Content uploaded by Victor Cui
Author content
All content in this area was uploaded by Victor Cui on Dec 03, 2022
Content may be subject to copyright.
ATTACKING YOUR PARTNERS: STRATEGIC ALLIANCES AND
COMPETITION BETWEEN PARTNERS IN PRODUCT MARKETS
Citation:
Cui, V., Yang, H., & Vertinsky, I. (2018). Attacking your partners:
Strategic alliances and competition between partners in product
markets. Strategic Management Journal, 39(12), 3116-3139.
1
ATTACKING YOUR PARTNERS: STRATEGIC ALLIANCES AND COMPETITION
BETWEEN PARTNERS IN PRODUCT MARKETS
ABSTRACT
Research summary: This study contributes to the literature on strategic alliances by examining
the impact of collaboration on competition between partners in product markets. We integrate the
alliance learning and social network perspectives to examine how different combinations of
exploratory and exploitative alliances between a firm and its partner influence the firm’s
competition against its partner in product markets. Using a longitudinal dataset collected in the
U.S. pharmaceutical industry (1984–2003), we find an inverted U-shaped relationship between
relative exploration (i.e., the proportion of exploratory alliances in the collaborative portfolio
between a firm and its partner) and the firm’s competition against its partner. This relationship is
negatively moderated by firms’ relational and structural embeddedness, but positively moderated
by their positional embeddedness.
Managerial summary: This study examines how different combinations of exploratory and
exploitative alliances between two firms affect their competition in the product market. Using a
20-year dataset collected in the U.S. pharmaceutical industry, we find that the proportion of
exploratory alliances (i.e., joint development of critical innovations) in the alliance portfolio
between a firm and its partner increases the firm’s competition against its partner, up to a tipping
point at which such competition starts to decline. Given a certain combination of the two types of
alliances, such competition is stronger if the firm has more alternative allies than its partner but
weaker if the firm and its partner have previously collaborated or share common allies in their
networks.
Page 1 of 40
2
Introduction
Research on strategic alliances has revealed that prior collaboration often leads to further
collaboration between firms (Barden and Mitchell, 2007; Gulati, 1995; Li and Rowley, 2002;
Podolny, 1994). However, ample anecdotal evidence indicates that alliances are often followed
by aggressive competition between the same firms in product markets. For example,
pharmaceutical firm Merck allied with Novartis, Pfizer, Bristol-Myers, and Depomed in the
1990s yet later competed with these firms in product markets. Similar instances appear in the
automobile (Hamel, Doz, and Prahalad, 1989), telecommunications (Hille and Taylor, 2011;
Ritman, 2006), and other industries. Despite the abundant evidence, however, researchers have
not paid adequate attention to the influence of alliances on competition between partners in
product markets.
A thorough study of this ‘collaboration–competition’ relationship between partners is of
great theoretical importance, contributing to the development of a more comprehensive model of
inter-firm behavior rendered by strategic alliances (Kogut, 1989). Prior studies on alliance
learning have provided some important insights into the tension between collaboration and
competition (Hamel et al., 1989; Katila, Rosenberger, and Eisenhardt, 2008; Khanna, Gulati, and
Nohria, 1998). For example, researchers maintain that competition within alliances stems from
the misalignment of interests between allies (Gulati and Singh, 1998; Hamel et al., 1989) and
have identified important factors that influence allies’ competitive learning within alliances, such
as asymmetric learning capabilities (Hamel, 1991; Khanna et al., 1998; Yang, Zheng, and Zaheer,
2015), the ratio between private and common interests (Khanna et al., 1998), and knowledge
similarities between allies (Dussauge, Garrette, and Mitchell, 2000). However, three important
issues in this sphere of research remain unaddressed.
Page 2 of 40
3
First, while researchers have examined aggressive learning between allies, whereby allies
compete to increase private rather than common benefits (Hamel, 1991; Khanna et al., 1998;
Lavie, 2007), prior studies focused on the hazards of misappropriation within alliances; the effect
of alliances on competition between partners in the realm of product markets remains poorly
understood.
Second, prior studies provide insights into competitive learning between partners by
focusing on research-based alliances while overlooking other types of collaboration (e.g., Yang
et al., 2015). It is assumed that research-based alliances entail intensive knowledge exchange and
therefore create many opportunities for competitive learning between allies, but firms often
simultaneously engage in multiple types of alliances (e.g., research-based, marketing,
manufacturing) that differ significantly in the incentives they generate for allies to compete with
one another (Hamel, 1991; Khanna et al., 1998; Mowery, Oxley, and Silverman, 1996). The
extent to which a firm competes with its partner is arguably determined by the composition of all
types of alliances binding them, which collectively define the overall orientation of their
interactions as well as the costs and benefits of competitive moves against each other (Chen and
Miller, 2015). Few studies have yet examined how the composition of the collaborative portfolio
between firms affects their competition with one another.
Third, many prior studies have examined the tension between cooperation and
competition by focusing on characteristics of the allying firms per se, such as their relative
learning capabilities (Yang et al., 2015) and knowledge similarities (Dussauge et al., 2000),
while largely overlooking the impact of the broad inter-firm alliance networks within which
competitive learning between partners is embedded (Uzzi, 1996). Only a few studies have
analyzed the collaboration–competition phenomenon within alliances from the network
Page 3 of 40
4
perspective. For example, Lavie (2007) provides insights into how a firm’s appropriation
capacity within alliances is affected by factors such as the availability of alternative alliances and
multilateral competition among its partners. Polidoro, Ahuja, and Mitchell (2011) study the
impact of network embeddedness on partnering firms’ competitive incentives and, in turn, on
joint venture dissolution.
1
Yet the focus of these studies remains centered either on allies’ market
performance or on the survival of joint ventures; the impact of partners’ network embeddedness
on the interplay between alliances and competition in the product market has rarely been studied.
Our study addresses these limitations by integrating the alliance learning and network
perspectives in order to study the mechanisms through which strategic alliances influence allies’
competition in the product market. We ask, how does the composition of the collaborative
portfolio between a firm and its partner affect the firm’s competitive aggressiveness against its
partner in the product market, and how does network embeddedness influence this relationship?
We capture the composition of two firms’ collaborative portfolio using the concept of
relative exploration, defined as the proportion of exploratory collaborations among all
collaborations between a firm and its partner (Uotila et al., 2009; Yang et al., 2011).
2
As
conceptualized in the alliance learning literature, exploratory collaboration is oriented toward
developing critical innovations, which require intensive interactions, tacit knowledge sharing,
and relational capital building for long-term benefits (Lavie and Rosenkopf, 2006), whereas
exploitative collaboration is oriented toward transacting existing resources for short-term
1
Other studies have investigated the impact of embedded relationships on competition in different
contexts, such as clique-level rivalries between alliances (Gimeno, 2004) and industry-level competition
not specified for a particular target (Gnyawali and Madhavan, 2001; Andrevski, Brass, and Ferrier, 2016).
2
Alternatively, we could use the proportion of exploitative alliances among all alliances between a firm
and its partner, or relative exploitation. The mechanisms will be the same, except that the predicted
relationship will flip.
Page 4 of 40
5
economic benefits (Mowery et al., 1996).
We maintain that increases in the proportion of exploratory alliances within the
collaborative portfolio between a firm and its partner facilitate identification of the partner’s
vulnerabilities and appropriation of its capacities, increasing the firm’s incentive to launch
competitive action against its partner (Yu, Subramaniam, and Cannella, 2009). However, there is
a cost/benefit trade-off involved in launching competitive attacks. As the proportion of
exploratory alliances increases, the escalating damage to long-term benefits and the risk of ‘tit-
for-tat’ retaliatory attacks from the partner may reach a threshold at which the expected cost
becomes higher than the expected benefit of launching further competitive attacks. We
accordingly propose that relative exploration demonstrates an inverted U-shaped relationship
with a firm’s product-market competition with its partner.
We further argue that the cost/benefit trade-off of launching competitive attacks is
bounded by firms’ network environments. Firms’ relational embeddedness (i.e., repeated alliance
ties), positional embeddedness (i.e., relative network centrality), and structural embeddedness
(i.e., common partners) within their alliance networks create important boundary conditions that
moderate the effect of relative exploration on their competition in different directions. We test
our hypotheses using a longitudinal dataset that integrates data on firms’ alliances and product
competition in the U.S. pharmaceutical industry over a period of two decades (1984–2003).
Theory and Hypotheses
Although strategic alliances are inter-firm partnerships formed to jointly develop, manufacture,
and distribute technologies and products (Gulati, 1998), partnering firms often simultaneously
collaborate and compete within their alliances (Hamel, 1991). For example, Hamel et al. (1991,
1989) observe that firms have a propensity to maximize their own benefits within alliances, even
Page 5 of 40
6
at the cost of common interests. Other researchers suggest that firms race to acquire valuable
intellectual properties and organization-specific knowledge from each other so that faster
learners can reap more private benefits (Khanna et al., 1998).
While competitive learning is an important phenomenon in various types of
collaborations, this study focuses on horizontal alliances established between firms operating
within the same industry. Firms in horizontal alliances normally have a strong competitive
inclination toward each other because of either existing or potential rivalries regarding products
or resources (Chen, 1996). The competitive nature of their relationship underpins their alliance
interactions, making horizontal collaboration an appropriate setting for studying the mechanisms
through which competitive alliance learning leads to competition in the product market.
