Content uploaded by Brian Silverman
Author content
All content in this area was uploaded by Brian Silverman on Jan 13, 2014
Content may be subject to copyright.
1
Alliance-Based Competitive Dynamics
••
Brian S. Silverman
Harvard Business School
Morgan Hall 243
Soldiers Field
Boston, MA 02163
(617) 495-6729 (voice)
(617) 496-5859 (fax)
bsilverman@hbs.edu
Joel A.C. Baum
Rotman School of Management
University of Toronto
105 St. George Street
Toronto, ON M5S 3E6
(416) 978-4914 (voice)
(416) 978-4629 (fax)
baum@mgmt.utoronto.ca
April 7, 2000
•
We are grateful to Ron Burt, Tony Calabrese, Hank Chesbrough, Ken Corts, Martin Evans, Ranjay
Gulati, Harvey Kolodny, Pam Popielarz, Tim Rowley, Toby Stuart, and workshop participants at Carnegie
Mellon University, Northwestern University, University of Chicago and University of Toronto for helpful
comments. We are also grateful to Fred Haynes (Contact International) and Denys Cooper (NRC) for
permitting us access to their data. We also thank Whitney Berta, Tony Calabrese, Jack Crane, and Igor
Kotlyar for their help with data collection and coding.
2
Alliance-Based Competitive Dynamics
Abstract
Do rivals’ alliances increase or decrease the competitive pressure experienced by a firm?
Drawing on the resource-based view and on organizational ecology, we propose that the effect of
rivals’ alliances on an industry’s competitive dynamics is determined by two conditions: the degree
to which they foreclose alliance opportunities for a focal firm, and the degree to which they
increase overall carrying capacity of the industry. We distinguish between horizontal, upstream,
and downstream alliances according to these conditions. Based on these distinctions, we
hypothesize that horizontal alliances increase competitive intensity the most, upstream alliances
less so, and downstream alliances least of all. We also hypothesize that a focal firm can coopt its
rivals’ alliances by judiciously partnering with well-linked rivals. An analysis of exit in the
Canadian biotechnology industry provides evidence broadly consistent with our predictions.
Key Words: Alliances, Competitive Dynamics, Organizational Mortality, Biotechnology
3
INTRODUCTION
In technology-based industries, do rivals’ alliances increase or decrease the competitive
pressure experienced by a firm? If rivals’ alliances increase competitive pressure, what if anything
can the firm do to counteract these effects? Although the last ten years have witnessed an
explosion of research concerning the effects of alliances on the firms that participate in them, the
literature is virtually silent regarding their competitive effects on rivals. Yet these implications are
crucial for scholarly understanding of the competitive dynamics of technology-based industries as
well as for managers competing in such environments.
To the extent that the competitive dynamics of alliances have been considered, the literature
has produced varying theoretical predictions. One line of argument, with roots in the resource-
based view of the firm, is that collaborative arrangements give participants more efficient or more
timely access to scarce resources (Kogut 1988). In a world of limited potential partners, a firm’s
alliances simultaneously make it a more formidable competitor and weaken its rivals by denying
them access to desirable partners and resources (Gomes-Casseres 1994). An alternate line of
argument, stemming from ecological conceptions of carrying capacity and legitimation, contends
that links between a population’s members and established organizations outside the population
increase the availability of resources for the population overall; in a sense, a rising tide of alliances
helps all competitors, even if the allying firm benefits more (Baum and Oliver 1992). In neither
case has attention been devoted to considering what actions firms might take to defend against the
negative (or to exploit the positive) competitive dynamics that obtain.
Our paper extends prior research on the competitive dynamics of alliances. In particular, we
explore the variegated competitive implications associated with different types of alliances.
Building on evidence that incumbents’ horizontal, vertical upstream, and vertical downstream
alliances yield systematically different effects on the likelihood of entry into new technology-based
industry subfields (Calabrese, Baum and Silverman 2000), we predict that these different
categories of alliances will generate systematically different effects on competitive intensity among
rivals. We link these disparate effects to the degree to which an alliance 1) forecloses rivals’
4
alliance opportunities and 2) expands the resource base available to industry participants. We also
explore the degree to which a firm can mitigate the competitive effects of rivals’ alliance networks
We test our hypotheses in an empirical study of all firms participating in the Canadian
biotechnology industry between January 1991 and December 1996. For this sample of firms, we
investigate the impact on a focal firm’s exit rate of its rivals’ alliances. We find evidence consistent
with several of our predictions. Notably, we find that the competitive intensity experienced by a
focal firm generally increases with the number of alliances that its rivals form, but that this effect is
moderated for those types of alliances that are more likely to expand the industry resource base
and are less likely to foreclose rivals’ alliance opportunities. We also find that a focal firm benefits
from the alliances of those biotechnology firms with which it collaborates. Our study thus 1)
highlights the multifaceted competitive impact of rivals’ alliances on a focal firm, 2) underscores
the importance of judicious partner selection in forming horizontal alliances to mitigate
deleterious competitive effects, and 3) identifies criteria that can inform such partner selection.
THEORY AND HYPOTHESES
Competitive Effects of Alliances
Interorganizational relationships appear to provide a myriad of advantages for participating
firms, primarily associated with the direct or indirect availability of resources. Resource-based
theorists propose that alliances facilitate the accessing of knowledge and other assets for which
arm’s-length contracting between parties may be hazardous (Kogut 1988). Alternatively, alliances
may confer upon a firm an aura of external legitimacy (Miner, Amburgey and Stearns 1990),
which in turn enables the firm to acquire other necessary resources. These advantages are
hypothesized to be particularly important when timely access to dispersed knowledge and
resources is crucial (Teece 1992), or when ambiguous technologies force actors to rely on more
indirect social indicators to assess organizational performance (Stuart, Hoang and Hybels 1999).
5
A common thread throughout this approach is the notion that alliances enable a firm to out-
compete its rivals in the quest for resources necessary for strong performance or survival. An
implicit, but not theoretically explored, corollary is that a firm’s alliances not only make it better
able to withstand competition, but also make it a stronger competitor for others. Put another
way, if “winning the alliance race” at one point in time increases a firm’s chance to win subsequent
races for partners, capital, or patents (Amburgey, Dacin and Singh 1996; Walker, Kogut and Shan
1997), then a firm that gains alliances today is better positioned to deny its rivals access to
ancillary resources tomorrow. Such an outcome is exacerbated in a world of limited partners.
Partner scarcity rewards a firm that moves quickly to secure high-quality partners by both
enhancing its own reputation as a desirable partner (Gulati 1995; Powell, Koput and Smith-Doerr
1996) and foreclosing rivals’ partnering opportunities (Gomes-Casseres 1994). Thus, much as
increased size may make a firm both more viable and a more intense competitor (Barnett and
Amburgey 1990), alliances not only make a focal firm organizationally viable, but also
ecologically potent.
An alternate argument in the literature proposes similarly beneficial results for participants in
alliances, but is more optimistic about the effect of such linkages on rivals. In this view, alliances
with established actors may benefit a focal firm’s rivals by increasing the carrying capacity of an
industry, either through increases in capital available to industry participants or through increased
“legitimation” of the industry (Baum and Oliver 1992). Alliances that involve established actors
thus embody a public good element, particularly for young industries that are characterized by
uncertainty about the fundamental viability of their technologies and organizational forms (Aldrich
and Fiol 1994). Thus, such alliances may simultaneously make a focal firm organizationally
viable and reduce the competitive pressure of the environment.
When are a firm’s alliances likely to increase competitive intensity, and when are they likely to
decrease it? We propose that alliances with different types of partners are characterized by
different levels of rival foreclosure and carrying capacity enhancement, and will consequently yield
systematically different effects on the level of competitive intensity introduced. Calabrese et al.
6
(2000) note that the prevalence of incumbents’ upstream alliances, downstream alliances, and
horizontal alliances yield systematically different effects on the likelihood of entry into subfields of
the Canadian biotechnology industry. Below we distinguish among horizontal, upstream, and
downstream alliances according to the degree to which these alliances forged by a focal firm
foreclose alliance opportunities to its rivals, and the degree to which these alliances increase the
carrying capacity of the industry. Based on this distinction, we then explore the effects of each
type of alliance on the competitive dynamics of technology-based industries. These arguments are
displayed graphically in Figure 1.
[PLACE FIGURE 1 ABOUT HERE]
Downstream alliances. Downstream alliances link firms in a technology-based industry to
sources of complementary assets, commercialization knowledge, and capital outside of the
existing industry boundaries. For example, biotechnology firms’ downstream alliances with
pharmaceutical, chemical, or marketing companies provide biotechnology firms with distribution
channels, production facilities, marketing expertise, and financing that facilitate the successful
development and commercialization of a product or process (Kogut, Shan and Walker 1992).
Such alliances provide a firm with increased access to resources, which are likely to increase the
firm’s viability and make it a stronger competitor.
At the same time, such alliances directly and indirectly increase the level of resources
available for industry participants. Alliances with downstream companies often involve significant
infusions of capital into an industry (Lerner 1994). In addition, alliances between young
technology-driven firms and established downstream companies appear to increase the interest
demonstrated by other sources of financing (Stuart et al. 1999), presumably because downstream
company involvement signals (or is perceived to signal) the commercial viability of the firms’
technologies. Similarly, to the extent that a downstream alliance also signals to other downstream
firms the commercial potential of a particular area of innovation, one downstream alliance will
7
lead other downstream firms to follow suit, thus increasing the resources available to an industry
(Amburgey et al.1996).
