Paper 3

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/336200338

What Drives Exploration? Convergence and Divergence of Exploration

Tendencies among Alliance Partners and Competitors

Article  in  The Academy of Management Journal · October 2019

DOI: 10.5465/amj.2017.1409

CITATIONS

4 READS

1,236

4 authors:

Some of the authors of this publication are also working on these related projects:

Big Data and Data Science in Management View project

Geert Duysters

Eindhoven University of Technology

113 PUBLICATIONS   9,138 CITATIONS   

SEE PROFILE

Dovev Lavie

Università commerciale Luigi Bocconi

59 PUBLICATIONS   7,761 CITATIONS   

SEE PROFILE

Anna Sabidussi

Tilburg University

25 PUBLICATIONS   149 CITATIONS   

SEE PROFILE

Uriel Stettner

Tel Aviv University

15 PUBLICATIONS   763 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Uriel Stettner on 25 October 2019.

The user has requested enhancement of the downloaded file.

1

WHAT DRIVES EXPLORATION? CONVERGENCE AND DIVERGENCE OF

EXPLORATION TENDENCIES AMONG ALLIANCE PARTNERS AND COMPETITORS†

Geert Duysters

Tilburg University

Dovev Lavie

Bocconi University

Anna Sabidussi

Tilburg University

AVANS University of Applied Sciences

Uriel Stettner

Tel Aviv University

First version: December 29, 2017

Revised: October 23, 2018

Revised: June 26, 2019

Revised: September 20, 2019

Accepted for publication in the Academy of Management Journal

† The authors are listed alphabetically. We acknowledge the research assistance of Nico Rasters and Gianluca

Tarasconi, and appreciate the feedback received from Shiko Ben-Menachem and Justin Jansen. We also thank seminar

participants at the Wharton School, INSEAD, Bocconi University, IESE Business School, ETH Zurich, NUS Business

School, BI Norway, the University of Vienna, and WU Vienna. An early version of this study was presented at the

2016 Strategic Management Society Conference in Berlin. The paper was also presented at the 2017 Israel Strategy

Conference in Haifa, the 2018 Special Strategic Management Society Conference in Oslo, and the 2018 Academy of

Management Conference in Chicago. The authors acknowledge research grants received from the Eli Hurvitz Institute

for Strategic Management and the Coller Foundation. Dovev Lavie acknowledges his fellowship with the Invernizzi

Center for Innovation, Organization, Strategy and Entrepreneurship at Bocconi University (ICRIOS).

2

WHAT DRIVES EXPLORATION? CONVERGENCE AND DIVERGENCE OF

EXPLORATION TENDENCIES AMONG ALLIANCE PARTNERS AND COMPETITORS

ABSTRACT

Management research has alluded to organizational and environmental conditions that drive firms’

tendencies to explore versus exploit. We complement this research by developing theory on

vicarious learning to explain how a firm adjusts its own exploration level based on the exploration

levels of its alliance partners and competitors. Using panel data on 180 electronics firms publicly

traded in the U.S., we reveal an inverted U-shaped association between the firm’s exploration

tendency and the exploration levels of its partners and competitors. Convergence is explained by

imitation and legitimation, while divergence is associated with risk perception and specialization

in the knowledge domain. We further show how the convergence of the exploration tendency

becomes stronger under firm-specific uncertainty but weaker when the exploration patterns

exhibited by the firm’s partners and competitors are incoherent. Finally, counter to expectations,

we show that this convergence is weakened by the technological proximity of the firm’s

competitors. Our findings inform research on vicarious learning and the antecedents of exploration

by underscoring the role of interdependence in firms’ exploration tendencies.

3

INTRODUCTION

The exploration-exploitation framework has gained much scholarly attention in recent years, with

its impact extending even beyond the management discipline (Wilden, Hohberger, Devinney, &

Lavie, 2018). Nevertheless, “there has been little attempt to uncover why some organizations

emphasize exploration while others mostly pursue exploitation” (Lavie, Stettner, & Tushman,

2010: 118). A key question is: what drives a firm’s tendency to explore in its knowledge domains?

According to research on organizational learning, exploration involves developing knowledge

elements that are new to the firm, whereas exploitation entails leveraging and refining the firm’s

existing knowledge (Levinthal & March, 1993). By exploring, the firm can avoid obsolescence and

remain competitive, while exploitation is essential for its efficiency and for securing its market

position (March, 1991). The need to allocate limited resources to these distinct learning activities,

which involve conflicting routines, creates inherent tradeoffs between them. Accordingly, some

scholars have conceptualized these activities as lying on a continuum that ranges from exploitation

to exploration (Lavie et al., 2010). Firms vary in their tendencies to explore versus exploit, and

adjust these tendencies over time (e.g., Lavie & Rosenkopf, 2006). Still, prior research has mostly

focused on the consequences of exploration and on the means by which firms balance exploration

and exploitation rather than on the factors that drive these tendencies.

