of more traditional nancial institutions, such asmerger. The database, like the industry, is heavily
centred in the U S, although in recent years there has NationsBank and J. P. Morgan. The oldest rm in the
sample is Scotland’s Standard Life Investments, foundedbeen expansion in Europe. In 1999, 80% of the com-
panies in our sample were located in the U S and 10% in 1825; in 1999, nine new rms entered the database.
in Europe. For the pur poses of this paper, we limit the
sample to US-based companies because of the ease of M E T H O DS
using U S zip codes as a means to determine geographic
location. During the period 1988–99, 213 U S biotech Our objectives are to establish the co-location of
biotech rms and VCs, to explore how geographical rms received funds from venture capital companies.
The reference source BioScan reports information agglomeration in uences whether V C nancing of
biotech rms is done locally or non-locally, and toon a r m’s ownership, formal contractual linkages to
collaborators, products and current research. In addi- demonstrate the relationship between the locality of
capital and characteristics of both the biotech rms andtion, detailed information is provided on a company’s
nancial history, and we drew from this source data on VCs. We use descriptive statistics to accomplish these
objectives, comparing both VCs and the biotechs theyventure capital investments in speci c biotech compan-
ies. We also utilize data on the founding date and fund based on their location, stage of development and
the nature of the funding relationships.employment levels of biotech companies. Our database
draws on BioScan’s April issue, in which new informa- To identify location, we use postal zip codes for U S
rms and telephone country pre xes for those VCstion is added for each calendar year.
For information on venture capital forms, we con- located outside the US. Using these codes, we exam-
ined frequencies of rms and V Cs by location, identify-sulted Pratt’s Guide to Venture Capital Sources, a reference
guide to U S and non-US VC rms. The guide was ing nine areas with signi cant agglomeration of either
VC or biotech rms. These nine agglomeration clusters rst published in 1970, followed by new editions in
1972, 1974 and 1977. Since the fth edition, it has include: (1) Boston; (2) the N Y C tri-state region,
including parts of New Jersey and Connecticut; (3)been updated annually, based on information provided
by the V C rms. In addition to information on the Philadelphia; (4) the District of Columbia region,
including part of Maryland proximate to the Nationallocation of home and branch oYces, key staVand
founding dates, the guide covers VC rms’ preferences Institutes of Health (N I H); (5) Chicago; (6) Houston;
(7) San Diego; (8) the San Francisco Bay Area, includ-in ter ms of their preferred role in nancing, the type
of nancing they provide, and whether they have ing Berkeley, Oakland and Silicon Valley; and (9)
Seattle. Each biotech r m and VC was then assignedgeographic or industry preferences. The guide also
reports the amount of capital the VC rm manages, a cluster code equal to the agglomeration region it was
in, if any, or ‘0’ if the rm or V C was located elsewhere.and whether the rm primarily invests money raised
from limited partners or its own money. The 1999 For each biotech–VC dyad, we de ne the funding as
local if the rm and V C are within a one-hour driveedition reports that ‘the VC rms included have been
selected because they are devoted primarily to venture of one another (by automobile, using Yahoo’s estimated
driving time between zip codes). nancing’, and it goes on to remark on the expansion
of V C-type activity by a wide range of diVerent Each biotech r m is then placed into one of three
mutually exclusive categories based on whether it isorganizations: ‘today, venture investment activity covers
a spectrum of interests that encompasses all phases of only involved in dyads with local V Cs, only involved
in dyads with non-local VCs, or involved in dyads withbusiness growth’. Pratt’s Guide adopts a more restrictive
de nition of venture capital investors than does BioScan, both local and non-local V Cs. We do this separately for
when the biotech rm is at two distinct stages ofwhich groups angel investors, pension funds and uni-
versity technology oYces under the category of development, before and after its initial public oVering
(IPO). For each biotech rm, we also measure ainvestors. We utilized the Pratt’s de nition because we
want to focus on those companies that are most number of rm attributes, including: its age, experience
in the industry’s inter-organizational network (con-oriented towards high-risk, high-involvement, early-
stage investment in entrepreneurial start-up r ms. necting biotech with universities, government agencies,
nanciers, nonpro t labs, and large phar maceutical andThere are 208 venture rms that nance the biotechs
in our sample. They vary in size from small rms such chemical corporations); number of employees; time
from founding to IPO; time from its r st network tieas Allergan Capital of Irvine, California, with one
The Spatial Clustering of Science and Capital 297
to IPO; number of VC partners; number of other Diego County as the three largest hubs, and smaller
partners (besides V C); and centrality in four inter- centres in the New York metropolitan area (including
organizational networks – R &D, nance, licensing and the tri-state area of Northern New Jersey, western
commercialization. Connecticut and the suburbs of New York City) and
Each VC rm is also placed into one of the three the area around the National Institutes of Health in
exclusive categories based on whether it only funds Rockville, Maryland.