Finally, a firm’s downstream alliances often do not pose a high foreclosure risk to its rivals.
For example, large pharmaceutical firms typically maintain dozens or hundreds of simultaneous
alliances with different biotech firms. Further, it is frequently the case that downstream activities,
such as marketing or distribution, are the most scale- and scope-intensive of a technology-based
industry value chain (Calabrese et al. 2000). The scale and scope economies associated with these
activities imply that it is economically feasible – and even desirable – for downstream firms to
partner with multiple players in an industry. Thus, a focal firm’s alliances with downstream
partners are not likely to significantly foreclose rivals’ access to the same partners. In sum,
alliances with downstream partners typically do not foreclose rivals from forming similar alliances,
and simultaneously expand the resource base available to an industry.
Upstream alliances. Upstream alliances link firms in a technology-based industry to sources
of research knowledge. Biotechnology firms’ alliances with universities, research institutes,
hospitals, government labs, and industry associations purportedly provide them with cutting-edge
scientific and technological expertise necessary for the successful discovery and patenting of new
products or processes (Argyres and Liebeskind 1998). Semiconductor firms’ alliances with
universities are credited with providing similar access to important technological knowledge
(Spencer and Grindley 1993).
Such alliances increase the accessibility of scientific inputs for firms in an industry. However,
to the extent that the resulting knowledge is not subject to spillovers, it is not likely to benefit
rivals of the allying firm. More important, alliances with upstream partners foreclose rivals’ access
to those partners to a greater degree than do downstream alliances. Zucker, Darby and Brewer
(1994) indicate that scientists, including “star” scientists whose research efforts have a tremendous
impact on biotechnology firm success, rarely transact with more than one firm at a time. The
relative lack of scale and scope economies in individual research project (as compared to, say,
marketing) imposes stark limits on the number of simultaneous alliances to which an upstream
8
research player can commit its particular scientific or technological expertise. Finally, in contrast
to downstream alliances, upstream linkages rarely increase directly the level of financing or other
resources available to industry participants. In sum, as compared to downstream alliances,
alliances with upstream partners are likely to foreclose rivals from forming similar alliances while
expanding modestly the resource base available to an industry.
Horizontal alliances. Horizontal alliances link firms to other firms in the same industry. In
contrast to vertical alliances, such alliances between potential competitors do not tap resources
outside of the existing industry. In addition, as the number of horizontal alliances increases, rivals
face an increasingly limited pool of potential (and desirable) partners. Further, since each
horizontal alliance involves two firms of the same partner type, each alliance consumes two links
out of the limited supply available. Thus, horizontal alliances are likely to foreclose rivals from
forming similar alliances, while having no beneficial effect on the resource base available to an
industry.
Effect of Rivals’ Upstream, Downstream and Horizontal Alliances on Competitive Pressure
The above discussion distinguishes among rivals’ upstream, downstream and horizontal
alliances in terms of their effect on the competitive pressure experienced by a focal firm. Each
type of alliance is expected to make the participating rival more potent ecologically, thus raising a
focal firm’s exit rate. However, vertical alliances are more likely than horizontal alliances to yield
offsetting benefits in terms of expansion of the industry resource base. Further, downstream
alliances are less likely than upstream alliances to foreclose on the focal firm’s opportunity to forge
alliances of its own, thus mitigating some of the deleterious effects of rivals’ alliances. Thus, we
predict:
H1a: A firm’s exit rate increases as its rivals’ number of alliances increases.
9
H1b: The increase in a firm’s exit rate as its rivals’ number of alliances increases is lower
for upstream alliances than for horizontal alliances.
H1c: The increase in a firm’s exit rate as its rivals’ number of alliances increases is lower
for downstream alliances than for either upstream alliances or horizontal alliances.
Alliance Network Efficiency
In addition to a firm’s individual alliances, the overall composition of its alliance network is
likely to affect the competitive intensity it generates. To the extent that a firm’s alliance network is
redundant – that is, to the extent that it includes multiple alliances that provide access to the same
information (Burt 1992) or capabilities (Gomes-Casseres 1994) – a firm will access less diverse
information and capabilities for greater cost than it would through a smaller nonredundant set. A
highly redundant configuration may even prevent a firm from obtaining new or novel information
critical to its adaptation by limiting the number of links to firms in touch with emerging
innovations (Uzzi 1996, 1997). More ‘efficient’ alliance configurations, which provide access to
more diverse information and capabilities per alliance, and thus produce desired benefits with
minimum costs of redundancy, provide more benefit to firms than do inefficient configurations.
Consistent with this expectation, Powell et al. (1996) showed that U.S. biotechnology firms in
human therapeutics with ties to a more diverse set of activities were better able to locate
themselves in resource and information-rich positions, and grew more rapidly. Relatedly, Baum,
Calabrese and Silverman (2000) found that more efficient alliance configurations improved
performance of Canadian biotechnology firms along several dimensions – even after controlling for
the main effects of alliance formation.
To the extent that more efficient alliance configurations create stronger competitors, we
expect that, after controlling for the main effect of the number of alliances formed, increased
efficiency of rivals’ alliance networks will increase the competitive intensity experienced by a focal
firm.
10
H2: A firm’s exit rate increases as its rivals’ alliance network efficiency increases.
Collaboration as a Response: Beneficial Effects of Rival Partners’ Alliances
The above hypotheses stress how a rival’s alliances may increase the competitive threat it
poses to a focal firm. At the same time, these alliances may also make the rival a better partner.
Powell et al. (1996) note how biotechnology firms that are more centrally positioned – crudely
speaking, those with more links to other well-linked players – have greater and faster access to
more fine-grained information and are better able to absorb technological knowledge and other
resources. More generally, centrality in the interfirm network generates access to resources. As a
result, well-connected rivals may represent 'high quality' partners because they possess leading-
edge technology, have rapid access to critical information, and have accumulated partnering
experience. To the extent that access to knowledge is a primary motive for interfirm alliances, as
widely contended in the alliance literature, then the benefit of a firm’s alliances will depend in part
on its partners' alliances.
As a result, a firm may be able to turn competitors’ alliance-based competitive strengths to its
own advantage by establishing alliances with its better-connected rivals. This is particularly true
for a rival whose access to interfirm resources and information is enhanced by an efficiently
configured alliance network. Alliances with well-connected partners provide promising
opportunities to learn new capabilities and acquire advanced technological knowhow. Such
advantages also arise to the extent that there exist signaling benefits to allying with well-linked,
and consequently highly visible, partners (Podolny 1994; Stuart et al. 1999). Thus we predict:
H3: A firm’s exit rate decreases as its partners’ number of alliances increases.
H4: A firm’s exit rate decreases as its partners’ alliance network efficiency increases.
11
METHODS
We test the above hypotheses in a study of competitors in the Canadian biotechnology
industry. In biotechnology, the significant resource and speed demands of patent races and
commercialization motivate biotechnology firms (BFs) to seek out partnerships with downstream,
upstream, and horizontal firms (Powell et al. 1996). Alliances with downstream partners provide
access to complementary assets critical to successful development and commercialization: market
access, marketing and distribution infrastructure, technology and production facilities, and/or
expertise in managing clinical trials (Pisano 1990). This is particularly true of collaborations with
pharmaceutical and chemical firms. Alliances with upstream partners are a source of up-to-date
information or knowledge critical to success in patent races but too tacit to be effectively
transferred through licensing or purchase (Liebeskind, Oliver, Zucker and Brewer 1996). This is
particularly true of collaboration with universities and research institutes. Alliances with other
BFs offer many of the above benefits as well as access to experience on how to operate and grow
a firm in the biotechnology industry.
Many scholars have noted the unique role of patents in biotechnology (Austin 1993; Fligstein
1996). Given the unusually strong appropriability regime associated with biotechnology, patents
provide BFs with significant bargaining power in negotiations for financing and other assets
required for product commercialization (Pisano 1990). In addition, frequent publicity surrounding
a BF’s pending patents (see, for example, the Canadian Biotechnology Handbook) indicates that
they are used to signal patent race leads that are exploited in the race for additional resources.
Consequently in our analysis we are careful to control for the patenting activity of a focal firm and
its rivals. Although treated as control variables, the competitive dynamics of patenting are
interesting in their own right, and we consider briefly these results in the discussion section.
Data
We tested our hypotheses using data describing the alliances and other organizational
characteristics of BFs operating in Canada during the six-year period between January 1, 1991
12
and December 31, 1996. We compiled event histories for the 613 BFs that existed at any time
during this period using data from Canadian Biotechnology, an annual directory of companies
active in the biotechnology field operating in Canada. Published since 1991, Canadian
Biotechnology is the most comprehensive historical listing in existence of Canadian BFs, their
products, growth, performance, and alliances. Canadian Biotechnology tracks BFs operating in
sixteen industry sectors: 1) agriculture, 2) aquaculture, 3) engineering, 4) environmental, 5) food,
beverage and fermentation, 6) forestry, 7) human diagnostics, 8) human therapeutics, 9) human
vaccines, 10) horticulture, 11) contract research organization, 12) veterinary, 13) energy, 14) bio-
materials, 15) cosmetics and 16) mining.
We cross-checked this information with The Canadian
Biotechnology Handbook (1993, 1995, 1996), which lists information for a more restrictive set of
core Canadian BFs – firms entirely dedicated to biotechnology – and found no significant
discrepancies in information for those firms represented in both sources.