Research on the antecedents of exploration alludes to environmental conditions that can

facilitate it, such as resource munificence, technological and market uncertainty, environmental

dynamism, and the intensity of competition (Jansen, Volberda, & Van Den Bosch, 2005; Kim &

Rhee, 2009; Sidhu, Volberda, & Commandeur, 2004; Voss, Sirdeshmukh, & Voss, 2008). These

exogenous factors uniformly shape firms’ tendencies to explore in a particular industry, and thus

cannot explain heterogeneity in their tendencies. However, some studies have shown that

organizational characteristics such as age, size, organizational structure, and culture can explain

4

deviation from the typical exploration tendency in an industry (Jansen, Van Den Bosch, &

Volberda, 2006; Sorensen & Stuart, 2000; Voss et al., 2008). Other studies have identified

managerial antecedents such as managers’ attention to innovation, advice seeking, leadership style,

socio-psychological aspects, and adoption of open innovation, which can influence a firm’s

tendency to explore (Alexiev, Jansen, Van den Bosch, & Volberda, 2010; Jansen, Kostopoulos,

Mihalache, & Papalexandris, 2016; Jansen, Vera, and Crossan, 2009; Khanagha, Volberda, and

Oshri, 2017; Li, Maggitti, Smith, Tesluk, & Katila, 2013). But even though prior research has made

progress in understanding heterogeneity in firms’ tendencies to explore, it has implied that firms’

tendencies are independent or are collectively shaped by industry conditions.

In the current study, we argue and demonstrate that firms’ tendencies to explore are in fact

interdependent and associated with the corresponding exploration levels of other firms in their main

reference groups. Specifically, we consider how the patterns of exploration exhibited by alters with

whom a firm maintains cooperative and competitive ties shape the firm’s inclination to explore.1

By focusing on the firm’s motivation to converge or diverge from the exploration levels of these

alters, we complement research on organizational learning that alludes to conditions that uniformly

shape firms’ exploration tendencies in a particular industry.

Convergence with the exploration tendencies in a reference group is far from trivial, because

each firm has an idiosyncratic exploration level that is considered desirable (Levinthal & March,

1993) and because the frequency of a behavior does not clearly indicate its efficiency, and thus

may be insufficient to encourage firms to follow suit (Gupta & Misangyi, 2018). This is especially

the case with exploration, whose outcomes are unforeseen in the short term (March, 1991).

1 We study the extent of exploration (how much a firm explores) rather than the knowledge domains in which

exploration is pursued (where a firm explores). Hence, convergence with the exploration tendencies of the reference

group does not necessarily entail entering the same knowledge domains. When the firm’s partners and competitors

enter new knowledge domains, the firm may invest in exploration that extends its own knowledge domains.

5

Research on vicarious learning suggests that firms tend to imitate successful behaviors (Greve,

2011), but this learning may be hampered when the success of the behavior is uncertain (Terlaak

& Gong, 2008). In turn, our theory suggests that imitation and legitimation facilitate convergence

between a firm’s tendency to explore and the level of exploration exhibited by its partners and

competitors; nevertheless, convergence increases exploration only up to a point, beyond which it

is mitigated as a result of aversion of perceived risk. We further propose that as the exploration

level of partners and competitors becomes excessive, the firm diverges from it and shifts to

exploitation. Such divergence is explained by efforts to specialize in the knowledge domain. We

also expect convergence to intensify when the firm faces increasing uncertainty and becomes more

technologically proximate to its partners and competitors. Finally, we suggest that variation in the

exploration levels of the firm’s partners and competitors weakens convergence.

We test our predictions with panel data on 180 electronics firms publicly traded in the U.S.

Our findings support our conjectures with the exception of technological proximity, which does

not affect convergence with partners and weakens convergence with competitors as a result of

firms’ differentiation efforts. Hence, although each firm is expected to have an idiosyncratic level

of exploration that serves its needs, we show that a firm aligns its exploration tendency with the

exploration levels of its unique set of partners and primary competitors, at least to an extent. We

also reveal that this convergence is subject to boundary conditions, and eventually, as exploration

levels become excessive, gives way to divergence of exploration tendencies.