local biotech rms, only funds non-local biotech rms, The map of venture capital rms that invest in
or funds both local and non-local biotech rms. We biotech, presented in Fig. 2, also shows regional con-
do this assignment separately for funded biotech r ms centration, but with some notable geographic diVer-
that are pre- and post-I PO. For each V C, we also have ences. Again the Bay Area and Boston are the two
measures of age, number of oYces, capitalization, and dominant areas, with Menlo Park, C A, far and away
whether it is primarily investing its founders’ own the most active location of all. But New York is third
money or other investors’ money. and San Diego’s position much smaller, a reversal of
their rol es i n the biotech world, re ecting N ew York’s
pre-eminence as a nancial centre. Several other areas
R E S U LT S are signi cant with respect to venture capital – Cleve-
land, Los Angeles, Minneapolis and Chicago, but these
We begin with a graphic presentation of the location are areas with scant biotech activity. And in 1988, there
of our samples of biotech and V C rms. Our biotech
are areas with some biotech rms – such as Seattle,
database starts with the year 1988. The oldest rm in
Philadelphia, Madison, W I, Atlanta, Miami, FL – with
our sample at that point is a Northern California
no local venture capital presence.
company, Alza, founded in 1968. The rst biotechno-
Fast forward to 1998 and you can see the growth of
logy rm to go public was Genentech in 1980. So
the biotech industry, accompanied by only modest
Fig. 1, which shows the location of rms by zip code,
geographic expansion. The growth is pronounced in
is a map of the industry in its adolescent stage. The
Boston, where newspaper accounts now routinely
larger the dots, the more rms located in that zip code.
cheer its advance on the Bay Area as the most active
These maps are simple counts of the number of rms
locale for biotech.4The Bay Area and San Diego grow
in an area, and not selected for rm size or market
rapidly as well, but so does the Philadelphia area, the
value. There is a strong pattern of spatial clustering,
with the Bay Area, the greater Boston area and San Washington-Baltimore corridor, northern New Jersey,
0 300miles
0 500kms
Fig. 1. US dedicated biotechnology rms, 1988
298 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
0 300miles
0 500kms
Fig. 2. U S vent ure capital rms fu nding biot ech, 1988
and the Research Triangle of North Carolina on the both VCs and bio rms in such circumstances need to
hunt externally for partners. At the same time, theeast coast, and the Houston area in Texas. Further west,
Boulder, CO, Salt Lake City, Utah, and especially most active areas are likely to be magnets for outside
investors, while rms seek support wherever capital isSeattle emerge as smaller hubs. But the overall pattern
is one of cluster-based growth. As the number of available. We turn now to an examination of the
biotech-venture capital relationships that result frombiotech rms in our sample climbs by 146, the percent-
age of U S companies located outside the main regional the simultaneous searching of biotechs for funds and
VCs for oppor tunities.clusters remains steady at approximately 28%.
Venture capital took oVdramatically in the 1990s. For the entire time period, 213 biotech rms have
relationships with V Cs that meet Pratt’s criteria. TheGO M P E R S and L ER N E R , 2001, report that there were
34 funds in 1991 and 228 in 2000. Fig. 4 portrays the number of biotech r ms nanced by V Cs grows,
almost monotonically, from 27 in 1988 to 118 in 1999,VC rms that funded biotech companies in 1997, and
shows massive growth in the Bay Area, and along the with a dip in 1997. Of these rms, 54% of the biotech
rms received local V C support at some point. Thisnorth-east corridor from Washington to Boston. There
still remain several ‘mismatches’, however, that is gure varies by location and over time. Among biotech
rms located in a cluster, 58% have funding from aregions with V Cs but little biotech (Chicago, Cleve-
land, St Louis); areas with very active biotech but local V C at some point, compared to only 48% for
rms outside of any single cluster. The percentage ofnot a great preponderance of venture capital (Seattle,
Research Triangle, even San Diego has much more VC-backed biotechs with local funding ranges over
time from 33% in 1988 to over 62% in the mid 1990s,biotech); and areas with no VC but some biotech (Salt
Lake City, Atlanta, Madison). before settling back to 48% in 1999.