It is likely that aggressive learning within horizontal alliances can be translated into
competition between allies in product markets, for two reasons. First, proprietary technological
knowledge obtained from a partner can be applied outside the alliance when contracts do not
adequately direct and constrain the use of knowledge-based assets (Khanna et al., 1998). Firms
can apply such knowledge to develop or improve their own products, constituting a threat to their
partners’ existing products. Second, learning about partners’ organizational systems, behavioral
patterns, and intentions can also help firms design calibrated attacks on their partners in the
product market in order to maximize their own performance (Yu and Cannella, 2007).
Relative exploration and product competition between allies
We classify alliances into two categories: exploitative and exploratory (Koza and Lewin, 1998;
March, 1991). Exploitative collaborations, such as marketing and licensing alliances, are arm’s-
length relationships designed to exchange existing knowledge and resources for short-term
economic returns (Mowery et al., 1996). Firms engaged in exploitative collaborations typically
Page 6 of 40
7
do not require close interactions, since they can be guided by contracts to accomplish their
respective duties in a relatively stand-alone fashion (Rothaermel and Deeds, 2004). In contrast,
exploratory collaborations are designed to synthesize knowledge assets from both parties to
develop critical innovations of great strategic importance for a firm’s long-term stakes, and
consequently demand more intensive interactions for sharing tacit know-how (Lavie and
Rosenkopf, 2006; Rothaermel and Deeds, 2004). Exploratory collaborations cannot be perfectly
regulated by contracts, since they consist of non-routinized searches and learning that involves
experimentation with new and often unforeseen alternatives and outcomes.
Different compositions of the collaborative portfolio between two firms have different
implications for their competition in the product market. When the proportion of exploratory
alliances is low, the overlap in the two firms’ long-term stakes is relatively small, so that a short-
term horizon and a ‘transaction-oriented’ perspective dominate their interactions. This
orientation allows both parties to tolerate self-serving behaviors in their interactions, or even to
take them for granted. Their interactions then focus on maximizing immediate profits, so that
each firm is motivated to use what it learns from its partner to increase private benefits. In this
type of orientation, increases in the proportion of exploratory alliances provide opportunities and
incentives to compete in product markets. Specifically, there are two reasons.
First, increases in the proportion of exploratory alliances enhance a firm’s awareness of
opportunities to benefit from competition. Exploratory collaborations involve more intensive
interactions, which enable the firm to gain a deeper and more comprehensive understanding of
its partner’s technologies, organizational operations, leadership, and motives (Davis and
Eisenhardt, 2011). This knowledge helps the firm more precisely identify the strengths and
weaknesses in its partner’s organizational systems and innovation schemes (Dussauge et al.,
Page 7 of 40
8
2000; Yang, Lin, and Peng, 2011), information that can be used to launch calibrated attacks and
provides strong incentives for competitive actions.
Second, exploratory collaboration enhances a firm’s capacity to develop competing
products because it requires sharing of proprietary know-how. Exploratory collaboration
involves articulating and transferring complex knowledge between allies, which requires
intensive hands-on coaching to facilitate joint problem solving. Such coaching often draws on
latent knowledge bases, showing the connections between divisional areas of knowledge and
organizations (Agrawal, 2006; Uzzi, 1997). This fine-grained knowledge transfer enhances the
absorption of tacit know-how (Lane and Lubatkin, 1998) and facilitates the appropriation of
fundamental knowledge that spills over during close interactions. Firms’ incentives to apply such
technological know-how in designing substitute products is high, since doing so can cause
destructive damages to incumbent products and reward the attackers with enhanced private
benefits (Chen and Hambrick, 1995; Lichtenberg and Philipson, 2002; Yu and Cannella, 2007).
The above arguments suggest that increasing the proportion of exploratory alliances in
the collaborative portfolio enhances the firm’s incentives to compete with its partner when their
partnership is transaction oriented. However, as the proportion of exploratory alliances increases,
overlap in the firms’ long-term stakes also rises, enlarging their mutual dependence in
developing critical innovations (Pfeffer and Salancik, 1978). The risk of retaliation from an
attacked partner also increases, as the partners now hold more critical information about each
other. The partners’ interdependence, on the one hand, and the risk of retaliation, on the other,
may escalate to a point at which the expected costs exceed the benefits of launching a
competitive action, suffocating the firms’ incentives to further intrude into each other’s product
market domains.
Page 8 of 40
9
Specifically, as firms engage more in exploratory collaboration, they become more reliant
on one another for developing exploratory innovations. Developing such innovations involves
great uncertainty, and thus demands more cooperative efforts (March, 1991). Competition
accordingly becomes a growing threat to firms’ common stakes, curbing their incentive to launch
attacks. When exploratory alliances dominate the collaboration between two firms, their shared
stakes in developing exploratory innovations rise to the point where their partnership becomes
strategically important to both parties’ long-term prosperity. A symbiotic relationship featuring a
long-term perspective and ‘relation-oriented’ interactions between them is likely to emerge
(Davis and Eisenhardt, 2011; McEvily and Marcus, 2005). The damage caused by competition
on greater long-term benefits may reach a threshold at which escalating competitive attacks
would be too costly to justify. As the marginal cost of competition continues to increase, the
likelihood that a firm will compete with its partner is reduced.
Moreover, heavy engagement in exploratory collaboration allows both partners to hold a
great deal of high-value information about each other’s operations (Li et al., 2008). Information
about, for example, technological pitfalls and managerial incompetence can be used to design
targeted retaliatory attacks (Chen and Hambrick, 1995). The stakes of ‘betrayal’ become
substantively higher as the strategic importance of the exploratory partnership increases, so that
the tit-for-tat retaliation risk escalates substantially. The consequences of such retaliatory attacks
can be tremendously destructive, resulting in significant costs to the attacker (Chen and
Hambrick, 1995). An expectation of such retaliation also decreases incentives to undertake
competitive actions.
In sum, firms are likely to develop a transaction-oriented partnership when the
collaborative portfolio between them is dominated by exploitative alliances (i.e., low relative
Page 9 of 40
10
exploration), and increases in the proportion of exploratory alliances enhance a firm’s incentives
to compete with its partner. However, as the proportion of exploratory alliances increases, the
two firms’ long-term stakes become more aligned, so that the damages caused by competition
increase. When the collaborative portfolio is dominated by exploratory alliances (i.e., high
relative exploration), the partnership becomes strategically important and a relation-oriented
interaction pattern is fostered. The escalating damage to long-term benefits and the risk of
retaliatory attacks from the partner may reach a level at which the cost of competition outweighs
its short-term benefit, and at this level the motivation to compete starts to decline, resulting in an
inverted U-shaped relationship between relative exploration and the firm’s competition with its
partner; the competition peaks at the medium level of relative exploration (Grant and Schwartz,
2011; Hanns, Pieters, and He, 2016).
Hypothesis 1: There is a curvilinear relationship (taking an inverted U-shape) between
relative exploration and the aggressiveness of a firm’s competition against its partner in the
product market.
The moderating role of network embeddedness
We further contend that this curvilinear relationship is subject to the influence of the inter-firm
alliance networks within which the firm and its partner are embedded. Alliance learning does not
take place in a vacuum but is shaped by the broader social context, as firms are not atomistic
players but relational entities (Granovetter, 1985; Uzzi, 1996). Polanyi (1957) was the first to use
the concept of embeddedness in describing the social structure of modern markets, and many
studies have further conceptualized inter-firm embeddedness and examined how it affects
economic actions (e.g., Lavie, 2007; Polidoro et al., 2011; Uzzi, 1997).
We build a comprehensive model examining the impact of alliance network
Page 10 of 40
11
embeddedness on the relationship between relative exploration and product competition between
allies by focusing on all three dimensions of embeddedness: relational, positional, and structural
(Gulati and Gargulio, 1999). Relational embeddedness, reflected in repeated alliance ties
between two firms, emphasizes cohesive and reliable relationships (Gulati and Gargiulo, 1999).
Positional embeddedness, reflected in network centrality, manifests a firm’s status and power
within the overall network. Structural embeddedness, reflected in the common ties shared by the
firm and its partner, highlights the information sharing, social monitoring, and reputational
effects that regulate inter-firm interactions (Gulati and Gargiulo, 1999; Polidoro et al., 2011).
Relational embeddedness. We argue that repeated collaboration attenuates the inverted
U-shaped relationship between relative exploration and the firm’s competition against its partner
by (a) lowering the firm’s incentive to act opportunistically when the partnership is more
transaction oriented and (b) increasing the cost of aggressive intrusions when the partnership is
more relation oriented.
Specifically, at low to medium levels of relative exploration (i.e., more transaction
oriented), the effect of relative exploration on a firm’s competition is smaller when the number
of repeated collaborations increases, for two reasons. First, repeated ties foster relational capital
between allies (Gulati, 1995; Gulati and Singh, 1998; Kale et al., 2000), stifling a firm’s
incentive to compete with its partner. Relational capital, such as trust, aligns the interests of allies
as well as promotes cooperation and reciprocity (Kogut, 1989; Uzzi, 1997; Zaheer et al., 1998).