The Canadian biotechnology industry is made up of relatively small firms when compared to
those in the U.S. and Japan. Most biotechnology firms are located in Ontario, followed by
Quebec and British Columbia. In 1991, 471 BFs were already in operation; thus, the life histories
for these firms were left-censored (i.e., founded before the study period). With available archival
information, we were able to confirm founding dates for all left-censored BFs. Although left
censoring presents a difficult estimation problem when founding dates of censored organizations
are not known, the estimation problem can be corrected using a conditional likelihood approach
when founding dates are known (Guo 1993). During the study period, 142 BFs were founded and
186 exited. Exit was defined as either the failure of a firm or closure of a subsidiary or joint
venture. Acquisitions or changes in name were not included as failures because the organization
itself continued to operate. This treatment corresponds to common practice in organizational mortality studies
(Baum 1996). Therefore, the final sample for the analysis included 613 BFs, of which 186 (30.6%) exited the
industry between January 1991 and December 1996.
In addition to the directory data, we used the Micropatent database to identify each patent
issued to these BFs in the United States that was applied for between January 1975 and December
13
31, 1996.
1
We used U.S. patent data because most Canadian BFs file patent applications in the
U.S. first to obtain a one-year protection during which they file in Canada, Europe, Japan, and
elsewhere (Canadian Biotech ‘89; Canadian Biotech ‘92). When these BFs were subsidiaries, we
included only those patents assigned to the subsidiary. During the 1975-90 period, these BFs
were granted 1,309 patents, and during 1991-1996, were granted 502 patents.
Two aspects of our industry definition merit discussion. One is our use of a national boundary
to define the industry boundary, and the second is our inclusion of firms in both human and non-
human biotechnology sectors as part of the same industry. Several factors point to the value of
studying the Canadian biotechnology industry as a ‘quasi-independent’ organizational population –
that is as a population that has its own internal dynamic, but is also shaped by, and active in,
industry activity beyond the national border. The first is the industry's significance, both in size,
and to the Canadian economy. During the 1991-97 period, Canada's national biotechnology
industry was second only to the U.S. in number of firms. The yearly average number of U.S. BFs
during this period was approximately 1,300 (Amburgey et al. 1996; Ernst and Young 1997). The
Canadian average was about 460, or roughly one-third the number in the United States. By way
of comparison, Canada's economy is approximately one-tenth the size of the U.S. economy.
Moreover, between 1989 and 1993, biotechnology led all Canadian industries in growth in
domestic sales (24%), exports (19%) and employment (14%) (Canadian Biotech ‘94).
In addition to its substantial size and economic significance, two further factors suggest the
Canadian biotechnology industry should be considered a quasi-independent population (Calabrese
et al. 2000). First, Canadian BFs draw almost exclusively on within-country sources of financing
(i.e., Canadian capital markets, banks, and venture capital firms). U.S. venture capital firms tend
to fund startups that operate within the U.S., and typically focus on startups located nearby
1
Pending U.S. patent applications are not public information. Rather, the U.S. Patent and Trademark Office
publishes patents only upon granting them. As a result, and consistent with prior research using patent statistics,
we only include information on ultimately successful patent applications. In addition, there is a time lag between a
patent application’s filing and its granting. During the 1984-1993 period, 93% of the patents in our sample were
granted within 3 years of application. This is consistent with broader patent patterns in the U.S. (Griliches 1990).
We used Micropatent through 1999 to search for patents granted to our sample firms that were applied for before
1996, thus identifying virtually all patents pending during our sample period.
14
(Gompers and Lerner 1998; Sorenson and Stuart 2000). Second, there is a very high degree of
within-Canada partnering among Canadian BFs. Within-country partnerships accounted for more
than 90% of the horizontal alliances established by Canadian BFs in our data. Barley et al. (1992:
325) report comparably few alliances between U.S. and Canadian biotechnology firms – just 33 at a
time when there were over 400 Canadian BFs. Our data indicate a similarly strong tendency
toward within-country partnerships across many types of alliances, particularly upstream. So, the
network of organizations active in the Canadian biotechnology industry is highly localized within
Canada's national boundaries.
Prior research on the U.S. biotechnology industry has employed two industry definition
strategies. One approach, employed by Stuart et al. (1999) and Powell et al. (1996) for example,
is to focus on the human sector in isolation. Stuart et al. (1999) study human diagnostics and
therapeutics, while Powell and his colleagues study only human therapeutics. In these studies, the
implicit assumption seems to be that there is something distinctive about the markets and/or the
regulatory environment surrounding the human applications sectors of the biotechnology industry
that justifies their isolated study. On the other hand, Barley et al. (1992), Amburgey et al. (1996)
and Walker et al. (1997) have adopted a similar industry definition strategy as our own that
emphasizes 'core technology/knowledge' as the key defining characteristic of an industry's
boundary. As Walker et al. (1997: 123) characterize it for example, “biotechnology includes all
techniques for manipulating micro-organisms.” Although we adopt this second approach, we also
recognize that human sectors of the biotechnology industry do have important features that
differentiate them from non-human sectors (e.g., a much more lengthy and stringent product
approval process). As described below, our empirical strategy is sensitive to sectoral differences,
defining all theoretical variables at the sector level, and incorporating time-varying, sector-specific
controls (e.g., financing, and competition), as well as sector-specific fixed effects in our baseline
model.
Independent Variables
15
To test our hypotheses we constructed firm-specific variables to measure each BF’s number of
alliances by type of partner and its alliance network efficiency. For each BF, we then aggregated
these variables for all of its rivals and for all of its partners. First we describe the variable
construction, and then describe the aggregation procedure:
Rivals’ alliances. To test H1a-c we constructed a set of firm-specific variables that counted,
separately, the aggregate number of alliances a BF’s rivals had at the start of each year with the
following types of organizations: 1) downstream partners: pharmaceutical firms, chemical firms,
marketing firms; 2) upstream partners: universities, research institutes, government labs, hospitals,
industry associations; and 3) horizontal partners: rival BFs.
Although the biotechnology firms in
our sample are all Canadian, our alliance data includes information on these firms’ alliances with
other organizations worldwide. For each focal firm, we defined its rivals as those BFs that 1)
operate in one or more of the same biotechnology sectors (e.g., agriculture; human diagnostics) as
the focal BF and 2) are not partners of the focal BF.
2
The resulting variables are analogous to
elaborations of the standard density dependence model (Hannan & Carroll 1992) such as
population mass (Barnett & Amburgey 1990), which respecifies density as the aggregate of the
sizes of competing organizations to estimate the effects of size asymmetries on competitive
dynamics. H1a-c predict that a firm’s exit rate should increase significantly as its rivals establish
more horizontal alliances, to a lesser degree as its rivals establish more upstream alliances, and
still less so as its rivals establish more downstream alliances. Consequently, we expect the
coefficients for this set of variables to adhere to the following pattern: horizontal > upstream >
downstream.
Rivals’ alliance network efficiency. Following Burt’s (1992) conception of structural
equivalence – in which firms participating in the same line of business are considered equivalent in
the set of skills, relationships, and assets they embody – we assume that alliance partners of a given
type are roughly structurally equivalent. We therefore construct a measure of alliance network
2
We found only two alliances between BFs that operate in non-overlapping sectors; consequently, we did not
estimate non-rival BF alliance effects.
16
efficiency that captures the diversity of each rival’s alliance partner types. This measure is based
on the Hirschman-Herfindahl index, and computes diversity as one minus the sum of the squared
proportions of a BF’s alliances with each of the nine partner types divided by the
of alliances.
3
We then aggregated the alliance network efficiency scores for all of a focal BF’s
rivals. Specifically:
Rivals’ Alliance Network Efficiency = ∑∑
i
( [1 – ∑∑
ij
(PA
ij
)
2
] / NA
i
)
where PA
ij
is the proportion of rival i's alliances that are with partner type j, and NA
i
is rival i's
total number of alliances. A rival with 6 alliances, 2 with pharmaceutical firms, 2 with hospitals,
and 2 with marketing firms would score [1-(2/6)
2
+(2/6)
2
+(2/6)
2
]/6 = 0.111. Another rival with a
less "efficient" network, comprised of 5 alliances with pharmaceutical firms and 1 marketing
alliance, would score [1-(5/6)
2
+(1/6)
2
]/6 = 0.046. H2 predicts a positive relationship between
rivals’ network efficiency and a focal firm’s exit rate. Therefore we expect the coefficient for this
variable to be positive.
Partners’ Alliances. To test H3, we constructed a set of firm-specific variables in the same
manner as the Rivals’ Alliances measures. This set of variables counted the aggregate number of
alliances a BF’s partners had at the start of each year with each of the 9 types of organizations. H3
predicts that a firm’s exit rate should decrease as its partners establish more alliances, be they
downstream, upstream, or horizontal. Consequently, we expect the coefficients for this set of
variables to be negative.
Partners’ Alliance Network Efficiency. To test H4, we constructed a measure of
aggregate partner alliance network efficiency in the same manner as the Rivals’ Network Efficiency
measure. Specifically:
Partners’ Alliance Network Efficiency = ∑∑
k
( [1 – ∑∑
kj
(PA
kj
)
2
] / NA
k
)
3
Scaling by total number of alliances permits the variable to capture the extent to which the alliances comprising
each BF’s network are, on average, redundant.