Our study contributes to research on the antecedents of exploration by explaining some

previously unobserved heterogeneity and by uncovering an important antecedent that underscores

interdependence in firms’ exploration tendencies. We reveal that a firm’s tendency to explore is

related not only to exogenous industry conditions (e.g., Jansen et al., 2005; Sidhu et al., 2004) and

organizational factors (e.g., Greve, 2007; Jansen et al., 2006), but also to the typical exploration

6

levels prevalent in the firm’s unique reference groups. This association varies from convergence to

divergence, depending on the observed exploration levels. We conclude that firms do not operate

in isolation, nor do they uniformly react to changing industry conditions. Rather, their tendencies

to explore are interactively constructed in a network wherein firms observe the exploratory

behavior of their unique partners and primary competitors and position themselves accordingly.

We thus offer a novel explanation for the heterogeneity in exploration tendencies.

Finally, we advance research on vicarious learning, which has underscored the roles of

imitation and legitimation in driving convergence of behaviors (e.g., Haunschild & Miner, 1997;

Lieberman & Asaba, 2006), but has paid less attention to boundary conditions that lead to divergent

behaviors. We show that imitation and legitimation are offset by risk aversion and efforts to

specialize in the knowledge domain when the outcomes of alters’ behaviors become unpredictable,

thus leading to divergence of behaviors. We further identify firm-specific uncertainty, variance in

behaviors, and proximity as important boundary conditions for vicarious learning. Hence, we offer

a nuanced perspective on the convergence and divergence of behaviors in reference groups.

Insert Figure 1 here

THEORY AND HYPOTHESES

Convergence and Divergence of Exploration Tendencies

A firm’s exploration tendency is a typical corporate behavior that evolves via experiential and

vicarious learning (Baum, Li, & Usher, 2000). In vicarious learning, firms adjust their corporate

behavior in response to behaviors prevalent in their reference groups (Srinivasan, Haunschild, &

Grewal, 2007). One important reference group that promotes vicarious learning is alliance partners,

which directly interact with the firm (Powell, Koput, & Smith-Doerr, 1996) and often serve as

trendsetters and role models for the firm (Abrahamson, 1996). Besides its partners, the firm’s

7

competitors are another important reference group (Fiegenbaum & Thomas, 1995; Porac, Thomas,

& Baden-Fuller, 1989). Firms track their competitors’ actions and position themselves vis-à-vis

competitors (Chen, 1996). They pay most attention to their primary competitors (Clark &

Montgomery, 1999) in order to closely monitor them and learn from their behavior (Haunschild &

Miner, 1997; Hsieh, Tsai, & Chen, 2015). Hence, vicarious learning from primary competitors

complements learning via direct interaction with partners (Baum et al., 2000).

When vicariously learning from partners, a firm can observe its partners’ corporate behavior

irrespective of the scope and content of its alliances with them (Khanna, Gulati, & Nohria, 1998;

Lavie, 2009). Similarly, vicarious learning from primary competitors may encompass corporate

behaviors (Miner & Mezias, 1996), as in learning from the failures of competitors, which helps the

firm reflect on causal processes and develop its own practices (Kim & Miner, 2007). According to

research on vicarious learning from reference groups, a firm is inclined to compare and adopt

behavioral patterns that are typical of the population average, which indicates a widely adopted

behavior (Hu, Blettner, & Bettis, 2011). Although some firms may rely on a small reference group

of leaders that exhibit superior performance (Massini, Lewin, & Greve, 2005), this is unlikely in

the case of exploration, whose performance outcomes are unforeseen in the short term.

Firms learn a range of corporate behaviors from their reference groups, such as alters’

innovation strategies (e.g., Semadeni & Anderson, 2010), product introductions (e.g., Giachetti &

Lanzolla, 2016), international expansion (e.g., Henisz & Delios, 2001), and market entry (e.g.,

Haveman, 1993), but there is heterogeneity in the extent to which these behaviors are followed

(Gupta & Misangyi, 2018). A firm is more likely to track and adopt the corporate behavior of its

partners and primary competitors when such behavior is visible (Baum et al., 2000) and entails

uncertainty (O’Neill, Pouder, & Buchholtz, 1998; Srinivasan et al., 2007). By definition,

exploration is observable yet inherently uncertain, forcing the firm to confront outcomes that