Trust encourages firms not to abuse each other’s vulnerabilities and to respect their respective
boundary of resources and proprietary knowledge (Krishnan et al., 2006). The incentive to
behave opportunistically in joint exploration is therefore reduced as the number of repeated ties
increases.
Page 11 of 40
12
Second, repeated collaborations restrain a firm’s incentive to compete by improving the
effectiveness of formal governance mechanisms, such as contracts, in preventing opportunistic
learning in exploratory activities. Repeated ties help firms learn how to work together effectively
by helping them discover their partners’ self-serving behaviors and patterns, which in turn leads
them to anticipate similar behavior in the future (Doz, 1996; Lioukas and Reuer, 2015). Firms
normally incorporate such knowledge into improving their contracts and other defensive
mechanisms to guard against competitive behaviors from the same partners (Mayer and Argyres,
2004). Repeated collaborations can therefore strengthen these governance mechanisms and
reduce a firm’s incentives to compete with its partner.
At medium to high levels of relative exploration (i.e., more relation oriented), too, the
effect of relative exploration on competition is reduced by the number of collaborations between
allies, as repeated collaboration increases the cost of competition in this type of partnership. First,
repeated collaborations further enhance the symbiotic relationship between partners in joint
knowledge discovery. Repeated collaboration facilitates the development and refinement of
important bilateral systems, such as problem-solving mechanisms, that typically consist of
routines of negotiation and mutual adjustments. These mechanisms enable allies to reduce
production errors and resolve problems more flexibly (Uzzi, 1997). Repeated collaboration also
fosters relational capital, which encourages allies to share proprietary knowledge in a more fine-
grained and timely fashion (Krishnan et al., 2006; Uzzi, 1997). These formal or informal
mechanisms further enhance the partnership’s synergistic benefits, thus heightening the cost of
damaging such a collaboration.
Second, repeated collaborations between firms can help both parties gain more
knowledge and intelligence about each other (Gulati, 1995; Podolny, 1994), further escalating
Page 12 of 40
13
the risk of retaliation. Specifically, multiple collaborations help the partner (a) see through
superficial phenomena and relationships to identify the firm’s underlying behavior patterns, (b)
connect separate units of know-how to detect the firm’s fundamental knowledge bases, and (c)
deepen understanding of the firm’s competencies and vulnerabilities (Uzzi, 1997). Such
knowledge further increases the risks of destructive retaliation, expanding the cost of aggressive
competition.
To summarize, a firm’s incentive to compete with its partner is reduced at low to medium
levels of relative exploration, while its competition costs are enlarged at medium to high levels
of relative exploration, when the number of repeated ties is higher. The effect of relative
exploration on a firm’s competition with its partner is thus attenuated at both sides of the
inverted U-shape. The overall slope of the relationship between relative exploration and a firm’s
competition with its partner is therefore likely to be flattened, with the peak of the slope
becoming lower.
Hypothesis 2: Repeated alliance ties between a firm and its partner negatively moderate
the relationship between relative exploration and the firm’s competition against this partner in
the product market, such that the inverted U-shape is flattened when the number of repeated
alliance ties is higher.
Positional embeddedness. A firm’s centrality within the inter-firm network affects the
amount of information and resources that it can access (Godart, Shipilov, and Claes, 2014),
which represents a significant source of bargaining power (Brass, 1992). Firms’ relative
centrality therefore reflects the (im)balance between them in terms of resource flows and power
(Gnyawali and Madhavan, 2001; Yang et al., 2011). We argue that a firm’s greater centrality
relative to its partner intensifies the inverse U-shaped relationship between relative exploration
Page 13 of 40
14
and the firm’s competition against its partner. It further enhances the firm’s incentive to take
advantage of its less central partner when their relationship is more transaction oriented, and also
reduces the costs of aggressive intrusions when their partnership is more relation oriented.
At low to medium levels of relative exploration, relatively greater centrality enhances the
firm’s incentive to compete with its partner, for two reasons. First, the more central firm has the
advantage in terms of information gathering and monitoring, and thus can cast a wider net to
capture knowledge leakage from its partner, uncover the partner’s real intentions, and identify
weaknesses in its innovation system (Polidoro et al., 2011; Zahra and George, 2002). Second,
greater network centrality enhances the firm’s advantage in adapting acquired knowledge to
develop substitute products. Situated at the confluence of information flows in the network, a
more central firm occupies an advantageous position in terms of identifying opportunities to
apply specific knowledge gained from a partnership into broader areas of application. In a more
transaction-oriented relationship, these advantages further enhance the firm’s incentive to take
action against its partner (Gnyawali and Madhavan, 2001).
At medium to high levels of relative exploration, greater relative centrality reduces a
firm’s competition costs, for two reasons. First, when the difference in network centrality
between the firm and its partner is large, the firm has more choices of alternative partners than its
peripheral partner (Lavie, 2007), while the peripheral partner is more likely to maintain its
collaboration with the firm for resource access and institutional endorsement (Ahuja, 2000;
Podolny, 1994). The larger the difference in their centrality, the less dependent the firm is on this
specific partnership (Brass, 1992). Such asymmetry enhances the firm’s bargaining power in
their interactions (Shipilov, 2009), which in turn reduces its cost of competition.
Second, when the firm’s centrality is greater than its partner’s, the risk of retaliation from
Page 14 of 40
15
the peripheral partner is lower. This partner is less likely to respond to aggressive actions of the
firm, because (a) due to its lack of information sources, the partner will “find it difficult to
interpret the causes and consequences of competitive actions correctly” (Gnyawali and
Madhavan, 2001: 436), and therefore may not be sufficiently informed to respond; (b) fear of
provoking further actions from the more powerful firm may motivate the partner not to respond
(Gnyawali and Madhavan, 2001); and (c) the partner may not be capable of mobilizing sufficient
assets to orchestrate a response.
To summarize, when a firm’s alliance network centrality is greater than that of its partner,
its incentive to compete with that partner is enhanced at low to medium levels of relative
exploration, while its costs of competition are reduced at medium to high levels of relative
exploration. At both sides of the inverted U-shape, that is, the effect of relative exploration on
the firm’s competition is enlarged by the firm’s relative centrality. The slope in the relationship
between relative exploration and competition is therefore likely to be steeper, and the peak of the
slope is higher.
Hypothesis 3: A firm’s relative centrality vis-à-vis its partner positively moderates the
relationship between relative exploration and the firm’s competition against this partner in the
product market, such that the inverted U-shape is steepened when the firm’s relative centrality is
higher.
Structural embeddedness. The term structural embeddedness refers to the extent to
which two firms are connected by common third parties: the more common ties the partners have,
the more structurally embedded they are. We argue that structural embeddedness can attenuate
the inverted U-shaped relationship between relative exploration and a firm’s competition against
its partner in the product market by (a) dampening the firm’s incentive to compete when the
Page 15 of 40
16
partnership is more transaction oriented, and (b) heightening the competition costs when the
partnership is more relation oriented.
At low to medium levels of relative exploration, the firm’s incentive to take advantage of
its partner is reduced when their structural embeddedness is higher. Specifically, a larger number
of common ties increases resource symmetry between the firm and its partner. Chen (1996)
argues that a firm is more motivated to compete with others when it has superior resources. As
the number of common ties increases, the firm and its partner become increasingly similar in
resources to which they both have access, which restrains the firm’s incentives to compete with
its partner. In addition, if one firm “exploits the vulnerabilities of the other, the occurrence of
such behavior can be revealed to common partners and, through them, reach a larger number of
firms in the network” (Polidoro et al., 2011: 206). The resulting adverse effects on the firm’s
reputation also reduce its incentive to compete.
At medium to high levels of relative exploration, costs of competition are further
heightened when structural embeddedness between partnering firms is higher, for two reasons.
First, the partnership becomes more symbiotic as structural embeddedness between a firm and its
partner increases. Both firms’ success depends increasingly on resource exchanges not only
within the dyad but with their common allies, which fosters an ecosystem in which the two firms
are more interconnected and their interests more aligned. At a given level of relative exploration,
the cost of competition is significantly increased, since any competition between firms may have
a multiplicative adverse effect within the system. Second, common partners increase the risk and
intensity of revenge for opportunistic behaviors. If the firm acts against its partner within this
symbiotic relationship, it is not only likely to be attacked in retaliation by the initial partner but
may also experience escalated attacks in revenge from third-party common allies, which also
Page 16 of 40
17
have (in)direct stakes in this partnership (Polidoro et al., 2011).
Taken together, as structural embeddedness between a firm and its partner increases, the
firm’s incentive to compete is reduced at low to medium levels of relative exploration, while its
competition costs are increased at medium to high levels of relative exploration. Thus, at both
sides of the inverted U-shape, the effects of relative exploration on the firm’s competition are
reduced as their structural embeddedness increases. The slope in the relationship between
relative exploration and the firm’s competition with its partner is likely to be less steep when the
number of common ties is higher, and the peak of the slope is likely to be lower.
Hypothesis 4: Structural embeddedness between a firm and its partner negatively
moderates the relationship between relative exploration and the firm’s competition against this
partner in the product market, such that the inverted U-shape is flattened when the level of
structural embeddedness is higher.