On the V C side, 208 VCs provide funds to ourThe maps presented above help frame our presenta-
tion of the ndings. There are a handful of locales subsample of U S-based biotech rms, with 50% of
those VCs funding biotechs that are local. This percent-abundant in rms and venture capital, and three of these
regions have ourished with this propitious situation for age is slightly higher when VCs are funding post-IPO
(52%), is higher for V Cs located in one of the clustersmuch longer than a decade. Other regional centres do
not enjoy a comparably rich co-location of capital (54%), and rises signi cantly over our period of obser-
vation, starting at just 30% in 1988.and science. Many parts of the U S have only one
endowment – money or rms – but not both. Clearly We now examine features of biotech rms that
The Spatial Clustering of Science and Capital 299
0 300miles
0 500kms
Fig. 3. US dedicated biotechnology rms, 1998
Fig. 4. U S vent ure capital rms fu nding biot ech, 1997
300 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
Table 1. Means and standard deviations (in parentheses) for with ‘outside’ VC nancing take the longest time to
go public – 6·5 years.
biotech rms receivi ng V C fun ding prior to I P O, by locality
of funding Those r ms at the pre-IPO stage with only local
VC backing have a diVerent pro le. These are the
Non-local Local Both local smallest of the three types in terms of number of
funding funding and non-
employees, but have the largest percentage of staVwith
Variable only only local funding
PhDs and/or M Ds. These biotech rms go public
Firm characteristics rapidly, on average in 4·7 years. They also have much
Age 5·5913 5·1852 4·6411 more exclusive relations with venture rms, having
(2·7813) (4·2678) (2·6174)
Time to IPO from 1·78 funders, compared to 2·6 for the non-local
founding date 6·5000 4·7273 5·2188 biotechs and 4·3 for those with both local and outside
(in years) (3·2027) (1·6787) (2·1211) nancing. The latter group apparently are high-pro le
Time since rst tie 4·4754 4·5185 3·9625 companies. Not only do they attract both sources of
(in years) (2·7131) (4·3688) (1·8327) funds, they are the youngest as well, only 4·6 years on
Time to I PO from rst 5·0588 3·6364 4·6250
tie (in years) (2·8491) (1·2863) (2·0439) average. The locally-backed rms have a strong scient-
Number of employees 53·81 44·04 53·13 i c pro le, suggesting a research orientation and a need
(43·17) (35·73) (37 ·06) for management assistance and oversight that is best
Number of P hDs/M Ds 15·24 16·70 15·35 provided by local V Cs. The more exclusive ties to one
(9·29) (16·03) (10·74) or two V Cs also suggests the V Cs are more involved
Partner counts in the managing of the rm.
Number of Pratt’s V Cs 2·6073 1·7778 4·2635 Turning to companies at the post-I PO stage, there
funding (1·9230) (1·1956) (2·1281) are 57 with external V C links, 14 with only local
Number of non-D B F 8·6208 6·7451 9·3368
partners (4·7759) (4·0578) (4·8503) support and 62 with both sources. Not surprisingly,
Number of D BF partners 0·8283 0·5170 0·5965 these post-I PO rms are considerably larger, as one
(1·3283) (0·8447) (0·7559) would expect from companies that are older with more
Number of types of ties 2·0955 1·9556 1·9264 nancial security. But again those with only local
(0·7488) (0·8233) (0·7172) funding are notably smaller, and with a higher percent-
Number of forms of 3·2017 2·4353 2·8714
partners (1·3736) (1·0084) (1·2699) age of staVwith advanced science degrees. The local-
only rms had much more exclusive relations ties to
Centrality measures VCs, with 1·2, while those with both sources had
R&D centrality 0·0035 0.0008 0.0022
(0·0052) (0·0030) (0·0044) nearly four V C funders.
Finance centrality 0·0072 0·0030 0·0082 Of the 208 V Cs that fund biotech rms, 178 of
(0·0079) (0·0039) (0·0080) them nance biotech rms before their IP O, while
Licensing centrality 0·0022 0·0014 0·0015 152 provide funding for subsequent rounds of nancing
(0·0052) (0·0041) (0·0036) to publicly held rms. Obviously, most V Cs do both
Commerce centrality 0·0004 0·0038 0·0006
(0·0014) (0·0002) (0·0022) kinds of disbursements. The features of the VCs vary
Number of D BFs 69 27 56 with both locality and the pre- vs. post-IP O distinc-
tion. When backing is provided prior to the biotech
rm’s I P O, the V Cs funding locally are about two
years older (14 vs. 12) and larger in terms of oYces
receive funding from V Cs, treating rms that are pre- (1·9 vs. 1·7), but have less capital ($229 million vs.
and post- I PO separately. Table 1 presents data on $336 million), and are more likely to spend their own
biotech rms with support from venture capital in money (84% vs. 65%) when compared to V Cs that
advance of going public. We group the results into fund non-local biotechs. When the support comes after
three categories: companies with non-local V C suppor t the biotech rm’s IPO, the story is more complicated.