8

cannot be foreseen in the short term (March, 1991). As a result, we expect a firm to engage in

vicarious learning that is driven by imitation and legitimation and that leads to convergence with

the typical exploration level of the firm’s partners and primary competitors.2

Specifically, vicarious learning of exploratory behavior is driven by imitation that enables

firms to seek adaptive responses to common challenges (Kraatz, 1998). By acquiring knowhow

from alliance partners or by scanning their competitive environment, firms are prompted to imitate

the observed behavior of alters (Huber, 1991). Imitation is invoked by the perception that the

information that is available to alters and that guides their exploration efforts is more valuable than

one’s own information. Assuming that alters possess superior information or expertise facilitates

imitation of their exploration level. As information is revealed about firms that adopt this level of

exploration, imitation is further reinforced in the reference group. Aligning the firm’s exploration

tendency with the exploration level exhibited by alters is thus driven by the desire to overcome

information asymmetry (Kraatz, 1998; Lieberman & Asaba, 2006), capitalize on opportunities for

expanding the firm’s own knowledge domains (Anand, Mesquita, & Vassolo, 2009), and imitate

an effective practice that has been tested by others (Greve, 1996, 1998).

Besides vicariously learning proven practices, convergence of exploration tendencies may

result from imitation of common practices in a search for legitimacy (Suddaby, Bitektine, & Haack,

2017), which is essential when a corporate behavior entails risk and uncertainty (Lieberman &

Asaba, 2006). Indeed, scholars have argued that “legitimacy‐based reference groups guide firms in

their mimetic behavior” (Barreto & Baden‐Fuller, 2006: 1559). Exploration can be considered

legitimate when there is a shared perception of its appropriateness. Adopting a behavior that is

2 We study the extent of exploration (how much a firm explores) rather than the practice of exploration (how a firm

explores). Adjusting a firm’s tendency to explore based on its alters’ level of exploration is more straightforward than

learning the practice, which involves tacit routines for knowledge creation (Brix, 2017; Rosenkopf & Nerkar, 2001).

9

deemed appropriate and desirable in a social context contributes to the adopter’s legitimacy in the

eyes of external stakeholders (DiMaggio & Powell, 1983; Suchman, 1995). Firms need to ensure

that their exploration tendencies are sufficiently similar to those of alters in their reference groups,

e.g., primary competitors, in order to signal to their stakeholders that they conform to industry

norms. Firms conform to these expectations because the external endorsement obtained via

legitimacy helps in gaining access to resources and carrying out the firms’ exploration efforts.

Thus, a firm that adheres to the exploration level of role models can enhance its legitimacy beyond

the expected performance gain associated with exploration (Deephouse, 1996; Haunschild &

Miner, 1997; Lieberman & Asaba, 2006). In sum, convergence with the exploration level of

partners and primary competitors may be driven by vicarious learning in which the firm imitates a

popular behavior in a search for legitimacy (Henisz & Delios, 2001; Suchman, 1995).

A remaining question is whether firms engage in frequency imitation of common practices or

rather outcome imitation (Haunschild & Miner, 1997). Because the outcomes of exploration cannot

be foreseen in the short term, it is unlikely that exploration merely reflects the firm’s intention to

do the right thing. Rather, imitation reflects unconscious cognitive processes such as herding in

addition to more consciously deliberate processes (Gupta & Misangyi, 2018). Regardless of its

underlying cause, imitation facilitates convergence with the typical exploration level exhibited by

partners and primary competitors. For both reference groups, imitation is driven by the perceived

value and available information. However, for partners with whom the firm maintains cooperative

relations, information on their exploration level is more readily accessible than for competitors,

which may limit the firm’s access to information. In turn, information on the exploration level of

competitors is more relevant and valuable to the firm that operates in similar domains.

Nevertheless, we expect the convergence between a firm’s tendency to explore and the

exploration level exhibited by its partners and primary competitors to increase only up to a certain

10

point, beyond which it is mitigated. Operating at high levels of exploration entails exorbitant risk

given that it involves entering several new knowledge domains. Thus, the firm may be unable to

support extensive exploration (Uotila, Maula, Keil, & Zahra, 2009). Because the likelihood of

successful exploration is lower than that of exploitation (Levinthal & March, 1993), the firm may

avoid aligning its exploration level with the excessive exploration levels of its partners and

competitors in order to minimize losses from failed exploration. Hence, the firm’s aversion to the

perceived risk of excessive exploration exhibited by its partners or primary competitors is likely to

mitigate the convergence associated with imitation and legitimation.