Methods
Sample
We chose the U.S. pharmaceutical industry as an appropriate setting for examining our
hypotheses because it features both extensive alliance activities and competition for new
products (Lichtenberg and Philipson, 2002; Mowery et al., 1996). We first collected data on
alliances within this industry from 1984 to 2003, using three data sources—SDC Platinum,
MedTrack, and ReCap—which (a) have very similar standards for reporting information,
including the year of alliance establishment, partners’ names, and descriptions of alliance deals,
and (b) each normally reports only a fraction of all alliance activities (Schilling, 2009). Although
databases that track alliances in the pharmaceutical industry are normally reliable (Schilling,
2009), we found that the SDC database covers more historical data while MedTrack and ReCap
Page 17 of 40
18
include more recent information. We constructed a more comprehensive alliance database by
combining these three data sources. For due diligence, we followed Lavie and Rosenkopf (2006),
searching for alliance announcements and status reports from LexisNexis News, corporate
websites, and Securities and Exchange Commission (SEC) filings accessed via the Edgar
database. Most alliance announcements were cross-validated by at least two additional sources.
By relying on multiple sources, we minimized the possibility of double-counting alliances and of
counting alliances that were announced but not realized.
To measure product competition, we began by selecting all FDA-approved drugs from
the FDA’s National Drug Code Directory. We specifically identified drugs approved under
clauses 505(b)(2) and 505(j) of the Federal Food, Drug, and Cosmetic Act (FD&C Act) with the
aid of the FDA’s Approved Drug Products with Therapeutic Equivalence Evaluations (the
‘Orange Book’). Clause 505(j) governs generic drug applications, while clause 505(b)(2)
governs the modifications of previously approved brand-name drugs. Under the 505(b)(2) and
505(j) pathways of drug application, the timeline for most new drug launches is reduced to
approximately one year.
3
We then collected information on each drug’s ingredients, mechanisms
of action, physiologic effect, and chemical structure from four pharmacological databases:
3
Drug application is regulated by the FD&C Act, s. 505 of which describes three types of new drug
application pathways: (1) Full applications under clause 505(b)(1), which requires full reports of
investigations into the safety and effectiveness of a drug conducted by the applicant (what is usually
known as the brand-name drug application); under this pathway the new drug development cycle is
approximately 10–15 years. (2) Generic drug applications under clause 505(j), which requires information
to show that the proposed product is therapeutically identical or equivalent to a previously approved drug.
(3) Applications under clause 505(b)(2), which allows pharmaceutical firms to submit new drug
applications by referencing the literature or previous agency findings on the safety and efficacy of a drug,
even if not developed by the applicant; typical 505(b)(2) applications include modifications of a
previously approved brand-name drug such as a new formulation, a change in dosing regimen, a new
active ingredient, or a new combination of ingredients. The objective of the 505(b)(2) drug application
pathway is to encourage innovation and speed up new drug introductions by eliminating the need for
duplicative studies to demonstrate what is already known about a drug.
Page 18 of 40
19
Drugs @ FDA, Mosby’s Drug Consult, Drugs-by-Condition on Drugs.com, and AHFS Drug
Information from the American Society of Health-System Pharmacists (Toh and Polidoro, 2013).
A researcher in the pharmacology discipline who was not part of our team used this information
to help classify all listed drugs into competing groups. For example, Zoloft and Celexa are
considered direct competitors in the antidepressant market, since they use the same mechanism
of action: both are selective serotonin reuptake inhibitors (SSRIs), which boost mood by
blocking the re-absorption of the neurotransmitter serotonin within the brain, thus helping brain
cells send and receive chemical messages. Both Pristiq and Cymbalta belong to another group of
competing antidepressants using serotonin–norepinephrine reuptake inhibitors (SNRIs), which
work by inhibiting the reuptake of both serotonin and norepinephrine to boost mood (See Table
A in the online appendix for examples of competing antidepressants and their underlying
mechanisms). We then linked the competing drugs to their producers and market introduction
dates with the aid of the ‘Orange Book.’
We also collected firm financial information from Compustat, as well as patent data from
the U.S. Patent and Trademark Office (USPTO) and the Thomson Innovation database. We
located 126 public firms for which we had complete information on alliances, drugs, patents, and
financial reports, and identified 3,752 observations as our sample, with an observation window
ranging from 1984 to 2003.
Dependent variable
Competitive aggressiveness. Following prior studies, we measured the extent of competition in
product markets that a firm undertakes against its partner using competitive aggressiveness,
describing both the intensity and diversity of the firm’s competitive actions (Yu et al., 2009).
This variable contains three items: the number of competitive actions (i.e., market entries) that
Page 19 of 40
20
the firm launched against its partner following the alliance announcement and the width and
depth of its competing actions. The width dimension measures the number of different
therapeutic areas in which the firm entered that compete against its partner; the depth dimension
measures the number of destructive competitions (i.e., modified brand-name drugs) launched by
the firm against its partner. The launch of modified brand-name drugs is a more aggressive
competitive action than the introduction of generic drugs because of the former’s much longer
period of market exclusivity (Lichtenberg and Philipson, 2002).
4
We ran a factor analysis of
these three items and found that all three loaded high (> 0.73) on one latent factor, while the
value of Cronbach’s alpha is 0.81, which suggests that it is a reliable construct. We used the
average score for these three items to measure the dependent variable.
Independent variable and moderators
Relative exploration is calculated as the ratio of the number of exploratory collaborations to the
total number of collaborations between the firm and its partner over a five-year moving window
(Ang, 2008; Yang et al., 2011). We used a five-year window because alliance databases seldom
report the termination dates of alliances, and the life span of an alliance is normally no more than
five years (Yang et al., 2011). We constructed this variable using a content analysis of the deal
summary to identify the nature of alliance learning for each alliance in our sample. Specifically,
research and development (R&D) alliances with the objective of exploring and creating critical
new knowledge were coded as exploratory collaborations, while alliances such as co-marketing,
manufacturing, and licensing, which mainly utilize existing knowledge and resources, were
4
While a generic drug can be detrimental to the market share of a brand-name drug, Lichtenberg and
Philipson (2002) have shown that creative destructions such as new forms or combinations of ingredients
can be more detrimental to market share. This is partly because the latter may have better therapeutic
effects and are granted longer market exclusivity: while generic drugs are given a maximum 180 days of
exclusivity, modifications to brand-name drugs are granted three to seven years of exclusivity.
Page 20 of 40
21
classified as exploitative collaborations (Lavie and Rosenkopf, 2006; Rothaermel, 2001;
Rothaermel and Deeds, 2004). Since partners may have different intentions within an alliance,
we undertook this coding from the focal firm’s perspective (Lavie and Rosenkopf, 2006). It is
important to note that not all R&D collaborations are exploration oriented; within our data set, if
two firms formed an R&D alliance for the purpose of making incremental changes to an existing
technology, that R&D alliance was coded as exploitative. The value of this variable ranges from
0 to 1.
Repeated alliance ties reflect the relational embeddedness between two firms. We
measured this variable by counting the number of repeated alliance ties between the firm and its
partner over the previous five years (Ahuja, Polidoro, and Mitchell, 2009; Gulati and Gargiulo,
1999).
Relative centrality captures the relative positional embeddedness of two firms. We first
constructed yearly alliance matrices within the pharmaceutical industry from 1984 to 2003, using
a five-year moving window. Altogether, we identified 47,862 pairs of alliances, including those
started between 1980 and 1983. We constructed a symmetric (non-directional) matrix of these
firms for each year using Ucinet 6 (Borgatti, Everett, and Freeman, 2002), then calculated the
degree centrality of each firm within the above alliance network matrices. Relative centrality is
calculated as the firm’s degree centrality minus its partner’s degree centrality.
Common ties capture the structural embeddedness between the firm and its partner.
Following prior studies, we measured this variable by counting the number of common partners
that the firm and its partner shared during the previous five years (Gulati and Gargiulo, 1999;
Polidoro et al., 2011).
Control variables
Page 21 of 40
22
We controlled for sample heterogeneity by including nine variables suggested by prior studies as
influencing the relationship between alliances and competition.
R&D intensity. A firm’s R&D investment can contribute to its new product development
and thus influence its competition against its partner in the product market. We therefore
controlled for the firm’s R&D investment, measured as its R&D expenditures scaled by sales
(Keil, Maula, Schildt, and Zahra, 2008).
Return on assets (ROA). A firm’s financial performance is positively associated with new
product development (Rothaermel, 2001); firms that perform better financially may be more
prepared to compete in the product market. We therefore controlled for a firm’s financial
performance using its return on assets.
Financial slack. Since researchers have argued that competitive behaviors are dependent
on resource availability (Smith et al., 1991), we controlled for a firm’s financial slack, computed
as current assets divided by current liabilities (Hambrick, Cho, and Chen, 1996).
Coopetition. If two firms are competing with each other in the market when they
establish an alliance, this ‘coopetition’ relationship may contribute to the risk of their future
competition in the market. We therefore controlled for coopetition (coded as 1 if two firms were
competing in the product market when they established an alliance and as 0 otherwise).
Technological overlap. Prior research suggests that resource similarity between firms
may influence their competition (Chen, 1996). We accordingly controlled for dyadic resource
similarity by measuring technological overlap, calculated as the percentage of patent cross-
citations between a firm and its partner (Mowery et al., 1996).