only (of which there are 69); companies with just local Those rms that provide backing exclusively locally or
support (27 in total); and companies with both local exclusively non-locally are about the same size (1·5
and non-local backing (56). We compare rms with oYces), age (roughly 12 years) and capitalization, but
these three kinds of funding arrangements in terms of those going local only are more likely to be spending
their size, age, number of scienti c staVand a host of their own money (81% vs. 60%). Those V Cs that
measures that capture varying forms of connectivity support publicly held rms both locally and non-locally
within the industry. Those companies that secure only are much older (17·3 years), larger (two oYces), more
non-local nance are, on average, larger, older and capitalized ($388 million) and are even more likely to
have a larger number of collaborations with diverse be spending their own money (87%). Thus, older,
types of organizations, suggesting that these collabora- more experienced venture capital rms that have the
tions may be both a signal to attract V C support and/ bene ts of being located in technology-rich locations
or a vehicle for obtaining other kinds of resources in are able to be more exible as to where they invest. In
addition, a strong persistent nding is that when theadvance of securing V C backing. Most notably, rms
The Spatial Clustering of Science and Capital 301
Table 2. Means and standard deviations (in parentheses) for Table 3. Means and standard deviations (in parentheses) for
V Cs funding pre- and post-I PO biot ech rms, by lo calit y ofbiotec h rms receiv ing V C funding after I P O, by locality of
funding funding
Non-local Local Both local Both local
Non-local Local and non-localfunding funding and non-
Variable only only local funding Variable funding only funding only funding
Firm cha racteristics Fun ding pre -IP O r ms
Age 12·3553 14·0408 15·6180Age 8·6101 8·5048 7·4516
(3·4951) (3·1038) (2·6016) (10·1932) (19·6373) (8·1896)
Number of oYces 1·6942 1·915 1·9018Time to IPO from 4·7857 5·2143 4·3387
founding date (1·0512) (1·3619) (1·2266)
Capital (US$ millions) 336·1133 228·5154 262·6174(in years) (3·1143) (2·7506) (2·3881)
Time since rst tie 6·6871 7·2190 6·5161 (852·3067) (440·1693) (210·3292)
% spending own money 64·77 83·72 82·98(in years) (3·0893) (3·3885) (2·4761)
Time to I P O from rst 2·8772 3·9286 3·4032 Number of V Cs 88 43 47
tie (in years) (3·2518) (3·0751) (2·1838) Funding post-I PO rms
Number of employees 164·58 128·92 173·48 Age 12·4262 11·6874 17·3147
(204·38) (135·30) (161·66) (7·1343) (7·2631) (9·0252)
Number of P h Ds/M Ds 26·68 29·30 31·85 Number of oYces 1·5124 1·4835 1·9370
(22·99) (27·22) (24·96) (0·6832) (0·7757) (1·5462)
Capital (US$ millions) 185·6892 210·2204 388·9044Partner counts
Number of Pratt’s V Cs 2·0161 1·2500 3·9734 (307·6961) (382·2478) (692·3246)
% spending own money 59·46 81·25 86·86funding (1·5298) (0·8026) (2·3466)
Number of non-D B F 11·7545 14·2679 13·2397 Number of V Cs 74 32 46
partners (6·2974) (9·5956) (6·6534)
Number of D BF
partners 1·5837 0·8095 1·3628
(1·7746) (0·9582) (1·5293) is a recursive relationship: as the biotech industry
Number of types of ties 2·8053 2·9167 2·6648 matures, the signi cance of geographic proximity
(0·9538) (0·6626) (0·6581)
Number of forms of 4·3180 4·8155 4·2586 declines somewhat as extra-local ties are developed.
partners (1·6488) (1·6757) (1·3349) On the other hand, as VC rms mature and become
more experienced, their willingness and ability to work
Centrality measures
R&D centrality 0·0033 0·0037 0·0042 with high-risk local start-ups increases.
(0·0057) (0·0062) (0·0052) One of the particularities of venture capital is that it
Finance centrality 0·0054 0·0032 0·0098 arose and grew in diVerent places at diVerent times.