Whereas perceived risk restricts convergence with the exploration levels of partners and

primary competitors, specialization in the firm’s knowledge domain prompts divergence from their

excessive exploration levels, and thus a tendency to revert to exploitation. Specialization in the

knowledge domain fosters experiential learning that counters vicarious learning and deters

imitation (Reed & DeFilippi, 1990). Although for both reference groups, specialization drives

divergence, it is ascribed to division of labor with partners as opposed to differentiation vis-à-vis

competitors. Specifically, at high levels of partner exploration, the firm can divide labor with its

partners that engage in extensive exploration, while it reverts to internal exploitation in a narrow

knowledge domain (Stettner & Lavie, 2014). Whereas at low levels of partner exploration, the firm

must rely on its internal exploration efforts, at high levels of partner exploration, the firm can rely

on partners for externally extending its knowledge domains (Rosenkopf & Almeida, 2003). As it

increases its reliance on partners for exploration, it can specialize in exploiting the knowledge that

it has accumulated internally. Hence, when partners engage in excessive exploration, the firm can

rely on their complementary expertise in new knowledge domains. Because of such division of

labor, some partners are prone to further increase their exploration level while the firm gravitates

toward exploitation (Hess & Rothaermel, 2011). Hence, as partners explore more, the firm can

11

capitalize on their boundary-spanning, which, in turn, further restricts its own exploration tendency.

Finally, at high levels of competitor exploration, the firm is likely to diverge from the

exploration tendency of its primary competitors and revert to exploitation. This specialization in

its existing knowledge domains is due to the firm’s intent to differentiate itself from competitors

while avoiding increased investments in innovation that can escalate technology races (Deephouse,

1999). When the firm diverges from its competitors’ excessive exploration level and concentrates

instead on its established knowledge domains, it can reinforce its corporate identity, rely on existing

skills and capabilities, enhance its unique value proposition to customers, and defend its industry

position. Differentiation explains the firm’s reversion to exploitation once its primary competitors

turn to excessive exploration. In sum, at high levels of exploration by both partners and primary

competitors, specialization in the knowledge domain offsets the vicarious learning associated with

imitation and legitimation. The firm is likely to adjust its tendency to explore per the exploration

levels of these alters up to a threshold, beyond which the firm is expected to revert to exploitation.

Hypothesis 1. A firm’s tendency to explore exhibits an inverted U-shaped association with the

exploration levels of (a) its alliance partners and (b) its primary competitors.

Uncertainty, Variation, and Technological Proximity as Boundary Conditions

Firms vary with respect to the extent to which their exploration tendencies converge with those

of their partners and primary competitors. Convergence is contingent on conditions that can

influence the firm’s motivation and ability to engage in vicarious learning (e.g., Baum et al., 2000;

Gioia & Manz, 1985; Haunschild & Miner, 1997) and thus align its exploration tendency with

those of alters. These conditions include firm-specific uncertainty, variation in the exploration

levels of partners and primary competitors, and technological proximity to these alters.

A firm’s inclination to follow the exploration level of alters depends in part on the uncertainty

that it encounters. Firms experience distinctive challenges in predicting future outcomes. They

12

often face market uncertainty (Srinivasan et al., 2007), but some uncertainty remains firm-specific

(Beckman, Haunschild, & Phillips, 2004). Firm-specific uncertainty refers to the “lack of assurance

about the probability and outcomes of corporate decisions” (Gulati, Lavie, & Singh, 2009: 1220).

It indicates the difficulty that managers face in predicting environmental trends in light of

idiosyncratic internal factors (Beckman et al., 2004), such as the firm’s decisions, capabilities, and

strategies. Firm-specific uncertainty can be only partially resolved by the firm’s actions (Cuypers

& Martin, 2010). Although firm-specific uncertainty is defined from the standpoint of managers,

it also exposes the firm to potential speculation by external stakeholders (Beckman et al., 2004).

Vicarious learning is likely under conditions of uncertainty, which prompts managers to seek

external guidance (Srinivasan et al., 2007). This suggests that convergence increases with firm-

specific uncertainty. Convergence with the exploration levels of partners and primary competitors

is especially likely under uncertainty given the nature of exploration. In reaction to the challenge

of predicting the outcomes of its exploration efforts, the firm is likely to seek more guidance as

firm-specific uncertainty increases. When uncertainty increases, managers become doubtful about

the prevalence of opportunities for exploration, the possibility of pursuing these opportunities, and

the opportunities’ prospects (McMullen & Shepherd, 2006). Firm-specific uncertainty restricts the

firm’s ability to recognize, assess, and pursue exploration opportunities. As a result, its managers

may become hesitant about missing market opportunities or assuming the risk of exploration. The

high uncertainty about the outcomes of exploration motivates the firm to learn from the conduct of

alters (Greve, 1996; Rosenkopf & Abrahamson, 1999). Moreover, as uncertainty increases, so does

the belief that partners and competitors possess more reliable information about exploration

opportunities (Haunschild & Miner, 1997; Lieberman & Asaba, 2006). Consequently, the firm is

more prone to imitate alters and align its exploration efforts with their exploration level.