Alliance scope. A wide scope of collaboration increases the likelihood of both
information leakage and appropriation within alliances (Oxley and Sampson, 2004), heightening
Page 22 of 40
23
the risk of product competition. However, collaborations across a wide range of areas may also
increase partners’ mutual dependence, which reduces opportunistic behavior (Pfeffer and
Salancik, 1978). Prior studies have not provided a clear direction for the effect of alliance scope
on competition between partners in the product market. In our model, we controlled for alliance
scope using a measure adapted from prior studies (Oxley and Sampson, 2004), which was coded
as 0 if a collaboration does not include R&D activities, as 1 if it includes only R&D activities,
and as 2 if it includes both R&D and non-R&D activities.
Multiparty alliance (‘Multiparty’). In cases where an alliance includes more than two
firms, we split the alliance into a set of dyads. Empirically, this may result in a confounding
effect with structural embeddedness (i.e., number of common ties). We parceled out the possible
confounding effect caused by multiparty alliances by including a control variable (‘Multiparty’)
that measures the number of additional parties (> 2) to a dyad within an alliance. We also ran the
models using a sample excluding multiparty alliances; the results are robust.
Network density. As Coleman (1988) has noted, a dense alliance network in which a dyad
is located can also facilitate information sharing and monitoring within the network, suffocating
competition between allies. We therefore controlled for the density of the ego network for every
dyad. This ego network includes the two focal firms within the alliance as well as all other firms
that have at least one alliance with at least one of the two focal firms. We first transformed the
firm-by-firm alliance network into an incidence network (i.e., the output is a two-mode firm-by-
alliance network) using the command ‘transform | graph theoretic | incidence’ in Ucinet 6. We
then transformed the two-mode incidence network into a one-mode alliance-by-alliance square
network by using the command ‘data | affiliations’ and selecting column as the unit of interest.
Finally, we calculated ego network density within this new network for each alliance in our
Page 23 of 40
24
sample.
Year dummies. We controlled for unobserved heterogeneity associated with years by
including 19 year dummies in our models.
Analysis
We tested our hypotheses using fixed-effects models. The unit of analysis is firm–partner–year,
and we allowed a one-year lag between our predictor variables and the dependent variable. A
fixed-effects estimator has superior controls for time-invariant variables (Mundlak, 1978) and is
an effective way to account for possible endogeneity problems. For example, if unobserved
heterogeneities, such as the attractiveness of partners to one another and their tendencies to
compete with each other, are constant within firm–partner dyads, then there might be an
endogeneity concern. A fixed-effects estimator can rule out such a possibility by eliminating
time-invariant heterogeneities. Fixed-effects models also allow us to account for intra-cluster
correlations caused by multiple observations of the same dyad over time. We therefore employed
dyad fixed-effects and clustered standard errors on dyads in our models.
Results
Table 1 reports descriptive statistics and correlations for all variables, including the quadratic and
interaction terms. We mean-centered the variables before creating quadratic and interaction
terms in order to reduce non-essential ill-conditioning between independent variables and their
higher-order terms (Aiken and West, 1991). The dependent and independent variables show
considerable variance, and the correlation coefficients are consistent with our expectations.
We ran fixed-effects models following a hierarchical approach: Model 1 includes only
the control variables, while Models 2 through 5 add the independent and interaction variables.
Model 6 is the full model, including all independent and interaction variables. Variance inflation
Page 24 of 40
25
factor (VIF) scores were calculated for all models; none of the maximum VIFs exceed the value
of 2.5, which is substantially lower than the rule-of-thumb cut-off of 10 (Ryan, 1997). We then
used the ‘coldiag’ procedure in Stata to conduct the Belsley, Kuh, and Welsch (1980)
multicollinearity diagnostic test, which showed that the condition number for our complete
model is 7.53, well below the threshold of 30. We also ran the fixed-effects models using non-
centered data; the results are consistent. Since centered estimations can make interpretation of
the results less straightforward (Echambadi and Hess, 2007), we report estimations using the
original variable values in Table 2.
[Insert Tables 1 and 2 about here]
In Model 2 of Table 2, we tested Hypothesis 1 by introducing both the linear and
quadratic terms of relative exploration. The result shows that competitive aggressiveness first
increases significantly with relative exploration (b = 1.508, p = 8E-08), then decreases
significantly as relative exploration continues to increase (b = - 1.439, p = 1E-07). This result
indicates a curvilinear relationship (inverted U-shape) between relative exploration and
competitive aggressiveness, with a medium effect size (Cohen’s d = 0.428). We examined the
marginal effects of this relationship following the three steps suggested by Lind and Mehlum
(2010). First, we examined whether or not the second-order term is significant and of the
expected sign; this is confirmed by the result. Second, we tested whether the slope is indeed
sufficiently steep at both ends of the data range of relative exploration. Using the ‘margins’
command in Stata 12, we confirmed that when relative exploration = 0, the slope dy/dx = 1.701
(p = 4E-05), and when relative exploration = 1, the slope dy/dx = - 1.561 (p = 3E-05). Third, we
tested whether or not the turning point is located within the data range of relative exploration.
We confirmed this using the ‘nlcom’ command in Stata 12 by showing that the inverted U-shape
Page 25 of 40
26
turns when relative exploration = 0.522 and that the 95% confidence interval for the turning
point [0.504, 0.539] is within the value range of relative exploration. We provide additional
support by plotting this relationship in Figure 1. These findings suggest that Hypothesis 1 is
supported.
In Model 3, the interaction terms between repeated alliance ties (relational
embeddedness) and both the linear and quadratic terms of relative exploration are introduced in
order to test Hypothesis 2: whether repeated alliance ties negatively moderates the inverted U-
shaped relationship. This moderation effect is supported if the second-order interaction term is
significantly positive (Hanns et al., 2016). As confirmed by our results, the second-order
interaction term is indeed positive (b = 0.692, p = 6E-06), with a small-to-medium effect size
(Cohen’s d = 0.364). Figure 2 illustrates this moderation effect, showing that the inverted U-
shape is flattened when the value of repeated alliance ties is higher, supporting Hypothesis 2.
Model 4 introduces the interaction terms between relative centrality (positional
embeddedness) and both the linear and quadratic terms of relative exploration in order to test
Hypothesis 3: whether the firm’s relative centrality positively moderates the inverted U-shaped
relationship. This moderation effect is supported if the second-order interaction term is
significantly negative (Hanns et al., 2016). We found that indeed the second-order interaction
term is negative (b = - 0.054, p = 0.019), with a small-to-medium effect size (Cohen’s d = 0.226).
Figure 3 illustrates this moderation effect, indicating that the inverted U-shape is steepened when
the firm’s relative centrality is higher, supporting Hypothesis 3.
Model 5 introduces the interaction terms between common ties (structural embeddedness)
and both the linear and quadratic terms of relative exploration in order to test Hypothesis 4:
whether common ties negatively moderates the inverted U-shape. As indicated in our results, the
Page 26 of 40
27
second-order interaction term is positive (b = 0.024, p = 0.009), with a small-to-medium effect
size (Cohen’s d = 0.318). Figure 4 illustrates this moderation effect, showing that the inverted U-
shape is flattened when the value of common ties is higher, supporting Hypothesis 4. Model 6 is
the full model, including all control, independent, and interaction variables; all results from
Models 2–5 hold.
[Insert Figures 1, 2, 3, and 4 about here]
Post-hoc analyses
We conducted six additional analyses, either as robustness checks or to gain additional insights
into the primary relationships. These analyses investigated (1) whether the results are robust to
alternative measures for relative exploration; (2) which firm in a dyad is more likely to initiate
competitive actions; (3) what factors determine a firm’s response to its partner’s actions; (4) the
extent to which the technological know-how acquired in one area and knowledge of a partner’s
managerial system can be applied to competition against the same partner in different
technological areas; (5) whether the results are robust to different paradigms of competition; and
(6) the potential moderating effects of network density and multiparty. Details of these analyses
are available in the online appendix.
Discussion
This study contributes to a core research area in strategic alliances: the influence of collaboration
on inter-partner dynamics (Gulati, 1995; Kogut, 1989). While previous studies provide important
insights into how prior collaboration can promote further collaboration between partners (Gulati,
1995; Gulati and Singh, 1998), we add to this line of research by investigating whether
collaboration can lead to intense product market competition between partners. We approach this
question by analyzing the composition of the collaborative relationships between two firms as
Page 27 of 40
28
well as by studying the impact of their interconnectedness in the context of alliance networks.
We find that in addition to the path-dependent paradigm of partners’ interactions whereby prior
collaboration leads to future collaboration (Gulati, 1995), certain compositions of collaborative
relationships between two firms can lead to a new path of interaction—a transition from
collaboration to competition in product markets. Furthermore, the firms’ network embeddedness
can modify this transition process.
Our research contributes to this line of investigation on three fronts. First, our research
extends the literature on the tension between collaboration and competition within alliances.
Prior studies have long argued that firms race to appropriate knowledge from each other within
alliances (Hamel, 1991; Khanna et al., 1998). We extend this literature by identifying (a) the
composition of collaborative portfolios and (b) network embeddedness between allies as
important mechanisms through which learning within alliances is translated into competitive
interactions in the product market. Our study thus bridges two streams of research: learning races
within alliances and competitive dynamics in the product market.