17
where PA
kj
is the proportion of partner k's alliances that are with partner type j, and NA
k
is partner
k's total number of alliances. H4 predicts that a firm’s exit rate should decrease with the efficiency
of its partners’ alliance networks. Consequently, we expect the coefficient for this variable to be
negative.
Control Variables
Many other factors may influence the fates of BFs. Accordingly, the analysis controls for a
variety of additional BF characteristics and industry- or sector-specific environmental features.
Firm characteristics. Prior research has emphasized the importance of a firm’s own alliances
to its performance (e.g., Baum et al. 2000). We therefore constructed for each focal firm a set of
own-alliance variables that count the aggregate number of alliances that a BF had at the start of
each year with each type of partner. To control for own alliance network efficiency, we
constructed Own Efficiency in the same manner as the Rivals’ Network Efficiency measure. To
control for a focal firm’s innovative capabilities or success, we included annually updated counts of
Older Patents (defined as patents granted more than 5 years ago), Recent Patents (granted with 5
years), and Pending Patents (applied for but not yet granted).
Prior research has frequently demonstrated that age and size affect organizational exit rates.
We therefore included Age, defined as the number of years since the date of a BF’s founding. Size
was controlled with three annually updated variables: 1) the number of dedicated (non-R&D)
Biotechnology Employees, 2) the number of dedicated R&D Biotechnology Employees, and 3) a
Manufacturing Facility dummy variable coded 1 if the BF owned a manufacturing facility and 0
otherwise. It is also possible that a BF’s propensity to exit varies with the characteristics of the
sectors in which it operates. We controlled this with a set of fixed effects for Primary
Biotechnology Sector. Each variable was coded 1 if the BF was most active in 1) agriculture, 2)
aquaculture, 3) engineering, 4) environmental, 5) food, beverage and fermentation, 6) forestry, 7)
human diagnostics, 8) human therapeutics, 9) human vaccines, 10) horticulture, 11) contract
18
research organization, 12) veterinary, 13) energy, and 14) biomaterials, cosmetics and mining
(combined since each represented <1% of the sample) (Canadian Biotechnology, 1996).
Revenue was measured as the natural logarithm of annual revenues in 1991 constant dollars.
R&D expenditure was measured as the natural logarithm of annual R&D budget in 1991 constant
dollars. Diversification was defined as the total number of biotechnology sectors in which a BF
was active in each year. Ownership effects were controlled with a set of time-varying dummy
variables. Each variable was coded 1 if, during a year, the BF was 1) Private, 2) Public before
1991, 3) IPO during 1991-1996, 4) Nonprofit, 5) Government/University/ Hospital, 6)
Subsidiary, and 7) Joint Venture, and 0 otherwise. Finally, to control for survivor bias, we
included Left-censored, coded 1 for BFs founded before 1991 and 0 otherwise.
Rivals’ and partners’ characteristics. Given the competitive importance of patents in
biotechnology, we included aggregate measures of rivals’ and partners’ patents. We constructed
annually updated aggregate counts of Rivals’ Older, Recent, and Pending Patents. Similarly, we
constructed annually updated aggregate counts of Partners’ Older, Recent, and Pending
Patents. In addition to their innovativeness (i.e., patenting success), the effect of alliances with
rival BFs may depend on the relative scope of the partners (Khanna, Gulati and Nohria, 1998).
Baum et al. (2000) find that the broader the scope of a startup BF vis-à-vis the scope of its
potential rival partner, the better the BF’s resulting performance. To control for such effects, we
computed a measure of the relationship between the scope of a focal firm and the average scope
of those BFs with which it is allied at the start of each year. This measure computes relative
scope as the number of biotechnology sectors in which a focal BF participates divided by the
average number of biotechnology sectors in which its BF partners participate. Specifically:
Relative Scope
f
= S
f
/ (∑∑ S
k
/ NABF
f
)
where k includes all of focal BF f’s partner BFs, S
f
and S
k
are the number of biotechnology sectors
in which BFs f and k are active, and NABF
f
is BF f's total number of alliances with rival BFs.
19
Environmental Factors. We also controlled for several factors that may potentially
influence the carrying capacity and competitiveness of the biotechnology industry in Canada.
First, we obtained from the National Research Council of Canada yearly information on aggregate
financing of BFs, by biotechnology sector, from all sources (e.g., venture capital, private
placement, IPO, public offering, and other). To control for the effect of available capital on
carrying capacity, and hence on BF exit, we included Aggregate Bio Financing, defined as the
total financing (in 1991 constant dollars) in all sectors in which a given BF was active in each
year. Our intuition is that BFs that convert patent applications into market capitalization benefit
most from this financing. We therefore control for this, albeit indirectly, by including the
interaction term Aggregate Bio Financing * Own Patent Applications in the model. A negative
coefficient for this interaction would indicate that firms possessing a greater number of patent
applications benefited more from increased financing in their sector(s).
Of course, the effect of adding a dollar of Bio financing depends on the intensity of
competition at the time the funding is added. If potential competition is not measured explicitly,
then the effect of adding resources to the environment is assumed to be constant over time, and
estimates for the effects of Bio financing will suffer specification bias. Therefore, we incorporate
time-varying information on potential competition faced by BFs. We defined a BF's potential
competition as both the number or “density” (Hannan & Carroll 1992) and aggregate size
(measured as the sum of R&D and non-R&D employees) or “mass” (Barnett & Amburgey 1990) of
other BFs operating in one or more similar biotechnology sectors as the focal BF at the start of
each year. These competition measures are called Number of Potential Rival BFs and Mass of
Potential Rival BFs, respectively.
Potential competition may also come from across the border, as U.S. biotechnology firms
compete with Canadian firms in U.S., Canadian, and international markets. Indeed, Canadian BFs
typically patent their innovations first in the U.S. Therefore, we developed two sector-specific
control variables. The first controls for sector-specific density-dependent competition from U.S.
BFs, and the second for sector-specific recent patenting by U.S. BFs. We computed these
20
variables for the start of each observation year based on annual counts of the number of U.S. BFs
and recent patents granted to U.S. BFs (i.e., granted in the last five years).
4
The U.S. BF data are
disaggregated into six subcategories: 1) agriculture, 2) food, 3) diagnostics, 4) therapeutics, 5)
veterinary, and 6) other. To match the U.S. and Canadian data, we assigned U.S. BFs and their
recent patents to Canadian sectors as follows:
• U.S. agriculture to Canadian agriculture, aquaculture, and horticulture
• U.S. food to Canadian food, beverage, and fermentation
• U.S. human diagnostics to Canadian human diagnostics
• U.S. human therapeutics to Canadian human therapeutics and human vaccines
• U.S. veterinary to Canadian veterinary
• U.S. other to Canadian forestry, engineering, environmental, energy, contract research,
biomaterials, cosmetics, and mining.
Because U.S. BFs and patents in each subcategory were assigned to multiple Canadian
sectors, we needed a method for allocating the U.S. counts to the Canadian sectors to construct
the sector-specific U.S. BF density and U.S. recent patent controls. We accomplished this two
ways. First, we divided the U.S. firms and patents equally across the Canadian sectors. In the
second, in each year, we weighted the allocations according to the proportion of Canadian BFs or
patents in the sectors to which the allocations were made. The two allocation methods yielded
similar estimates, so we report those based on the simpler equal allocation below.
Descriptive statistics for the alliance and patent variables are given in Appendix Table A2. The correlations
are generally small to moderate in magnitude (95% are less than .5 or 25% shared variance), although correlations
are generally stronger among the rival BFs' alliance and patent variables (the maximum, between rival BFs'
hospital alliances and rival BFs' university alliances is .78 or 61% shared variance). Such levels of
multicollinearity among explanatory variables can result in less precise parameter estimates (i.e. larger standard
errors) for correlated variables but will not bias parameter estimates (Kennedy 1992). So, although this does not
4
We grateful to Terry Amburgey for providing us these data on U.S. BFs. See Amburgey et al (1996) for a more
detailed description.
21
pose a serious estimation problem, it can make it difficult to draw inferences about the effects of adding particular
variables to the models. Therefore, when estimating results, we followed a strategy of estimating hierarchically-
nested models and testing for the overall significance of sets of added variables, and we examined standard errors
for evidence of inflation to check that multicollinearity was not causing less precise parameter estimates (Kmenta
1971:371).
Model Specification
We estimate BFs' industry exit rate using λ
j
, the discrete time hazard rate, as our dependent
variable. The discrete time hazard rate – the conditional probability of experiencing an event at
time t
j
– of a BF exiting the industry is defined as:
λλ
j
= Pr ( T = t
j
| T $$ t
j
)
The left-censored cases founded before 1991 in our data create an estimation problem. The
problem is that firms founded before 1991 tend to over-represent 'low risk' cases because they
survived long enough to appear in our sample. This 'sample selection' bias is likely to result in an
underestimation of exit rates at younger ages in a standard event history analysis. Correcting for
left censoring presents a difficult estimation problem when founding dates of censored firms are
not known. However, when founding dates are known, the estimation problem can be corrected
using a conditional likelihood approach (Guo 1993). The conditional likelihood approach
addresses the problem of sample selection by conditioning the estimates for left-censored firms on
their having survived to t
0
, the amount of time the firm survived in the pre-observation period
(Guo 1993). We estimate a conditional discrete time model using Logistic regression following
the procedure outlined by Guo (1993: 239).