Firm-specific uncertainty reinforces convergence with the exploration levels of alters not only

13

because it fosters imitation, but also because it increases the need for legitimacy. As uncertainty

increases, convergence with the exploration levels of alters becomes essential for convincing

stakeholders that the firm is able to cope with environmental challenges (Kondra & Hinings, 1998).

Indeed, alignment with alters’ behavior enhances the firm’s legitimacy in the eyes of stakeholders

(Gimeno, Hoskisson, Beal, & Wan, 2005). Hence, uncertainty reinforces the need for legitimacy

and spurs the firm’s efforts to converge with the exploration levels of its partners and competitors.

Overall, firm uncertainty reinforces vicarious learning by increasing the need for legitimacy

and the reliance on alters for information. This increases convergence of the firm’s exploration

tendency with that of its partners and competitors at any level of exploration pursued by them.

Hypothesis 2. Firm uncertainty increases the positive association between a firm’s tendency to

explore and the exploration levels of (a) its alliance partners and (b) its primary competitors.

Although a firm may learn to align its exploration tendency with that of alters, this may not be

straightforward when the partners or competitors exhibit varied levels of exploration. Research on

vicarious learning suggests that “faced with such streams of inconsistent inputs and with

maneuvering limited cognitive capacity, at least some managers are uncertain about the underlying

covariation between practice value and the relevant trait” (Terlaak & Gong 2008: 850). Indeed,

incoherent behavior makes it difficult for the firm to identify the typical tendency in the reference

group and to align its exploration tendency accordingly. Convergence entails discerning

exploration levels, interpreting them, synthesizing this input, and adjusting the firm’s exploration

tendency. Variation in the exploration levels of partners or competitors inhibits these processes.

Specifically, to learn about the desirable level of exploration, the firm needs to monitor the

behavior of its partners and competitors. Such monitoring requires discerning the typical tendency

in these reference groups, which becomes more challenging the more dispersed the observed

exploration pattern. Hence, convergence of exploration tendencies is impaired by inconsistent

14

information and unclear causal relationships (Gioia & Manz, 1985). Next, convergence involves

interpreting information, giving meaning to data, and synthesizing it (Maitlis & Christianson,

2014). When the exploration of partners or competitors reflects coherent tendencies, interpretation

and synthesis are straightforward. However, when their exploration levels vary, it is more difficult

to make sense of their tendencies and identify the desirable exploration level. As a result, the firm’s

ability to converge with the exploration levels of its partners and competitors is compromised.

Moreover, convergence entails devising organizational routines for adjusting the firm’s exploration

tendency based on comprehension and interpretation of the learned practice (Nelson & Winter,

1982). However, variation in the exploration levels of partners and competitors makes it difficult

to learn a set of routines that enable the firm to follow their exploration tendencies.

In addition, when alters in the reference group exhibit inconsistent patterns of exploration, it

becomes more difficult to gain legitimacy by following their exploration tendencies (Henisz &

Delios, 2001; Suchman, 1995). Stakeholders may associate such variation with randomness and

unreliability of the behavior, and thus perceive it as illegitimate (Rhee, Kim, & Han, 2006).

Variation in the exploration levels of alters may preclude consensus among stakeholders about the

desirable level of exploration, so that the firm cannot gain legitimacy by conforming to that level

(Deephouse, 1999). Indeed, when the firm’s partners or primary competitors disagree about the

desired level of exploration, the firm’s convergence with the typical exploration level does not

enhance legitimacy and validation of its exploration endeavors. Without a well-received reference

for exploration, convergence with the typical tendency is less likely to be deemed appropriate

(Suchman, 1995). Overall, variation in the exploration levels of partners and primary competitors

hampers the learning and interpretation of their tendencies, constrains the firm’s ability to imitate

them, and undermines their legitimacy, thus mitigating convergence with these tendencies.

Hypothesis 3a. Variation in the exploration level of alliance partners weakens the positive

15

association between the firm’s tendency to explore and the exploration level of its partners.

Hypothesis 3b. Variation in the exploration level of primary competitors weakens the positive

association between the firm’s tendency to explore and the exploration level of its competitors.