(0·0052) (0·0034) (0·0066) Consequently, there may be distinct patter ns of nan-
Licensing centrality 0·0033 0·0083 0·0036 cing based on location. To examine this, we collapse
(0·0057) (0·1158) (0·0048)
Commerce centrality 0·0027 0·0027 0·001 the regions into three areas – the Bay Area, Boston
(0·0064) (0·0054) (0·0036) and the rest of the country. Between the Bay Area and
Boston, over half of the ‘action’ occurs, so this tripartite
Number of D B Fs 57 14 62
division is sensible. Looking rst across the 12-year
period, there are some discernable patterns. With
respect to companies that only receive local support,
venture rms in the Bay Area tend to fund smaller,V Cs invest their own money, their disbursements are
very likely to be made locally. younger companies that have collaborations underway
to commercialize new products. In Boston, local onlyWe also checked to see what the relationship was
between the age of VCs and the age of biotechs at the funding goes to larger and older biotechs, which are
more involved in R & D collaborations and licensingtime of their IP Os. One speculation is that younger
VCs bring companies public earlier than older rms in agreements. Outside these two main centres, local V C
funding goes more to medium sized companies. Withorder to build a reputation and raise needed funds
(GO M P E R S , 1996). In our sample, in contrast, there regard to funding that originates outside the ‘home’
region, the biotech recipients within the Boston clusterwas a negative relation between VC age and the age
of the biotech rm at IP O. This relationship was are the younger and smaller biotechs, while in the Bay
Area cluster these rms tend to be older. In the rest ofdriven by experienced, older V Cs in the Bay Area and
San Diego that funded local younger r ms and East the country, outside support ows to older and larger
companies. Finally, the rms that receive nancingCoast VCs that manage funds with both local and
non-local younger biotechs. In sum, the gains from both locally and from the outside are older in both
Boston and the Bay Area. But, rms receiving bothexperience for older VCs include both the capacity to
oversee younger rms as well as more geographically types of nancing that are located elsewhere in the U S
are among the youngest, smallest and best connecteddistant rms. For the venture capital rms, then, there
302 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
Table 4. Means and standard deviations (in parentheses) for
biotech rms in t he Boston cluster, San Franci sco Bay Area
cluster, and outside any regional cluster that received V C
funding prior to I P O, by locality of funding
Both local
Non-local Local and non-local
Variable funding onl y funding only funding
Boston cluster
Age 8 8·83 9·08
(2·63) (0·23) (3·36)
Number of employees 160·44 360 200·92
(151·64) (226·27) (193·95)
R&D centrality 0·049 0·085 0·052
(0·071) (0·105) (0·056)
Finance centrality 0·064 0·031 0·093
(0·069) (0·020) (0·047)
Licensing centrality 0·044 0·058 0·048
(0·086) (0·081) (0·068)
Commerce centrality 0·017 0 0·014
(0·048) (0) (0·029)
Number of D BFs 9 2 12
New York (NY) Boston (B)
Rest of Country (C)
Bay Area (BA) San Diego (SD)
1988
B B
NY
C C
BA BA
SD
1990-95
B B
NY
C C
BA BA
SD
1989
B B
NY
C
NY
BA BA
SD
1996-99
C
Pre-IPO
San Francisco Bay Area
Age 7·3 7·16 7·22 Fig. 5. Regional patterns of venture capital
(2·86) (2·46) (2·31)
Number of employees 98·75 63·67 184·45
(49·09) (27·97) (186·36)
R&D centrality 0·059 0·0066 0·025
(0·081) (0·0121) (0·039)
Finance centrality 0·016 0·041 0·120
(0·0077) (0·048) (0·080)
Licensing centrality 0·013 0·0041 0·035
(0·018) (0·0071) (0·047)
Commerce centrality 0·0023 0·039 0·018
(0·0051) (0·067) (0·056)
Number of D BFs 5 3 22
Not in a cluster
Age 8·55 7·30 7·25
(3·41) (2·90) (0·96)
Number of employees 170·89 103·75 88·25
B B
NY
C
NY
BA BA
SD
1996-99
C
B B
NY
C
BA BA
SD
1992-95
C
NY B
BA SD
1988-91
C
Post-IPO
(141·42) (100·59) (60·15)
R&D centrality 0·015 0·0008 0·074 Fig. 6. Regional patterns of venture capital–biotech funding
(0·032) (0·0015) (0·073)
Finance centrality 0·049 0·0072 0·072
(0·043) (0·0071) (0·044)
Licensing centrality 0·032 0·1347 0·039 I P O, while Fig. 6 covers post-IPO. Essentially, there
(0·040) (0·1693) (0·031)
Commerce centrality 0·032 0·052 0·0010 are ve clusters – the Bay Area, San Diego, Boston,
(0·066) (0·072) (0·0020) the New York metro area, and the rest of the country.
Number of D BFs 20 5 4 Beginning in 1988 with relationships at the pre-I P O
stage, there are only two main regions for venture
capital – the Bay Area and New York City. Funds from
the Bay Area owed principally to San Diego and otherinto the world of R& D. Clearly, the threshold for
receiving both types of nancing is higher for compan- parts of the country at this stage (no doubt, due to left
censoring of the data, we miss earlier links betweenies located outside the Bay Area or Boston.