To estimate the conditional discrete time model, we use the year as a discrete unit of time,
viewing each time unit as an independent trial. A BF that exits in its kth year is treated as having
experienced k trials. Corresponding to these trials are k success indicators, which are coded 1 if
22
an exit occurs during a trial and 0 otherwise. Accompanying each trial is a vector of time-varying,
annually updated covariates containing independent and control variables. For left-censored BFs
first observed at t
0
at the end of their kth year, to obtain conditional discrete time estimates, the
first k trials must be dropped from the analysis (Guo, 1993). We used TDA 5.7 (Rohwer 1995) to
estimate the vectors of parameter estimates using logistic regression procedures.
RESULTS AND DISCUSSION
Table 1 reports conditional discrete time estimates for our analysis of BF industry exit rates.
Model 1 provides a baseline model that includes basic environmental and firm-level controls. For
ease of presentation, coefficients for firm and sector controls are reported in Appendix A1.
Model 2 adds controls for a focal firm’s own alliances and patents. As indicated by the likelihood
ratio test, this model offers a significant improvement over Model 1. Model 3 adds rival BFs'
alliances and alliance configuration efficiency to test H1a-c and H2, and also controls for rivals’
patenting activity. Model 3 is again a significant improvement. Finally, to test H3 and H4, we
include partners’ alliances in Model 4. This model, which also controls for partners’ patenting
activity and the relative scope of a focal firm vis-à-vis its partners, yields a significant
improvement in fit and thus represents our best-fitting model. Coefficients are generally robust
across models, so, we interpret model 4.
[INSERT TABLE 1 ABOUT HERE]
Effects of rival BFs’ alliances (H1a-c and H2). Rivals’ alliances with six of the nine partner
types are positively associated with a focal firm’s likelihood of exit. Of greater interest, these
results vary systematically across alliance category. Rivals’ alliances with other BFs
unambiguously raise the focal firm’s exit rate. Rivals’ alliances with upstream partners – universities,
research institutes, government labs, hospitals, and industry associations – generally raise the focal
firm’s exit rate, but results are mixed. University alliances are significantly negatively associated
23
with exit, and the coefficients for hospital and industry association linkages are significant at only
p < .10. Finally, rivals’ alliances with downstream partners – pharmaceutical companies, chemical
companies, and marketing firms – generally are unrelated to a focal firm’s likelihood of exit. The
coefficients for pharmaceutical and chemical firm alliances are negative and not significant.
Although rivals’ marketing firm alliances are positively associated with a focal firm’s exit rate, the
coefficient is an order of magnitude smaller than that for rivals’ horizontal alliances.
To further explore these effects, we consider the effect size multipliers at the mean number of
alliances formed. Multipliers are computed as e
β∗µ
, where µ is the variable’s mean value. At the
mean, the multiplier for alliances with other BFs is nearly three times larger than that for any other
type of alliance. The multipliers for alliances with upstream partners are smaller than that for
horizontal alliances, and with the exception of university alliances are all larger than the multipliers
for downstream alliances. Thus, the effects of rivals’ alliances on a focal firm’s exit rate generally
conform to the predictions in H1a-c: horizontal alliances increase competitive intensity the most,
followed by upstream alliances, with downstream alliances generating the least competitive
intensity.
Consistent with H2, rivals’ alliance configuration efficiency is positively associated with a
firm’s likelihood of exit. Thus, controlling for the number of each type of alliance that a BF's rivals
form, rivals with more efficient alliance configurations are more intense competitors than those
with less efficient configurations.
Effects of partner BFs’ alliances (H3 and H4). Consistent with H3, partner BFs’ alliances
generally decrease a focal firm’s exit rate. Partners’ alliances with 5 of the 9 alliance types are
negatively associated with a focal firm’s exit rate, and the coefficients for 2 other types are
negative although not statistically significant. The two types of partners’ alliances that are
positively associated with a focal firm’s exit rate are marketing alliances and alliances with other
BFs. These findings may reflect, in part, the hazards a BF faces when its partners go on to form
new collaborative relationships with partners who are also the BF's rivals (Singh and Mitchell
1996), or change strategic direction (e.g., from product development to marketing). Although we
24
made no predictions concerning the relative size of partner alliance effects across types, Table 2
presents the effect sizes at the means for partners’ alliances. These are clustered much more
tightly than were rivals’ alliance effects. No discernible differences between partners’ upstream and
downstream alliances are evident.
The coefficient for partners’ alliance configuration efficiency has the predicted negative sign
but is not statistically significant. Thus, we reject H4.
Other effects on BF’s exit rates. Three results related to the control variables are
worth noting. First, a focal firm’s own alliances generally enhance its survival. Alliances with
pharmaceutical firms, chemical firms, and universities are all significantly and negatively
associated with a firm’s exit rate. The primary exceptions to this are a firm’s alliances with other
BFs and with government labs. These results are consistent with those in Baum et al. (2000),
who interpret the former as evidence of the difficulty of managing cooperative ventures with
direct competitors (Doz, Hamel and Prahalad 1988; Mowery, Oxley and Silverman 1996). Not all
alliances with other BFs are harmful, however. The significant, negative coefficients for partners'
recent and older patents, relative scope and several types of alliances indicate that judicious
partnering with skilled innovators, with BFs less likely to initiate learning races, or with well-
connected BFs can attenuate or reverse the hazardous main effect (Baum et al. 2000; Khanna et
al. 1998).
Second, a focal firm’s patenting activity is generally negatively associated with the firm’s exit
rate. The more pending patents, the lower is a firm’s exit rate. Similarly, the larger number of
older patents it has, the lower is its exit rate. Recently issued patents do not significantly affect
the exit rate, however. Thus, BFs that successfully commercialize past patent race victories and
those that currently hold state-of-the-art patent race leads gain survival advantages. We suspect
that the insignificance of recent patent race victories may reflect the uncertainty, time lag, and
market risks associated with development and commercialization of biotechnology patents.
Patenting activity of a firm’s partners, as we noted above, is also generally negatively associated
25
Third, rival BFs’ recent and pending patenting activities are positively associated with a focal
firm’s exit, while rivals’ older patents are negatively associated with a focal firm’s exit rate. The
increased exit rate associated with rivals’ recent and pending patents presumably reflects the ability
of rival firms to use their patent race wins (or leads) to preclude the focal firm from technological
progress or to attract the lion’s share of other resources necessary for survival. The negative
association between rivals’ older patents and a focal firm’s exit rate is consistent with Amburgey et
al.’s (1996) finding that although one firm’s patenting success in an area often gives it an advantage
in the short run, such success appears to signal technological and commercial potential of the
entire area and lead to growth in the long run.
CONCLUSION
Despite the recent explosion of research on alliances in technology-based industries, the
competitive dynamics of these phenomena remain poorly understood. We were therefore
motivated to explore the effects of a focal firm's rivals’ alliances, and its horizontal partners’
alliances, on patterns of firm survival in the Canadian biotechnology industry.
We find, as hypothesized, that rivals’ alliances are often harmful to a focal BF, but not
universally so. In particular, the effects of rivals’ alliances on a focal firm vary systematically with
the type of alliance: horizontal, upstream, or downstream. Rivals’ alliances with downstream
partners are less likely to increase a focal firm’s exit rate than their alliances with upstream
partners. In turn, rivals’ upstream alliances are less likely to increase a focal firm’s exit rate than
horizontal alliances. We also find that rivals whose alliance networks are configured more
efficiently, in terms of minimizing redundancy of type of partner, provide more intense
competition for a focal firm than those with less efficient alliance configurations.
We also find, as hypothesized, that a focal BF benefits from the alliances of those BFs with
which it collaborates. The two exceptions to this result being that a focal BF's survival chances
are harmed as its BF partners form more alliances with other BFs and marketing firms. As we
noted, this may reflect the dependence of a BF's fate on its partners' striking up new alliances with
26
the BF's rivals or shifting strategic focus after becoming the BF's partner. More generally, these
findings reinforce the precariousness of collaborating with potential rivals who collaborate, in
turn, with other potential rivals (Singh and Mitchell 1996). Contrary to our expectation, we find
that the efficiency of partners’ alliance configurations has no effect on the survival of a focal firm.
Beyond these rival and partner alliance effects, we find evidence that increases in rivals'
recent and pending patents generate stronger competition for a focal BF. However, rivals’ older
patents yielded a mutualistic effect. We interpret this effect as reflecting benefits arising from
diffusion of technological knowledge, or from signals to capital markets of a sector's viability.
Increases in partners’ recent and pending patents are negatively associated with a focal firm’s exit
rate, indicating that a focal firm benefits from partnering with more innovative BFs.
Before commenting on directions for future research, it may prove useful to comment on the
potential generalizability of our empirical strategy for examining asymmetric competitive
dynamics. Our approach, which is grounded in recent research on the dynamics of competitive
heterogeneity (e.g., Barnett and Amburgey 1990; Baum and Mezias 1992; Baum and Singh 1994;
Podolny, Stuart and Hannan 1996; Barnett 1997), suggests the value of the general method of
using firm-specific characteristics to gauge the relative competitive strengths of and potential for
interfirm competition. To date, this approach has been operationalized using variation in time-
varying organizational features including size, target market domains, technological proximity,
geographic proximity, and past operational and competitive experience. This research strategy
holds real promise for realizing a general approach to competitive dynamics that emphasizes the
role of firms' characteristics in defining organizations' positions relative to each other in a
competitive field.