Whereas variation in exploration levels can impede convergence, technological proximity can

facilitate it. Research on vicarious learning suggests that “for another organization’s actions to

influence a potential imitator, the organization and its context must be seen as sufficiently similar

to the imitator’s” (Baum et al., 2000: 775). A key aspect of similarity is technological proximity,

which refers to the extent of overlap in firms’ technical knowledge domains (Rosenkopf &

Almeida, 2003), or “the degree to which their technological problem-solving focuses on the same

narrowly defined areas of knowledge” (Makri, Hitt, & Lane, 2010: 606). Prior research has noted

that similarity in knowledge domains facilitates knowledge transfer (Mowery, Oxley, & Silverman,

1996; Phene, Fladmoe-Lindquist, & Marsh, 2006; Rosenkopf & Almeida, 2003), but this can also

apply to imitation of exploratory behavior. The more proximate a firm to alters, the more relevant

their behavior, and the easier it is for the firm to monitor and follow that behavior (Baum et al.,

2000). Thus, technological proximity increases the firm’s motivation and ability to align its

exploration tendency with the tendencies of its partners and primary competitors.

In particular, technological proximity is expected to facilitate convergence of exploration

levels because operating in comparable environments and engaging in similar activities affect

judgment about the relevance of alters (Greve, 2005). According to the homophily principle, ties

to similar alters are considered more significant (McPherson, Smith-Lovin, & Cook, 2001).

Technological proximity to alliance partners and competitors thus reinforces the perception that

the firm can rely on these alters as a relevant source of information about corporate behavior

(Kilduff, Elfenbein, & Staw, 2010). Hence, technological proximity is expected to facilitate the

firm’s monitoring and learning of the exploration levels of partners and competitors. In fact,

16

technologically proximate firms explore in similar knowledge domains (Phene et al., 2006;

Rosenkopf & Almeida, 2003) and thus are likely to develop similar perceptions about opportunities

and adopt common behaviors (Giachetti & Lanzolla, 2016). Technological proximity to partners

and competitors can even foster identification and common perceptions of opportunities in the

industry, which create a shared vision about exploration prospects (Dobrev, 2007; Kraatz, 1998;

O’Neill et al., 1998) and lead to consensus about competencies for entering new knowledge

domains, which further promote imitation in the reference group. The greater the technological

proximity, the more relevant the expertise of partners and competitors and the better the firm can

comprehend their exploration, which prompts it to more closely follow their exploration levels.

Hence, technological proximity reinforces imitation and convergence of exploration tendencies.

Finally, the firm’s search for legitimacy may gain from technological proximity because

convergence with the behavior of technologically proximate alters can legitimize the firm’s

decisions to enter new knowledge domains (Deephouse, 1996). Convergence with the exploration

tendencies of technologically proximate partners and competitors is especially important given the

inherent riskiness of exploration. The firm’s managers may become more confident about pursuing

risky exploration when the firm’s closest partners and competitors engage in exploration, assuming

that conformity will reduce these managers’ liability in case of failure (Schimmer & Brauer, 2012).

In sum, technological proximity to partners and competitors increases the attention that the

firm pays to their exploration, facilitates comprehension of exploration opportunities that those

alters have identified, and legitimizes exploration. Accordingly, technological proximity reinforces

vicarious learning and convergence with the exploration levels of partners and competitors.

Hypothesis 4a. Technological proximity of alliance partners strengthens the positive association

between the firm’s tendency to explore and the exploration level of its partners.

Hypothesis 4b. Technological proximity of primary competitors strengthens the positive

association between the firm’s tendency to explore and the exploration level of its competitors.

17

RESEARCH METHODS

Sample and Data

We tested our hypotheses with panel data on firms that are publicly traded in the U.S. and

operate in sectors of the electronics industry during 1990–2006. These sectors encompass

manufacturers of electronic devices, semiconductor components, and computer hardware,

including industrial and commercial machinery and computer equipment (SIC 35); electronic and

electrical equipment and components (SIC 36); and measurement, analysis, and control instruments

(SIC 38). The intensive competition and alliance formation in these sectors (Stuart, 2000) ensure

variance in firms’ patents, competitors, and partners. These sectors were chosen because at least

40% of all firms in these sectors apply for patents, which is essential for calculating patent-based

measures (Cockburn & Griliches, 1987). To develop meaningful measures of exploration, we

limited our sample to firms with financial data that applied for patents for at least five consecutive

years and that had a median patent application count of at least four patents per year. We corrected

potential sampling bias with our first-stage model. The resulting sample included 184 firms.

We relied on NBER patent data to consistently assess firms’ exploration tendencies (Katila &

Ahuja, 2002; Rosenkopf & Nerkar, 2001), and we corrected or supplemented incomplete data from

Comets Patent database (Griliches, 1998).3 We study patent applications rather than granted patents

because the year in which a patent was applied for is closer to the time of invention (Hall &

Ziedonis, 2001; Jaffe, Trajtenberg, & Henderson, 1993).4 Because patents are often assigned to

subsidiaries rather than to the headquarters, we identified each firm’s subsidiaries using the NBER

3 The focus on U.S. patents is justified by firms’ incentives to secure legal protection in the U.S. and the reputation of

the U.S. judicial system in providing effective protection of intellectual property (Gallini, 2002). We limit our data to

utility patents, while excluding design, reissue, and plant patents (Hall, Jaffe, & Trajtenberg, 2001). 4 The delay between the time of invention and the patent application does not generally exceed three months, but the

time lag between a patent application and the granting of the patent by the U.S. Patent Office may be three to four

years. The patent application date thus better reflects the time of knowledge creation (Griliches, 1998).