Turning from the cross-sectional portrait to a more Bay Area V Cs and biotechs), while New York money
went to Boston and the rest of the nation. In 1989,dynamic account, Figs. 5 and 6 present the sequence
of funding patterns during key periods in the industry’s Boston-based VCs enter the picture and fund local
companies, a pattern that holds for all subsequent timeevolution. These patterns were generated by examining
cross tabulations of the locations of each partner for periods. New York money continues to head north to
Boston and throughout the country, and Bay Areaall funder–fundee dyads separately for each year. We
highlight the predominant ow of VC funds in each funding picks up locally and continues in San Diego
and elsewhere. Over the years 1990–95 the only changetime period with a thick line. A dashed line indicates
a less active pattern. Fig. 5 captures relationships before is New York money heads west to San Diego. But in
The Spatial Clustering of Science and Capital 303
the most recent period, 1996–99, the picture changes rms that received local support with those that
attracted non-local nancing. The locally-funded rmsand Bay Area money moves to Boston and other parts
of the country, while New York money enters the Bay were smaller, younger, more science focused (measured
by the percentage of PhDs and MDs on their payrollArea and begins seeding rms in the New York metro
region. Over all the years, money from outside New and their number of R &D collaborations) and likely
to have more exclusive relations with only one or twoYork or the Bay Area goes to other parts of the country
and never ‘invades’ the home turf of the most active V Cs. The biotechs that garnered external support were
larger in size, older, and had advanced to a stage wherebiotech clusters.
Turning to post-I PO nancing, again New York their work had moved further down the product life
cycle (measured by their ties to other organizations toand the Bay Area are the primary locales for venture
funds for the years 1988–91. In 1992, both Boston assist in commercializing products). Thus, local VC
support is directed to much earlier stage companies,money and funds located elsewhere become active.
Once again, Boston money generally ‘stays home’. while exter nal support ows to companies that have to
‘show’ more in order to attract nancing.New York and Bay Area money moves more, especially
in this last period, 1996–99. In this later stage, Bay For venture capital rms, there is evidence that, as
the V Cs grow older an d larger, they invest more inArea money ows locally – to San Diego and to the
rest of the country and, less signi cantly, to Boston. both younger and more distant biotech companies.
These gains from experience are tempered somewhatNew York money begins to go to New York rms,
and continues to Boston and the rest of the country. by location. Boston V C money evinces a strong tend-
ency to stay home. New York money is restless, movingIn sum, V Cs located in the Bay Area hunt in their
own backyards, in San Diego and all over the nation, around to Boston, San Diego, and the rest of the
country. Bay Area VCs start out in California, wherepoaching in Boston as well. New York money moves
widely, and in later stages, as a biotech presence biotech activity is very expansive, but by the latter part
of the 1990s, California money goes to Boston anddevelops in the New York area, local rms are sup-
ported too. The rest of the nation stays out of the other parts of the country. The reciprocal move never
happens, as outside money rarely encroaches on theirestablished clusters, and Boston money remains local.
home turf in the Bay Area. We report a rather similar
pattern in an examination of the portfolio of collabora-
S U M M A RY A N D D I S C U S S I O N tions that U S biotech rms are involved in over the
period 1988–99 (OW E N -SM I T H et al., 2002). Initially,Venture capital r ms have become a key component
of the innovation process, and play an important role nearly half of all inter- rm alliances were locally based
and clustered in a few dense regions. By the end of thein high technology regions in the U S. VC-backed
R&D is three times as likely to generate patents as 1990s, most alliances were extra-local. But this process
was driven by a ‘reaching out’ from established clusterscorporate-sponsored R & D (K O R T U M and LE R N E R,
2000). In large part, this eVect is due to the direct stake to other new areas.
These patter ns suggest the diYculty of trying toentrepreneurs have in start-up rms and the fact that
entrepreneurs in large organizations receive only a small intentionally create high-tech regions. Despite abun-
dant attempts by policy makers and entrepreneurs inshare of the rewards from corporate innovation. But
venture capital support also has a catalytic eVect. Many many parts of the world, the relationships between
nance and R &D are, in many respects, based on per-companies report that V C funding is a key milestone,
and symbolically more important than other kinds of sonal ties, fostered in regions with extensive two-way
communication among the relevant parties. Such rela- nancing (HE L L M AN and PU R I, 2000). Our results
show that V C backing is a strong signal, attracting tions are not easily created by formal policies. More-
over, in the case of biotech, there has been a strong co-other V Cs from outside the local area and sustaining
a process where subsequent rounds of support are evolution of the worlds of science and nance. The
presence in the most active regions of key publicgarnered at the post-I PO stage.
We nd a strong pattern of spatial concentration in research organizations, such as research universities and
non-pro t institutes, that are buVered from marketbiotech and venture capital. Given that V Cs and
biotech are both found in considerable number in the forces means that the science plays a critical and auto-
nomous role in industry evolution. This dual contribu-Bay Area and Boston, it is not surprising that much
VC support is locally-based. A little more than half tion of money and ideas makes biotech rather diVerent
from other high-tech elds that are less steeped in basicthe biotech r ms in our sample received local VC
disbursements, and that percentage rose to 58% within research. Other elds in which product development is
more rapid and more in the hands of commercialour key geographic clusters. But the tendency of VCs
to nance local companies increased over the decade inventors may not have the same co-location patterns
of biotech.of the 1990s, indicating the continuing strong role of
VC in sponsoring R & D within a region. We see this The recurrent collaboration and mutual inter-
dependence of money and ideas raise a number ofpattern most clearly when comparing the pro les of
304 Walter W. Powell, Kenneth W. Koput, James I. Bowie and Laurel Smith-Doerr
Acknowledgements – Research support provided by
interesting questions for further research. What do the
National Science Foundation (No. 9710729, W. W. Powell
performance pro les of biotech r ms and V Cs look
and K. W. Koput, Co-PIs). We appreciate the helpful
like in the dense regions compared to areas that are less comments of Gernot Grabher and Joerg Sydow.