That said, two specific directions for further research appear particularly promising.
Empirically, more nuanced measures of alliances and intellectual property could incorporate
insights from recent research on the structure of interorganizational networks (e.g., Burt 1992;
Gulati 1995; Powell et al. 1996; Podolny et al. 1996). For example, development of alliance and
patent measures that reflect structural holes, cumulative alliance experience, network centrality,
27
and technology network position is likely to both enhance the predictive power of our current
model and support more extensive integration of sociological and economic models of
competitive dynamics. Of course, a major obstacle to realizing such elaboration is the detailed
data required. Nevertheless, pursuit of these connections may yield important insight into sources
of firms' competitive advantages from alliances as well as intellectual property.
Theoretically, the results of this study indicate that competitive dynamics occur at the group-
versus-group level as well as firm-versus-firm. To the extent that alliance networks increase
members' interdependence as well as their capabilities, members’ success depends increasingly on
the competitive strength that groups of organizations build collectively, and competition surfaces
among alliance groups (Barnett and Carroll 1987; Gomes-Casseres 1994; Powell and Smith-
Doerr 1994). Such "group-versus-group" competition does not lessen the importance of firm-
level competition, but it does alter the nature of competition in a way that increases the
competitive significance of a firm's alliances. This suggests the importance of recasting our
analysis at the alliance-group level, as well as developing additional measures designed to capture
the collective competitive strength of alliance groups. Notably, it is possible to draw on
organization ecology (Barnett and Carroll 1987) and the resource-based view (Dyer and Singh
1998) to develop theoretical implications regarding group-vs.-group dynamics, just as we drew on
these theories for firm-level insights in this study.
In conclusion, this study, the first to investigate the competitive dynamics of alliances, moves
organization theory and strategic management scholarship forward in their understanding of the
influence of rivals’ and partners’ alliances on focal firm survival and interfirm competition.
28
References
Aldrich, H.E. and C.M. Fiol. 1994. “Fools ruch in? The institutional context of industry creation.”
Academy of Management Review, 19:645-670.
Amburgey, T.L., T. Dacin and J. Singh. 1996. “Learning races, patent races, and capital races:
Strategic interaction and embeddedness within organizational fields.” In J.A.C. Baum and J. E.
Dutton (eds.) Advances in Strategic Management, 13: 303-322.
Argyres, N.S. and J.P. Liebeskind. 1998. “Privatizing the intellectual commons: Universities and
the commercialization of biotechnology.” Journal of Economic Behavior and Organization, 35:
427-454.
Austin, D.H. 1993. “An event-study approach to measuring innovative output: The case of
American Economic Review, 83: 253-258.
Barnett, W.P. 1997. “The dynamics of competitive intensity.” Administrative Science Quarterly, 42:
128-160.
Barnett, W.P. and T.L. Amburgey. 1990. "Do larger organizations generate stronger
competition?" In J. V. Singh (ed.), Organizational Evolution: New Directions: 78-102.
Newbury Park, CA: Sage.
Barnett, W.P. and G.R. Carroll 1987. Competition and mutualism among early telephone
companies." Administrative Science Quarterly, 32: 400-421.
Baum, J.A.C. 1996. “Organizational ecology.” In S. Clegg, C. Hardy, and W. Nord (eds.)
Handbook of Organization Studies: 77-114. London: Sage.
Baum J.A.C. and S.J. Mezias 1992. "Localized competition and organizational mortality in the
Manhattan hotel industry, 1898-1990," Administrative Science Quarterly, 37: 580-604.
29
Baum, J.A.C. and C. Oliver. 1992. “Institutional embeddedness and the dynamics of organizational
populations.” American Sociological Review, 57: 540-559.
Baum, J.A.C. and J.V. Singh. 1994. "Organizational niches and the dynamics of organizational
mortality." American Journal of Sociology, 100: 346-380.
Baum, J.A.C., T. Calabrese and B.S. Silverman. 2000. “Don’t go it alone: Alliance networks and
startups’ performance in Canadian biotechnology.” Strategic Management Journal, 21: 267-
294.
Blossfeld, H.P. and G. Rohwer. 1995. Techniques of Event History Modeling: New Approaches
to Causal Analysis. Mahwah, NJ: Lawrence Erlbaum Associates.
Burt, R.S. 1992. Structural Holes: The Social Structure of Competition. Cambridge MA: Harvard
University Press.
Calabrese, T., J.A.C. Baum and B.S. Silverman. 2000. “Canadian biotechnology start-ups, 1991-
1997: The role of incumbents’ patents and strategic alliances in controlling competition.” Social
Science Research, forthcoming.
Canadian Biotech. Various years. Toronto, ON: Ernst & Young.
Canadian Biotechnology. Various years. Toronto, ON: Contact Canada.
Doz, Y., G. Hamel and C.K. Prahalad. 1989. “Collaborate with your competitors–and win.” Harvard
Business Review, 67(Jan-Feb): 133-139.
Dyer, J.H. and H. Singh. 1998. “The relational view: Cooperative strategy and sources of
interorganizational competitive advantage,” Academy of Management Review, 23: 660-679.
Fligstein, N. 1996. “Markets as politics: A political-cultural approach to market institutions.”
American Sociological Review, 61:656-673.
30
Gomes-Casseres, B. 1994. "Group versus group: How alliance networks compete." Harvard
Business Review, (July-Aug): 62-74.
Griliches, Z. 1990. "Patent statistics as economic indicators: A survey. Journal of Economic
Literature, 28: 1661-1707.
Gulati, R. 1995. "Social structure and alliance formation patterns: A longitudinal analysis."
Administrative Science Quarterly, 40: 619-652.
Guo, G. 1993. "Event-history analysis for left-truncated data." Sociological Methodology, 23:
217-244.
Hannan, M.T. and G.R. Carroll. 1992. Dynamics of Organizational Populations. New York:
Oxford University Press.
Hawley, A.H. 1986. Human Ecology: A Theoretical EssayHuman Ecology: A Theoretical Essay.
Chicago; University of Chicago Press.
Kennedy, P. 1992. A Guide to Econometric Methods (3d edition). Cambridge, MA: MIT Press.
Khanna, T., R. Gulati and N. Nohria. 1998. “The dynamics of learning alliances: Competition,
cooperation and relative scope.” Strategic Management Journal, 19: 193-211.
Kmenta, J. 1971. Elements of Econometrics. New York: Macmillan.
Kogut, B. 1988. “Joint ventures: Theoretical and empirical perspectives.” Strategic Management
Journal, 9: 319-332.
Kogut, B., W. Shan, and G. Walker. 1992. “The make-or-cooperate decision in the context of an
Nohria and R.G. Eccles (eds.) Networks and Organizations:
Structure, Form and Action: 348-365. Boston: HBS Press.
31
Leik, R. K. 1992. “New directions for network exchange theory: Strategic manipulation of
Social Networks, 14: 309-323.
Lerner, J. 1994. “Venture capitalists and the decision to go public.” Journal of Financial Economics,
35: 293-316.
Liebeskind, J. P., A.L. Oliver, L. Zucker, and M. Brewer 1996. “Social networks, learning and
flexibility: Sourcing scientific knowledge in new biotechnology firms.” Organization Science,
7: 428-443.
Miner, A.S., T.L. Amburgey and T. Stearns. 1990. “Interorganizational linkages and population
dynamics: Buffering and transformational shields.” Administrative Science Quarterly, 35: 689-
713.
Mowery, D.C., J.E. Oxley, and B.S. Silverman 1996. “Strategic alliances and interfirm knowledge
transfer.” Strategic Management Journal, 17 (Special Issue): 77-92.
Oxley, J.E. 1997. “Appropriability hazards and governance in strategic alliances: A transaction cost
approach.” Journal of Law, Economics and Organization, 13: 387-409.
Pisano, G.P. 1990. “The R&D boundaries of the firm: An empirical analysis.” Administrative
Science Quarterly, 35: 153-176.
Podolny, J.M. 1994. “"Market uncertainty and the character of economic exchange."
Administrative Science Quarterly, 39: 458-483
Podolny, J.M., T. Stuart, and M.T. Hannan 1996. "Networks, knowledge, and niches." American
Journal of Sociology, 102: 659-689.
32
Powell, W.W. and L. Smith-Doerr 1994. "Networks and economic life." In N.J. Smelser and R.
Swedberg (eds.) Handbook of Economic Sociology: 368-402. Princeton NJ: Princeton
University Press.
Powell, W.W., K.W. Koput, and L. Smith-Doerr. 1996. “Interorganizational collaboration and the
locus of innovation: Networks of learning in biotechnology.” Administrative Science
Quarterly, 41: 116-145.
Rohwer, G. 1995. TDA (Transitional Data Analysis) 5.7. Florence: European University Institute.
Singh, K. and W. Mitchell. 1996. “Precarious collaboration: Business survival after partners shut
down or form new partnerships.” Strategic Management Journal, 17 (Summer): 99-115.
Sorenson, O. and T.E. Stuart. 2000. “Syndication networks and the spatial distribution of venture
capital investments.” Working paper. Anderson School of Management, UCLA and Graduate
School of Business, University of Chicago.