Second, we extend prior studies on learning races, which tend to focus on competition
within research-based alliances. We examine the roles played by all types of alliance between
two firms, in terms of their combinations, in determining a firm’s competition with its partner.
Our findings suggest that a firm’s product market competition with its partner is not necessarily
high in research-intensive collaborations; in fact, the firm is more likely to adopt a long-term
outlook that encourages relationship building and cooperation when the collaboration portfolio is
dominated by exploratory alliances, reducing its motivation to compete with its partner. Our
finding of a curvilinear relationship between relative exploration and competitive aggressiveness
highlights the role that the overall collaborative portfolio between partners plays in determining
Page 28 of 40
29
their competitive interactions.
Third, this research contributes to the literature by studying the moderating role of allies’
network embeddedness in the ‘collaboration–competition’ relationship. We demonstrate that
network embeddedness—namely relational, positional, and structural embeddedness—affects the
relationship between relative exploration and product market competition. Our research speaks
directly to the importance of looking beyond dyadic interactions when examining dyadic
competition, representing a valuable addition to the growing body of studies exploring how allies’
network characteristics influence their interaction patterns and performance (Lavie, 2007;
Polidoro et al., 2011).
Our post hoc analyses (see online appendix) make further contributions to the research on
collaboration and competition by unpacking the competitive dynamics (i.e., actions and
responses) between allies. For example, post hoc analysis 2 found an asymmetry between allies
with respect to who will be more likely to ‘defect’ from collaboration: in addition to firms with
relatively great network centrality, those with more experience competing with other allies are
also more likely to compete with their current partners. This suggests that the likelihood that
allies will compete against each other is not evenly distributed within a dyad: the firm with either
more advantageous network positioning or richer experience in managing the transition from
collaboration to competition is more likely to do so.
Post hoc analysis 3 revealed an interesting relationship with respect to responses to
competitive actions: more aggressive actions tend to provoke more acts of retaliation. This
finding seems to be different from suggestions by traditional studies on competitive dynamics
that aggressive competition (e.g., irreversible actions) tends to suffocate reactions (Chen and
MacMillan, 1992). This is likely because those studies assume that competitors have no
Page 29 of 40
30
knowledge of each other ex ante, and thus compete as if they were strangers each time they
encounter each other (Ketchen et al., 2004). As recent research indicates, traditional competitive
dynamics studies assume a Markov or memory-less process of competition, in which only the
current state matters in predicting subsequent states (Kilduff, Elfenbein, and Staw, 2010). Prior
studies on competitive dynamics have overlooked the role of inter-firm learning (Nelson and
Winter, 1982) in determining firms’ competitive interactions. Our finding suggests that when
firms are alliance partners, the attacked firm is likely to respond to the aggressor’s actions,
having developed the capacity to do so through alliance learning. This observation is echoed by
two additional findings: first, that the attacked firm is more likely to respond if the number of
repeated ties with the aggressor is high, because these ties can provide the attacked firm with
ample knowledge regarding the aggressor, and therefore enable it to take effective retaliatory
action; and, second, that the attacked firm is less likely to respond if it is either less centrally
located within the alliance network or less experienced in managing the transition from
collaboration to competition than the aggressor—situations in which the attacked firm is less
informed and/or less capable of retaliating. Our findings thus complement the classic literature
on competitive dynamics.
Our findings provide strong support for the recent conceptualization of competition as
‘relational’ (Chen and Miller, 2015). Different from the traditional ‘rivalrous view’ of
competitive dynamics, which highlights a firm-centric perspective (Chen and Miller, 2015), the
‘relational’ view emphasizes the necessity of understanding others’ needs and preferences, as
well the interdependencies between self and others, before deciding on competitive moves. Our
research extends this line of argument by suggesting that firms can increase their ‘relational’
savvy through learning in various forms of interactions, including alliances (Lioukas and Reuer,
Page 30 of 40
31
2015), competitions (Tsai et al., 2011), and the transition between the two; all of these
relationships may inform subsequent competitive encounters.
Limitations and directions for future research
The findings of our study should be interpreted in light of its limitations, which suggest
opportunities for future research. First, we examined our hypotheses only in the context of
horizontal alliances in the U.S. pharmaceutical industry. Although the findings are strongly
supportive of our hypotheses, we must be cautious about generalizing them to other collaboration
types (e.g., vertical alliances) and other industries. Researchers may further advance this
emerging research paradigm by considering the roles of alliance- and industry-specific
conditions in the relationship between collaboration and competition.
Second, while the alliance learning perspective has provided important insights for the
development of our theoretical framework, we are also aware that alliances are complex
partnerships. This learning approach can be well complemented by other relevant theories, such
as transaction-cost economics and the resource-based view. For example, researchers might
explore how firms structure their alliances by taking governance and resources into consideration
in order to prevent competition from allies.
Third, while we studied an important dimension of inter-firm competition (i.e.,
competitive aggressiveness) that has been extensively examined in the competitive dynamics
literature, there are other dimensions of competition that are currently not included in our
research. For example, we did not study competitive actions using such variables as action
execution speed, action visibility, and competitor’s acumen (Chen and Hambrick, 1995; Tsai et
al., 2011). There are also ample opportunities to study how alliance learning may affect different
paradigms of product competition, such as the ‘racing’ and ‘incumbent’ types of competition
Page 31 of 40
32
(see online appendix for descriptions of these paradigms). While our preliminary analysis
suggests that the relationship between collaboration and competitive aggressiveness does not
differ between these two scenarios, it may prove fruitful to examine whether and how the
relationship between collaboration and the other dimensions of competition (e.g., speed,
visibility, and acumen) may differ in these two scenarios or in other competitive paradigms.
Studying these characteristics of competition will enrich our understanding of the competitive
consequences of alliances.
Finally, while we compiled a comprehensive longitudinal dataset depicting alliances and
competitions, the use of second-hand data constrained our ability to interpret the micro-level
processes through which information and knowledge are leaked or appropriated through
interactions among individuals, as well as the mechanisms through which decisions regarding
new product introductions are made. For example, we were not able to measure tacit
technological and managerial know-how in our data because of its low codifiability and visibility.
Nor were we able to investigate the extent to which such tacit knowledge travels across
technological and organizational boundaries. Despite our modest attempts to address this issue in
a post hoc analysis, we acknowledge that this is an important limitation of our study. Field
research, such as intensive case studies and interviews, might uncover some of the latent and
important processes that cannot be completely revealed by our present work. Future studies in
this direction will be warranted.
Conclusion
This study investigates an important yet overlooked relationship in the alliance literature: the
impact of alliances on partners’ competition in the product market. Integrating alliance learning
and network perspectives, we argue that the composition of the collaborative relationship
Page 32 of 40
33
between a firm and its partner affects the firm’s competition against this partner. We found an
inverted U-shaped relationship between the firm’s relative exploration and its competitive
aggressiveness against its partner in the market. This curvilinear relationship is flattened by their
relational embeddedness (i.e., repeated ties) and structural embeddedness (i.e., common ties), but
steepened by their positional embeddedness (i.e., relatively great centrality of the firm) within
the alliance network. This study contributes to the literature on inter-firm dynamics between
collaboration and competition, calling for more studies to investigate this intriguing area.
Acknowledgements
We thank our guest editors, anonymous reviewers, and participants in the 2016 SMJ special issue
workshop in Rome for their constructive feedback and comments. This research is supported by
the Social Sciences and Humanities Research Council (SSHRC) of Canada (grant number 435-
2017-1301) and the Research Grants Council of the Hong Kong Special Administrative Region,
China (CityU 11508415).
References
Agrawal A. 2006. Engaging the inventor: exploring licensing strategies for university inventions
and the role of latent knowledge. Strategic Management Journal 27: 63–79.
Ahuja G. 2000. Collaboration networks, structural holes, and innovation: a longitudinal study.
Administrative Science Quarterly 45: 425–455.
Ahuja G, Polidoro Jr F, Mitchell W. 2009. Structural homophily or social asymmetry? The
formation of alliances by poorly embedded firms. Strategic Management Journal 30: 941–
958.
Aiken LS, West SG. 1991. Multiple regression: Testing and interpreting interactions. Newbury
Park, CA: Sage.
Andrevski G, Brass DJ, Ferrier WJ. 2016. Alliance portfolio configurations and competitive
action frequency. Journal of Management 42: 811–837.
Ang SH. 2008. Competitive intensity and collaboration: impact on firm growth across
technological environments. Strategic Entrepreneurship Journal 29: 1057–1075.
Barden JQ, Mitchell W. 2007. Disentangling the influences of leaders’ relational embeddedness
on interorganizational exchange. Academy of Management Journal 50: 1440–1461.
Belsley DA, Kuh E, Welsch RE. 1980. Regression diagnostics: Identifying influential data and
Page 33 of 40
34
sources of collinearity. Wiley, New York.
Borgatti SP, Everett MG, Freeman LC. 2002. Ucinet for Windows: software for Social Network
Analysis. Analytic Technologies: Harvard, MA.
Brass DJ. 1992. Power in organizations: a social network perspective. In Research in Politics
and Society. Moore G., J. A. Whitt (eds.), JAI Press: Greenwich, CT.