active? Clearly, a certain level of activity is necessary
for mobilization, but is there a point where a ‘crowding
out’ eVect sets in? Understanding the point at which
density might become a deterrent would provide lever- N O T ES
age in explaining when and where new concentrations
might emerge. In other developed countries, such as 1. For accounts of these foundings, see H A L L , 1987;
Germany and Sweden, the state has played a very active TEI T E L M A N , 1989; W E RT H , 1994; R O B B I N S - RO T H,
2000.
role in trying to stimulate venture capital disbursements.
2. See FE L D M A N , 1999, for an excellent survey of empirical
Signi cant sums of money have been made available in
studies of spillovers.
the for m of matching grants. We do not as yet know
3. To supplement infor mation about biotech companies or
whether this public policy-driven process of nancing their various partners, we consulted other courses, includ-
innovation operates in a similar manner as the private ing various editions of Genetic Engineering and Biotechnology
equity market. One might speculate that policy makers Related Forms Worldwide, Dun and Bradstreet’s Who Owns
would be less content with strong patterns of regional Whom? and Standard and Poor’s. In addition, we utilized
concentration on distributional grounds. In contrast, annual reports, Securities and Exchange Commission
however, if the criteria for evaluation are rates of lings and, when necessary, made phone calls.
founding of new organizations, then the U S ‘model’ 4. See, for example, the story in the Boston Globe about
of spatial co-location of capital and science has been ‘biotech bragging rights’, contending that by including
an expansive and robust one. In the case of biotechno- small private rms, Boston has a greater number of rms
logy, it is safe to say that without venture capital and than the Bay Area, but recognizing that the market value
regional agglomeration, the industry would not exist of the public companies in the Bay Area was nearly
double that of Boston (A O K I , 2000).in the for m that it does today.
R E F E R E N CE S
ALM E I D A P. and KO G U T B. (1997) The exploration of technological diversity and the geographic localization of innovation,
Small Bus. Econ. 9, 21–31.
AN GE L D. P. (1991) High-technology agglomeration and the labor market: the case of Silicon Valley, Environ. Plann. A 23,
1,501–16.
AOKI N. (2000) N.E. see gaining as top U S biotech area, Boston Globe, 31 October, p. E1, p. E6.
ART H U R B. (1990) Positive feedback in the economy, Scienti c American, Februar y, pp. 92–99.
AUD R E T S C H D. and FE LD M A N M. (1996) R&D spillovers and the geography of innovation and production, Am. Econ. Rev.
86(3), 630–40.
AUD R E T S C H D. B. and ST E P HA N P. E. (1996) Company-scientist locational links: the case of biotechnology, Am. Econ. Rev.
86(2), 641–52.
BAU M J. and O L I V E R C. (1991) Institutional linkages and organizational mortality, Admin. Sci. Quart. 42, 304–33.
BRO W N J. S. (2000) Foreword, in KE N N E Y M. (Ed) Understanding Silicon Valley, pp. ix–xvi. Stanford University Press,
Stanford, C A.
BYG R AV E W. D. and TI M M O N S J. A. (1992) Venture Capital at the Crossroads. Harvard Business School Press, Boston, M A.
FEL D M A N M. P. (1999) The new economics of innovation, spillovers, and agglomeration, Econ. Innovation & New Technol. 8,
5–25.
FLO R I DA R. L. and K E N N EY M. (1988) Venture capital, high technology and regional development, Reg. Studies 22, 33–48.
FR EE M A N J. H. (1999) Venture capital as an economy of time, in LE N DE R S R. and G A L B AY S. (Eds) Corporate Social Capital,
pp. 460–79. Kluwer, Norwell, MA.
GI LS O N R. J. and BL AC K B. S. (1998) Venture capital and the structure of capital markets: banks versus stock markets,
J. Financial Econ. 47, 243–77.
GO M PE R S P. A. (1995) Optimal investment, monitoring, and the staging of venture capital, J. Finance 50, 1,461–89.
GO M PE R S P. A. (1996) Grandstanding in venture capital industry, J. Financial Econ. 42, 133–56.
GO M PE R S P. and LE R N E R J. (1999) The Venture Capital Cycle. MIT Press, Cambridge, MA.
GO M PE R S P. and LE R N E R J. (2001) The venture capital revolution, J. Econ. Perspectives 15(2), 145–68.