Spencer, W.J. and P. Grindley. 1993. “SEMATECH after five years: High technology consortia
and U.S. competitiveness” California Management Review, 35(4): 9-33
Stuart, T.E., H. Hoang and R.C. Hybels. 1999. “Interorganizational endorsements and the
performance of entrepreneurial ventures.” Administrative Science Quarterly, 44: 315-349.
Teece, D.J. 1992. “Competition, cooperation, and innovation: Organizational arrangements for
regimes of rapid technological progress.” Journal of Economic Behavior and Organization, 18:
1-25.
Uzzi, B. 1996. "The sources and consequences of embeddedness for the economic performance of
organizations: The network effect." American Sociological Review, 61:674-698.
33
Uzzi, B. 1997. "Social structure and competition in interfirm networks: The paradox of
embeddedness." Administrative Science Quarterly, 42: 35-67.
Walker, G., B. Kogut and W.J. Shan. 1997. “Social capital, structural holes and the formation of
an industry network.” Organization Science, 8: 109-125.
Williamson, O.E. 1991. “Comparative economic organization: The analysis of discrete structural
Administrative Science Quarterly, 36: 269-296.
Zucker, L.G., M.R. Darby, and M.B. Brewer. 1994. “Intellectual Capital and the Birth of US
Biotechnology Enterprises.” NBER Working Paper #4653.
34
36
Table 1. Conditional Discrete Time Models of BFs' Industry Exit, 1991-1996
Model 1
Model 2
Model 3
Model 4
Firm Characteristics
Included Included Included Included
Sector Characteristics
Included
Included
Included
Included
Own Alliances & Patents
Biotechnology .159 0.164 .345*
University -.214* -.222* -.209*
Research Institute -0.092 -0.131 -0.131
Government Lab 0.204 0.237 .329+
Hospital -0.153 -0.166 -0.154
Industry Association -0.611 -0.635 -0.630
Pharmaceutical -.255* -.288* -.283*
Chemical -.269+ -.273+ -.268+
Marketing -0.028 -0.035 -0.028
Alliance Configuration Efficiency x10 -.243* -.273* -.264*
Pending Patents -.341* -.386* -.371*
Recent Patents -0.015 -0.017 -0.009
Older Patents -.080* -.036+ -.034+
Rival BFs' Alliances & Patents
Multipliers
Biotechnology (Horizontal) .038* .041* 16.86
University (Upstream) -.036* -.034* .10
Research Institute (Upstream) .127* .130* 3.35
Government Lab (Upstream) .056* .062* 6.08
Hospital (Upstream) .091+ .099+ 2.12
Industry Association (Upstream) .127+ .129+ 1.93
Pharmaceutical (Downstream) -.026 -.023 --
Chemical (Downstream) -.029 -.021 --
Marketing (Downstream) .006* .006* 1.72
Alliance Configuration Efficiency .241* .242*
Pending Patents .029* .030*
Recent Patents .024* .023*
Older Patents -.023* -.019*
Partner BFs' Patents/Alliances
Biotechnology x10 (for rescaling) .092* 1.33
University x10 -.182* 0.86
Research Institute x10 -.137* 0.97
Government Lab x10 -0.141 --
Hospital x10 -0.077 --
Industry Association x10 -.362* 0.96
Pharmaceutical x10 -.160* 0.87
Chemical x10 -.224* 0.91
Marketing x10 .033* 1.05
Alliance Configuration Efficiency -0.144
Pending Patents -0.026
Recent Patents -.063*
Older Patents -.074*
Relative Scope -.329*
Constant -2.277* -2.152* -3.118* -3.305*
Log Likelihood -587.14 -574.46 -559.04 -541.86
Likelihood Ratio Test vs. model 1 25.34;13
df*
Likelihood Ratio Test vs. model 2
30.84;13df*
Likelihood Ratio Test vs. model 3 34.36;14
df
*
+ p<.10 * p<.05 The sample included 2,163 yearly spells and 186 industry exits.
Coefficients for firm and sector controls are given in Appendix Table A2
Values for all variables are computed at the start of each yearly spell.
Alliance effect multipliers are and evaluated at each variable's mean based on Model 4 coefficients.
37
Appendix Table A1. Sector Fixed Effects Coefficients for Table 1
Model 1
Model 2
Model 3
Model 4
Firm Characteristics
Firm Age -0.005 -0.005 -.009+ -.009+
Left Censored .548* .557* .569* .571*
Biotech Employees -.039* -.051* -.056* -.051*
Biotech R&D Employees -.040* -.037* -.031* -.029*
Manufacturing Facility -.233+ -0.208 -.241+ -0.228
Revenues ($mil) 0.066 0.057 0.071 0.073
R&D Expenditures ($mil) 0.089 .162* .168* .168*
Diversification (# of Sectors) -.189* -.163* -0.086 -0.077
Private (omitted ownership dummy) -- -- -- --
IPO 1991-96 -1.007* -.869* -.829* -.804*
Public (IPO before 1991) -0.027 -0.032 -0.078 -0.095
Nonprofit -0.344 -0.529 -0.541 -0.531
Government/University/Hospital -1.161* -1.093* -1.296* -1.263*
Subsidiary .396* .434* .431* .436*
Joint Venture -0.939 -0.751 -0.803 -0.791
Agriculture -.478* -.592* -.646+ -0.623
Aquaculture -0.119 -0.169 -0.533 -0.488
Engineering -1.351* -1.562* -1.867* -0.945
Environment -.555* -.714* -1.287* -1.238*
Food/Beverage/Fermentation 0.116 -0.031 -0.238 -0.082
Forestry .727* .648* .927* 1.263*
Human Diagnostics 0.088 0.092 0.324 0.313
Human Therapeutics (omitted sector dummy) -- -- -- --
Human Vaccines -0.621 -0.685 -0.707 -0.522
Horticulture -1.164* -1.351* -1.372* -1.285*
Contract Research Organization -1.477* -1.579* -1.673* -1.596*
Veterinary -0.833 -0.759 -0.769 -0.623
Energy -0.235 -0.337 -0.344 -0.279
Biomaterials/Cosmetics/Mining 0.107 0.204 0.183 0.744
Sector Characteristics
Aggregate Bio Financing ($mil) -.0037* -.0031* -.0035* -.0032*
BF's Pending Patents x
Aggregate Bio Financing ($mil) -.0011* -.0009* -.0010* -.0008*
Number Canadian BFs -0.0013 -0.0003 -0.0084 -0.0070
Mass Canadian BFs .204* .211* .253* .261*
Number U.S. BFs 0.0003 0.0001 0.0002 0.0003
Number U.S. BFs' Recent Patents 0.009* 0.009* 0.006* 0.006*
+ p,.10; * p<.05; the sample included 2,163 yearly spells and 186 industry exits.
38
Appendix A2. Descriptive Statistics for Own, Partners' and
Rivals' Alliance and Patent Variables
Own Alliance and Patent Variables
Variable Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1. Biotech 0.51 1.19 1.00
2. Pharmaceutical 0.30 0.84 0.29 1.00
3. Chemical 0.15 0.52 0.19 0.05 1.00
4. University 0.62 1.38 0.14 0.16 0.02 1.00
5. Institute 0.32 0.69 0.04 0.04 0.08 0.21 1.00
6. Government 0.11 0.39 0.03 -0.01 0.07 0.08 0.24 1.00
7. Hospital 0.07 0.38 0.04 0.02 -0.03 0.19 0.10 0.02 1.00
8. Industry Assoc. 0.05 0.34 -0.01 -0.01 -0.05 0.02 0.03 0.04 -0.02 1.00
9. Marketing 0.81 2.62 -0.03 0.01 -0.02 -0.02 0.00 -0.02 0.03 -0.03 1.00
10. Alliance Eff. 0.08 0.39 0.11 0.09 0.08 0.12 0.21 0.17 0.07 0.03 -0.07 1.00
11. Older Patents 0.36 4.43 0.01 0.02 0.02 0.00 -0.03 0.00 -0.01 0.00 0.00 0.00 1.00
12. Patent Apps. 0.22 1.48 0.05 0.13 -0.02 0.05 0.06 0.03 -0.01 0.01 0.01 0.05 0.09 1.00