Chen EL, Katila R, McDonald R, Eisenhardt KM. 2010. Life in the fast lane: origins of
competitive interaction in new vs. established markets. Strategic Management Journal 31:
1527–1547.
Chen M. 1996. Competitor analysis and interfirm rivalry: toward a theoretical integration.
Academy of Management Review 21: 100–134.
Chen M, Hambrick DC. 1995. Speed, stealth, and selective attack: how small firms differ from
large firms in competitive behavior. Academy of Management Journal 38: 453–482.
Chen MJ, MacMillan IC. 1992. Nonresponse and delayed response to competitive moves: the
roles of competitor dependence and action irreversibility. Academy of Management Journal
35: 539–570.
Chen MJ, Miller, D. 2015. Reconceptualizing competitive dynamics: a multidimensional
framework. Strategic Management Journal 36: 758–775.
Coleman, JS. 1988. Social capital in the creation of human capital. American Journal of
Sociology 4: 95–120.
Davis JP, Eisenhardt, KM. 2011. Rotating leadership and collaborative innovation:
recombination processes in symbiotic relationships. Administrative Science Quarterly 56:
159–201.
Dussauge P, Garrette B, Mitchell W. 2000. Learning from competing partners: outcomes and
durations of scale and link alliances in Europe, North America and Asia. Strategic
Management Journal 21: 99–126.
Echambadi R, Hess JD. 2007. Mean-centering does not alleviate collinearity problems in
moderated multiple regression models. Marketing Science 26: 438–445.
Gimeno J. 2004. Competition within and between networks: the contingent effect of competitive
embeddedness on alliance formation. Academy of Management Journal 47: 820–842.
Gnyawali DR, Madhavan R. 2001. Cooperative networks and competitive dynamics: a structural
embeddedness perspective. Academy of Management Review 26: 431–445.
Godart FC, Shipilov AV, Claes K. 2014. Making the most of the revolving door: the impact of
outward personnel mobility networks on organizational creativity. Organization Science
25(2): 377–400.
Grant, AM, Schwartz B. 2011. Too much of a good thing: the challenge and opportunity of the
inverted U. Perspective on Psychological Science 6(1): 61–76.
Granovetter MS. 1985. Economic action and social structure: the problem of embeddedness.
American Journal of Sociology 91: 1360–1380.
Gulati R. 1995. Does familiarity breed trust? The implications of repeated ties for contractual
choice in alliances. Academy of Management Journal 38: 85–112.
Gulati R. 1998. Alliances and networks. Strategic Management Journal 19: 293–317.
Gulati R, Gargiulo M. 1999. Where do interorganizational networks come from? American
Journal of Sociology 104: 1439–1493.
Gulati R, Singh H. 1998. The architecture of cooperation: managing coordination costs and
appropriation concerns in strategic alliances. Administrative Science Quarterly 43: 781–814.
Hambrick DC, Cho TS, Chen M. 1996. The influence of top management team heterogeneity on
Page 34 of 40
35
firms’ competitive moves. Administrative Science Quarterly 41: 659–684.
Hamel G. 1991. Competition for competence and interpartner learning within international
strategic alliances. Strategic Management Journal 12: 83–103.
Hamel G, Doz YL, Prahalad CK. 1989. Collaborate with your competitors and win. Harvard
Business Review 67: 133–139.
Hanns RFJ, Pieters C, He ZL. 2016. Thinking about U: theorizing and testing U- and inverse U-
shaped relationships in strategy research. Strategic Managemenf Journal 37: 1177–1195.
Hille K, Taylor P. 2011. Relief for Huawei as it settles with Motorola. Financial Times,
http://www.ft.com/cms/s/0/9b767044-65f6-11e0-9d40-00144feab49a.html#axzz1ebSG34pJ.
Retrieved on November 24, 2011.
Kale P, Singh H, Perlmutter H. 2000. Learning and protection of proprietary assets in strategic
alliances: building relational capital. Strategic Management Journal 21: 217–237.
Katila R, Rosenberger JD, Eisenhardt K. 2008. Swimming with sharks: technology ventures,
defense mechanisms and corporate relationships. Administrative Science Quarterly 53: 295–
332.
Keil T, Maula M, Schildt H, Zahra SA. 2008. The effect of governance modes and relatedness of
external business development activities on innovative performance. Strategic Management
Journal 29: 895–907.
Ketchen DJ, Snow CC, Hoover VL. 2004. Research on competitive dynamics: recent
accomplishments and future challenges. Journal of Management 30: 779–804.
Khanna T, Gulati R, Nohria N. 1998. The dynamics of learning alliances: competition,
cooperation, and relative scope. Strategic Management Journal 19: 193–210.
Kilduff GJ, Elfenbein HA, Staw BM. 2010. The psychology of rivalry: a relationally dependent
analysis of competition. Academy of Management Journal 53: 943–969.
Kogut B. 1989. The stability of joint ventures: reciprocity and competitive rivalry. Journal of
Industrial Economics 38: 183–198.
Koza M, Lewin A. 1998. The co-evolution of strategic alliances. Organization Science 9: 255–
264.
Krishnan R, Martin X, Noorderhaven NG. 2006. When does trust matter to alliance performance?
Academy of Management Journal 49: 894–917.
Lavie D. 2007. Alliance portfolios and firm performance: a study of value creation and
appropriation in the U.S. software industry. Strategic Management Journal 28: 1187–1212.
Lavie D, Rosenkopf L. 2006. Balancing exploration and exploitation in alliance formation.
Academy of Management Journal 49: 797–818.
Li D, Eden L, Hitt MA, Ireland RD. 2008. Friends, acquaintances, or strangers? Partner selection
in R&D alliances. Academy of Management Journal 51: 315–334.
Li SX, Rowley TJ. 2002. Inertia and evaluation mechanisms in interorganizational partner
selection: syndicate formation among U.S. investment banks. Academy of Management
Journal 45: 1104–1119.
Lichtenberg F, Philipson T. 2002. The dual effects of intellectual property regulations: within-
and between-patent competition in the U.S. pharmaceutical industry. Journal of Law &
Economics 45: 643–672.
Lind JT, Mehlum H. 2010. With or without U? The Appropriate Test for a U-Shaped
Relationship. Oxford bulletin of economics and statistics 72: 109–118.
Lioukas CS, Reuer JJ. 2015. Isolating trust outcomes from exchange relationships: social
exchange and learning benefits of prior ties in alliances. Academy of Management Journal 58:
Page 35 of 40
36
1826–1847.
March JG. 1991. Exploration and exploitation in organizational learning. Organization Science 2:
71–87.
Mayer KJ, Argyres NS. 2004. Learning to contract: evidence from the personal computer
industry. Organization Science 15: 394–410.
McEvily B, Marcus A. 2005. Embedded ties and the acquisition of competitive capabilities.
Strategic Management Journal 26: 1033–1055.
Mowery DC, Oxley JE, Silverman BS. 1996. Strategic alliances and interfirm knowledge
transfer. Strategic Management Journal 17: 77–91.
Mundlak Y. 1978. On the pooling of time series and cross section data. Econometrica 46: 69–85.
Nelson RR, Winter SG. 1982. An Evolutionary Theory of Economic Change. Belknap Press:
Cambridge, MA.
Oxley JE, Sampson RC. 2004. The scope and governance of international R&D alliances.
Strategic Management Journal 25: 723–749.
Pfeffer J, Salancik GR. 1978. The External Control of Organizations. Harper & Row: New York.
Podolny JM. 1994. Market uncertainty and the social character of economic exchange.
Administrative Science Quarterly 39: 458–483.
Polanyi K. 1957. The economy as instituted process. In K Polanyi, CM, Arensberg HW, Pearson
(eds.), Trade and Market in the Early Empires. The Free Press: Mankato, MN.
Polidoro F, Ahuja G, Mitchell W. 2011. When the social structure overshadows competitive
incentives: the effects of network embeddedness on joint venture dissolution. Academy of
Management Journal 54: 369–392.
Ritman A. 2006. Motorola and Huawei collaborate on 3G. ITP.net, http://www.itp.net/487110-
motorola-and-huawei-collaborate-on-3g. Retrieved on June 21, 2017.
Rothaermel FT. 2001. Incumbent’s advantage through exploiting complementary assets via
interfirm cooperation. Strategic Management Journal 22: 687–699.
Rothaermel FT, Deeds DL. 2004. Exploration and exploitation alliances in biotechnology: a
system of new product development. Strategic Management Journal 25: 201–221.
Ryan TY. 1997. Modern regression analysis. New York: Wiley.
Schilling MA. 2009. Understanding the alliance data. Strategic Management Journal 30: 233–
260.
Shipilov AV. 2009. Firm scope experience, historic multimarket contact with partners, centrality,
and the relationship between structural holes and performance. Organization Science 20: 85–
106
Smith KG, Grimm C, Gannon M, Chen M. 1991. Organizational information processing,
Competitive responses and performance in the U.S. domestic airlines industry. Academy of
Management Journal 34: 60–85.
Toh PK, Polidoro F. 2013. A competition-based explanation of collaborative invention within
the firm. Strategic Management Journal 34: 1186–1208.
Tsai WP, Su K, Chen MJ. 2011. Seeing through the eyes of a rival: competitor acumen based on
rival-centric perceptions. Academy of Management Journal 54: 761–778.