GRA N O V ET T E R M. (1985) Economic action and social structure: a theory of embeddedness, Am. J. Sociol. 91, 481–510.
GUL AT I R. (1995) Does familiarity breed trust – the implications of repeated ties for contractual choice in alliances, Acad.
Mgt. J. 38, 85–112.
GUL AT I R. and G A R G IU L O M. (1999) Where do interorganizational networks come from?, Am. J. Sociol. 104, 1,439–93.
HA LL S. (1987) Invisible Frontiers: The Race to Synthesize a Human Gene. Atlantic Monthly Press, New York.
HE L LM A N T. and PURI M. (2000) The interaction between product market and nancing strategy: the role of venture capital,
Rev. Financial Studies 13, 959–84.
The Spatial Clustering of Science and Capital 305
JA FF E A., T R A J T E N B ER G M. and HE ND ER S O N R. (1993) Geographic localization of knowledge spillovers, as evidence by
patent citations, Quart. J. Econ. 108, 863–911.
KEN N E Y M. (1986) Biotechnology: The University–Industry Complex. Yale University Press, New Haven, C T.
KORT U M S. and L E R N ER J. (2000) Does venture capital spur innovation?, Rand J. Econ. 31, 674–92.
KRU G M A N P. (1991) Increasing returns and economic geography, J. Pol. Econ.99(3), 483–99.
LEA D B E AT ER C. (2000) The Weightless Society. Texere, New York.
MA RS HA L L A. (1920) Principles of Economics, 7th edition. Macmillan, London.
NE LS O N R. and W I N T E R S. (1982) An Evolutionary Theory of Economic Change. Harvard University Press, Cambridge, M A.
OWE N - SM I T H J., RI C C A B ON I M., P A M M O L L I F. and P O W E LL W. W. (2002) A comparison of US and Europe university-
industry relations in the life sciences, Mgt. Sci. 48(1), 21–43.
PO DO L N Y J. M. (1994) Market uncertainty and the social character of economic exchange, Admin. Sci. Quart. 39, 458–83.
PO RT E R M. (1998) Clusters and the new economics of competition, Harv. Bus. Rev. November /December, 77–90.
POW E L L W. W. (1996) Interorganizational collaboration in the biotechnology industry, J. Inst. & Theoretical Econ. 120, 197–215.
POW E L L W. W., KO P U T K. W. and SM IT H - DE R R L. (1996) Interorganizational collaboration and the locus of innovation:
networks of learning in biotechnology, Admin. Sci. Quart. 41, 116–45.
PR E VE Z E R M. (1996) The dynamics of industrial clustering in biotechnology, Small Bus. Econ. 8, 1–17.
RO BB I N S - RO T H C. (2000) From Alchemy to I P O. Persens, Cambridge, MA.
SAH L M A N W. A. (1990) The structure and governance of venture capital organizations, J. Financial Econ. 27, 473–521.
SAX E N IA N A. (1994) Regional Advantage. Harvard University Press, Cambridge, MA.
SCO T T A. J. and STO R P ER M. (1987) High technology industry and regional development: a theoretical critique and
reconstruction, Int. Soc. Sci. J. 34, 215–32.
SIN T O N P. (2000) Venture capital epicenter, San Francisco Chronicle, 3 November, p. B1, p. B4.
SO RE N S E N O. and S TUA R T T. E. (2001) Syndication networks and spatial distribution of venture capital investments, Am. J.
Sociol. 106(6), 1,546–88.
STI N C H C O M B E A. (1965) Social structure and organizations, in MA R C H J. G. (Ed) Handbook of Organizations, pp. 142–93.
Rand McNally, Chicago, IL.
STO R P E R M. and SAL A I S R. (1997) Worlds of Production. Harvard University Press, Cambridge, M A.
STUA R T T. E. (1998) Network positions and propensities to collaborate: an investigation of strategic alliance formation in a
high-technology industry, Admin. Sci. Quart. 43, 668–98.
TEI T E L M A N R. (1989) Gene Dreams: Wall Street, Academics, and the Rise of Biotechnology. Basic Books, New York.
UZZI B. (1999) Embeddedness in the making of nancial capital, Am. Sociol. Rev. 64, 481–505.
VON HIP P E L E. (1994) Sticky information and the locus of problem solving, Mgt. Sci. 40(4), 429–39.
WER T H B. (1994) The Billion-dollar Molecule. Touchstone, New York.
WRI G H T M. and RO BB IE K. (1998) Venture capital and private equity: a review and synthesis, J. Bus. Finance & Accounting
24, 521–70.
ZUC K E R L. G., DA RB Y M. R. and B R E W ER M. B. (1998) Intellectual human capital and the birth of U S biotechnology
enterprises, Am. Econ. Rev. 88(1), 290–306.