13. Recent Patents 0.56 2.71 0.05 0.11 0.02 0.01 0.04 0.03 -0.02 0.02 0.00 0.08 0.30 0.46 1.00
14. Relative Scope 0.54 2.14 0.15 0.15 0.17 -0.01 0.02 -0.02 0.00 -0.02 0.00 0.03 -0.01 0.00 0.00 1.00
Partner BFs' Alliance and Patent Variables
Variable Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
15. Older Patents 0.24 3.78 0.17 0.16 0.02 0.06 0.03 0.00 -0.01 0.00 0.02 0.00 0.01 0.02 0.01 0.10
16. Patent Apps. 0.21 3.99 0.17 0.25 0.01 0.07 0.01 0.03 0.00 0.00 -0.01 0.02 -0.01 0.01 0.01 0.13
17. Recent Patents 0.37 4.49 0.31 0.32 0.03 0.11 0.03 0.04 0.02 0.01 -0.01 0.01 0.00 0.02 0.02 0.14
18. Biotech 0.31 1.23 0.23 0.13 0.05 0.04 0.06 0.00 0.00 0.00 0.05 0.13 0.00 0.08 0.07 0.08
19. Pharmaceutical 0.09 0.51 0.16 0.11 0.01 0.02 0.02 0.02 0.01 -0.01 0.02 0.27 0.10 0.09 0.06 0.02
20. Chemical 0.04 0.30 0.10 0.02 0.06 0.01 0.06 -0.02 -0.02 0.01 -0.02 0.10 0.00 0.02 0.02 0.03
21. University 0.08 0.48 0.20 0.12 0.02 0.04 0.07 0.00 -0.01 0.02 0.03 0.08 0.00 0.11 0.16 0.11
22. Institute 0.02 0.13 0.12 0.00 0.00 -0.02 0.01 0.02 -0.01 0.00 -0.01 0.05 0.00 0.05 0.04 0.07
23. Government 0.06 0.30 0.17 0.07 0.04 0.03 0.05 -0.01 0.02 -0.01 0.00 0.10 -0.01 0.03 0.03 0.11
24. Hospital 0.02 0.14 0.12 0.08 -0.01 0.00 0.05 -0.02 0.00 0.00 0.04 0.08 0.00 0.05 0.07 0.16
25. Industry Assoc. 0.01 0.05 0.10 0.02 0.01 0.00 0.01 0.02 -0.01 0.01 0.02 0.02 0.00 0.04 0.07 0.02
26. Marketing 0.16 1.75 0.07 0.05 0.00 -0.01 0.03 -0.02 0.02 -0.01 0.01 0.07 0.00 0.00 0.00 0.03
27. Alliance Eff. 2.2 2.3 0.06 0.04 0.07 0.03 0.02 0.04 0.02 0.03 -0.02 -0.01 0.02 0.03 0.03 0.04
Rival BFs' Alliance and Patent Variables
Variable Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
28. Biotech 68.9 31.2 0.00 -0.01 0.10 0.01 0.01 -0.01 -0.03 0.04 -0.08 0.04 0.00 0.01 0.08 -0.01
29. Pharmaceutical 34.3 25.6 0.08 0.13 0.06 0.13 0.04 -0.04 0.06 -0.01 -0.06 0.08 -0.01 -0.02 0.06 0.02
30. Chemical 25.9 19.1 -0.02 -0.07 0.10 -0.03 0.01 0.02 -0.07 0.05 -0.08 0.01 0.00 0.00 0.07 -0.01
31. University 67.2 35.7 0.03 0.03 0.10 0.04 0.04 0.00 -0.01 0.01 -0.07 0.06 0.00 0.00 0.10 0.01
32. Institute 9.3 4.0 -0.01 -0.06 0.11 -0.01 0.03 0.01 -0.06 0.04 -0.10 0.01 0.00 -0.01 0.07 -0.01
33. Government 29.1 11.3 0.00 -0.04 0.11 0.00 0.02 0.02 -0.05 0.04 -0.10 0.01 0.00 0.00 0.08 -0.01
34. Hospital 7.6 7.2 0.09 0.19 -0.04 0.15 0.01 -0.06 0.09 -0.05 0.00 0.10 -0.01 0.04 0.06 0.03
35. Industry Assoc. 5.1 2.9 -0.02 -0.05 0.09 -0.02 0.01 0.02 -0.05 0.01 -0.08 0.02 0.00 0.02 0.10 -0.01
36. Marketing 90.4 72.1 0.04 0.01 0.04 0.03 -0.04 -0.04 0.00 0.02 0.02 0.03 0.01 0.02 0.06 0.00
37. Alliance Eff. 4.1 2.2 0.04 0.07 -0.11 0.04 -0.09 -0.08 0.13 -0.03 0.20 0.06 0.03 0.06 0.00 0.02
38. Older Patents 14.8 11.9 0.00 0.04 -0.10 0.03 -0.06 -0.05 0.11 0.00 0.07 0.04 -0.34 0.02 -0.07 0.00
39. Patent App. 13.2 16.5 0.00 0.08 -0.10 0.07 -0.07 -0.08 0.15 -0.06 0.09 -0.03 0.06 0.06 -0.01 0.02
40. Recent Patents 29.7 22.1 -0.01 0.13 -0.13 0.08 -0.07 -0.09 0.21 -0.02 0.10 0.05 -0.02 0.03 -0.02 -0.01
39
Appendix A2. Descriptive Statistics for Own, Partners' and
Rivals' Alliance and Patent Variables (continued)
Partner BFs' Alliance and Patent Variables
Variable 15 16 17 18 19 20 21 22 23 24 25 26 27
15. Older Patents 1.00
16. Patent Apps. 0.11 1.00
17. Recent Patents 0.36 0.66 1.00
18. Biotech 0.13 0.26 0.27 1.00
19. Pharmaceutical 0.08 0.46 0.41 0.57 1.00
20. Chemical -0.04 0.02 0.04 0.41 0.22 1.00
21. University 0.18 0.35 0.30 0.38 0.29 0.23 1.00
22. Institute 0.12 0.03 0.02 0.21 0.09 0.20 0.19 1.00
23. Government 0.14 0.26 0.20 0.45 0.33 0.38 0.42 0.33 1.00
24. Hospital 0.16 0.22 0.12 0.25 0.21 0.02 0.22 0.22 0.36 1.00
25. Industry Assoc. 0.07 0.15 0.10 0.14 0.07 0.01 0.11 0.03 0.04 0.09 1.00
26. Marketing 0.03 0.07 0.06 0.10 0.09 0.02 0.06 0.09 0.11 0.19 0.02 1.00
27. Alliance Eff. 0.02 0.02 0.09 0.13 0.04 0.05 0.07 0.05 0.04 0.02 0.05 -0.04 1.00
Rival BFs' Alliance and Patent Variables
Variable 15 16 17 18 19 20 21 22 23 24 25 26 27
28. Biotech -0.02 -0.01 -0.05 -0.03 -0.03 -0.01 -0.02 0.01 -0.03 -0.01 0.00 0.00 0.03
29. Pharmaceutical -0.01 0.02 0.05 0.05 0.02 0.01 0.04 0.04 0.01 0.04 0.03 0.00 0.07
30. Chemical 0.00 -0.02 -0.02 -0.04 -0.04 -0.01 -0.03 0.00 -0.03 -0.03 -0.02 0.00 -0.01
31. University 0.00 0.01 0.02 0.02 0.00 0.01 0.00 0.03 0.00 0.00 0.01 0.01 -0.03
32. Institute 0.01 -0.02 -0.02 -0.03 -0.05 -0.01 -0.03 -0.01 -0.04 -0.03 -0.01 -0.01 0.01
33. Government -0.01 -0.01 -0.01 -0.02 -0.03 0.00 -0.02 0.01 -0.03 -0.03 0.00 0.00 0.01
34. Hospital 0.04 0.06 0.09 0.09 0.08 0.00 0.06 0.04 0.03 0.05 0.05 -0.01 0.06
35. Industry Assoc. 0.00 -0.02 -0.01 -0.03 -0.03 -0.01 -0.02 0.00 -0.03 -0.02 -0.02 0.01 -0.02
36. Marketing 0.00 0.00 0.01 -0.02 -0.02 -0.04 -0.02 0.01 -0.02 0.03 0.01 -0.03 0.03
37. Alliance Eff. 0.02 0.05 0.03 0.05 0.09 -0.05 0.02 0.00 0.02 0.08 0.01 0.06 -0.01
38. Older Patents 0.01 0.00 -0.02 0.01 0.02 -0.03 0.01 0.01 0.01 0.05 -0.01 0.03 0.04
39. Patent App. -0.03 -0.04 -0.06 0.02 0.06 -0.04 0.00 -0.02 -0.01 0.03 -0.01 0.03 0.02
40. Recent Patents -0.01 -0.04 -0.06 0.04 0.04 -0.03 0.01 0.01 0.01 0.06 0.01 0.03 0.04
Rival BFs' Alliance and Patent Variables
Variable 28 29 30 31 32 33 34 35 36 37 38 39 40
28. Biotech 1.00
29. Pharmaceutical 0.66 1.00
30. Chemical 0.29 -0.29 1.00
31. University 0.73 0.73 -0.08 1.00
32. Institute 0.42 0.12 0.50 0.35 1.00
33. Government 0.45 0.42 0.51 0.62 0.71 1.00
34. Hospital 0.56 0.75 -0.44 0.78 0.00 0.40 1.00
35. Industry Assoc. 0.37 0.05 0.61 0.22 0.54 0.57 -0.06 1.00
36. Marketing 0.40 0.43 -0.26 0.37 -0.26 -0.07 0.48 -0.21 1.00
37. Alliance Eff. 0.65 0.61 -0.14 0.59 -0.10 0.23 0.55 -0.02 0.69 1.00
38. Issued >5 yrs. 0.37 0.44 -0.18 0.41 -0.05 0.15 0.45 0.02 0.35 0.42 1.00
39. Patent App. 0.50 0.63 -0.39 0.58 -0.17 0.20 0.65 -0.18 0.47 0.69 0.33 1.00
40. Issued <5 yrs. 0.66 0.66 -0.29 0.68 0.02 0.41 0.68 0.15 0.53 0.68 0.55 0.65 1.00
Note: The sample included 2,163 yearly spells. Standard Deviations are not given for
dummy variables. Correlations >.04 are significant at p<.05.
40
Figure 1: The Effect of Different Types of Alliances on Competitive Intensity
Degree of foreclosure of rivals
Increase in carrying capacity
Competitive intensity
Degree of foreclosure of rivals
Increase in carrying capacity
Degree of foreclosure of rivals
Increase in carrying capacity
downstream
alliances
upstream
alliances
horizontal
alliances
Low
High