Uotila J, Maula M, Keil T, Zahra SA. 2009. Exploration, exploitation, and financial performance:
analysis of S&P 500 corporations. Strategic Management Journal 30: 221–231.
Uzzi B. 1996. The sources and consequences of embeddedness for the economic performance of
organizations: the network effect. American Sociological Review 61: 674–698.
Uzzi B. 1997. Social structure and competition in interfirm networks: the paradox of
Page 36 of 40
37
embeddedness. Administrative Science Quarterly 42: 35–67.
Yang H, Lin Z, Peng MW. 2011. Behind acquisitions of alliance partners: exploratory learning
and network embeddedness. Academy of Management Journal 54: 1069–1080.
Yang H, Zheng Y, Zaheer A. 2015. Asymmetric learning capabilities and stock market returns.
Academy of Management Journal 58: 356–374.
Yu T, Cannella AA. 2007. Rivalry between multinational enterprises: an event history approach.
Academy of Management Journal 50: 665–686.
Yu TY, Subramaniam M, Cannella Jr AA. 2009. Rivalry deterrence in international markets:
contingencies governing the mutual forbearance hypothesis. Academy of Management
Journal 52: 127–147.
Zaheer A, McEvily B, Perrone V. 1998. Does trust matter? Exploring the effects of
interorganizational and interpersonal trust on performance. Organization Science 9: 141–159.
Zahra SA, George G. 2002. Absorptive capacity: a review, reconceptualization, and extension.
Academy of Management Review 27: 185–203.
Page 37 of 40
38
TABLE 1. Descriptive Statistics and Correlations
Variables Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12
1 Competitive aggressiveness 0.141 0.921
2 R&D intensity 0.897 5.874 -0.020
3 ROA 0.027 0.242 0.058 -0.385
4 Financial slack 3.201 3.650 0.081 0.170 -0.169
5 Coopetition 0.058 0.233 0.620 -0.031 0.086 0.001
6 Technological overlap 0.000 0.012 -0.004 -0.003 -0.003 0.017 -0.006
7 Alliance scope 0.914 0.696 -0.013 0.041 -0.108 0.056 -0.011 -0.029
8 Repeated alliance ties 0.343 0.702 -0.008 -0.038 0.067 -0.104 0.034 -0.012 -0.010
9 Relative centrality -0.191 5.365 0.019 -0.133 0.364 -0.309 0.012 -0.028 -0.079 0.080
10 Common ties 14.892 16.240 0.011 -0.084 0.241 -0.244 0.146 -0.018 0.022 0.321 0.067
11 Multiparty 0.039 0.193 0.021 -0.019 -0.001 0.006 0.027 -0.005 0.075 0.049 -0.032 0.03
12 Network density 0.642 0.090 -0.103 0.041 -0.156 0.097 -0.078 0.005 0.028 -0.083 -0.135 -0.224 -0.007
13 Relative exploration 0.320 0.366 0.014 0.021 -0.104 0.052 0.004 -0.020 0.403 0.095 -0.100 0.117 0.080 0.047
14 Relative exploration squared 0.236 0.338 -0.065 0.037 -0.113 0.055 -0.058 -0.006 -0.073 -0.129 -0.074 -0.004 0.007 0.069
15 Relative exploration × Repeated alliance ties 0.134 0.364 -0.014 0.004 0.011 -0.050 0.012 0.009 -0.088 0.195 0.031 0.124 0.013 -0.087
16 Relative exploration squared × Repeated
alliance ties
0.084 0.266 0.023 -0.02 0.036 -0.106 0.064 -0.006 0.006 0.671 0.076 0.230 0.023 -0.096
17 Relative exploration × Relative centrality -0.257 2.714 0.020 -0.033 0.167 -0.079 0.019 0.023 -0.010 0.026 0.047 0.070 -0.016 -0.105
18 Relative exploration squared × Relative
centrality
-0.224 2.358 0.012 -0.122 0.365 -0.250 0.022 -0.014 -0.032 0.056 0.698 0.081 -0.024 -0.139
19 Relative exploration × Common ties 5.461 10.400 0.012 -0.023 0.119 -0.075 0.064 0.013 -0.024 0.111 0.073 0.204 -0.023 -0.077
20 Relative exploration squared × Common ties 3.951 8.563 0.044 -0.082 0.243 -0.210 0.137 -0.01 0.029 0.193 0.092 0.708 -0.016 -0.199
Note: (a) N = 3,752
(b) Means were calculated using un-centered variable values.
Variables 13 14 15 16 17 18 19
13 Relative exploration
14 Relative exploration squared 0.650
15 Relative exploration × Repeated alliance ties -0.151 -0.197
16 Relative exploration squared × Repeated
alliance ties
-0.069 -0.237 0.534
17 Relative exploration × Relative centrality -0.073 -0.123 0.107 0.062
18 Relative exploration squared × Relative
centrality
-0.150 -0.173 0.070 0.100 0.521
19 Relative exploration × Common ties -0.005 -0.032 0.296 0.21 0.136 0.142
20 Relative exploration squared × Common ties 0.064 -0.02 0.185 0.286 0.145 0.181 0.563
Page 38 of 40
39
TABLE 2. Fixed-effects analyses of firms’ competitive aggressiveness
1 2 3 4 5 6
Constant 39.286 37.947 38.294 36.838 39.115 38.105
(19.012) (16.323) (16.081) (15.986) (15.504) (15.493)
R&D intensity -0.001 -0.002 -0.002 -0.001 -0.002 -0.001
(0.001) (0.002) (0.002) (0.002) (0.002) (0.002)
ROA 0.024 -0.003 -0.005 -0.004 -0.013 -0.010
(0.029) (0.035) (0.037) (0.029) (0.036) (0.032)
Financial slack 0.041 0.040 0.040 0.040 0.040 0.040
(0.018) (0.017) (0.017) (0.017) (0.017) (0.017)
Coopetition 2.832 2.786 2.776 2.785 2.775 2.769
(0.740) (0.704) (0.702) (0.704) (0.708) (0.705)
Technological overlap -0.113 -0.145 -0.129 -0.207 -0.089 -0.150
(0.184) (0.172) (0.178) (0.186) (0.167) (0.188)
Alliance scope 0.015 -0.139 -0.156 -0.138 -0.142 -0.155
(0.014) (0.036) (0.035) (0.033) (0.033) (0.030)
Repeated alliance ties 0.017 -0.054 0.032 -0.053 -0.045 0.027
(0.025) (0.015) (0.039) (0.016) (0.017) (0.038)
Relative centrality 0.010 0.010 0.009 0.005 0.010 0.005
(0.006) (0.005) (0.006) (0.006) (0.005) (0.007)
Common ties -0.001 -0.001 -0.001 -0.000 0.001 0.001
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Multiparty -0.048 -0.047 -0.029 -0.044 -0.042 -0.021
(0.077) (0.087) (0.082) (0.084) (0.086) (0.078)
Network density -1.001 -1.005 -1.015 -0.997 -0.933 -0.984
(0.348) (0.384) (0.381) (0.392) (0.382) (0.386)
Relative exploration 1.508 1.803 1.492 1.880 2.011
(0.278) (0.273) (0.271) (0.243) (0.245)
Relative exploration sq. -1.439 -1.735 -1.430 -1.831 -1.970
(0.267) (0.252) (0.273) (0.239) (0.254)
Relative exploration ×
repeated alliance ties
-0.670 -0.560
(0.161) (0.148)
Relative exploration sq. ×
repeated alliance ties
0.692 0.572
(0.151) (0.143)
Relative exploration ×
relative centrality
0.056 0.058
(0.015) (0.014)
Relative exploration sq. ×
relative centrality
-0.054 -0.058
(0.023) (0.023)
Relative exploration ×
common ties
-0.023 -0.017
(0.008) (0.008)
Relative exploration sq. ×
common ties
0.024 0.018
(0.009) (0.009)
R
2
0.423 0.434 0.436 0.434 0.435 0.437
Log likelihood -3883 -3849 -3843 -3847 -3845 -3839
χ
2
(2) (compared with the baseline) 67 79 72 75 88
Prob >
χ
2
2.236e-15 5.655e-18 2.914e-16 4.396e-17 9.878e-20
Note: (a) N = 3,752
(b) Robust standard errors in parentheses
(c) Year dummies included in regressions but not reported
(d) “sq.”: squared
Page 39 of 40
40
FIGURE 1. The relationship between relative exploration and
competitive aggressiveness
FIGURE 2. The moderation effect of repeated alliance ties
FIGURE 3. The moderation effect of relative centrality FIGURE 4. The moderation effect of common ties
0 .1 .2 .3 .4
Competitive Aggressiveness
0 1
Relative Exploration
0 .1 .2 .3 .4
Competitive Aggressiveness
0 1
Relative Exploration
Repeated Ties = mean - 1 s.d. Repeated Ties = mean + 1 s.d.
0 .1 .2 .3 .4
Competitive Aggressiveness
0 1
Relative Exploration
Relative Centraltiy = mean - 1 s.d. Relative Centraltiy = mean + 1 s.d.
0 .2 .4.1 .3
Competitive Aggressiveness
0 1
Relative Exploration
Common Ties = mean - 1 s.d. Common Ties = mean + 1 s.d.
Page 40 of 40