18

and the Corporate Affiliations databases, and cross-validated acquisitions in the SDC database.

Thus, we accounted for patents of the firm’s subsidiaries, but discarded patents of acquired firms

that were applied for prior to the acquisition (Puranam & Srikanth, 2007). This enabled us to study

exploration relative to the firm’s existing knowledge domains in a given year. We gathered data on

firms’ patents since 1985 to measure exploration experience starting the five years preceding the

study’s timeframe. In total, the 184 sampled firms and their subsidiaries applied for 280,080 patents

during 1985–2006. We aggregated patent counts to the firm-year level. The same procedure served

for measuring the exploration tendencies of partners and primary competitors.

To identify the firm’s alliance partners, we compiled alliance records from the SDC database,

considering active alliances as those formed in the past five years (e.g., Stuart, 2000). In total, 162

of the 184 sampled firms formed 6,735 alliances with 1,351 partners during 1985–2006.5 On

average, a firm formed 3.895 alliances per year and had a portfolio of 20.933 alliances during

1990–2006, with 74.499% of alliances formed with partners outside its two-digit SIC. To construct

our measures, we pooled records across all alliances in a firm-year.6

Next, we identified the firm’s competitors based on resource similarity (Chen, 1996),

acknowledging that firms tend to rely on supply-based (e.g., technologies developed) rather than

demand-based (e.g., customers served) definitions of competitors (Clark & Montgomery, 1999)

5 Following prior research, we excluded 633 alliances with 236 privately held partners for which data was unavailable

in the NBER database (e.g., Conti, 2014; Schilling, 2015; Sears & Hoetker 2014). Excluding the 8.591% alliances with

private partners is consistent with our theory because, unlike private partners, publicly traded partners are required to

disclose information about their exploration endeavors, while the exploration of private partners is less visible.

Furthermore, firms are likely to benchmark against alters with similar or higher status as role models for imitation and

legitimation (Haunschild & Miner, 1997; Haveman, 1993; Henisz & Delios, 2001). Therefore, public firms are more

likely to follow public partners than private partners. We applied a similar logic when focusing on public competitors.

Because we consider only the exploration levels of the primary competitors, a private competitor is unlikely to serve

as a benchmark. Comparisons of the 6,735 selected alliances to the 7,368 alliances including private partners reveal

no significant differences, so excluding private partners has limited implications. Finally, we account for potential

selection bias due to the exclusion of observations on private partners with our two-stage Heckman model. 6 We identified 718 patents (0.379%) that were jointly assigned to 28 firms (18.300%) and their 46 partners (3.651%).

Our findings remained unchanged when we exclude these jointly assigned patents from our data.

19

when observing exploration in knowledge domains. In the electronics industry, resource similarity

is more important than market communality as a result of the prolonged R&D process that precedes

product introduction to a common market. Because knowledge is the most relevant resource in this

industry (Dothan & Lavie, 2016), we relied on technological similarity captured by the overlap in

firms’ patent classes (e.g., Grimpe & Hussinger 2014; Polidoro, Ahuja, & Mitchell, 2011). For

each of the 184 firms, we identified competitors by tracking publicly traded firms that had at least

one patent class overlapping with those of the focal firm in the past five years. In line with research

on competitors’ reference groups and competitor identification (Clark & Montgomery, 1999; Porac

& Thomas, 1990), we selected the five primary competitors with the largest overlap. The overlap

was computed using the Jaffe (1986) measure: Pij = 𝐹𝑖𝐹𝑗

′

[(𝐹𝑖𝐹𝑖 ′ )(𝐹𝑗𝐹𝑗

′)]

1 2

, where Pij captures the annual

technological similarity of firm i to competitor j based on the angular separation between their

knowledge domain vectors, Fi and Fj. These knowledge domains represent the cumulative number

of patent applications across patent classes in a five-year window. For every firm-year, we selected

the firm’s five primary competitors with the highest Jaffe scores and pooled records across them.7

Finally, we gathered financial data from the Compustat and CRSP databases. For each firm-

year, we pooled the data across all partners and primary competitors. After listwise deletion of 275

records with missing data (10.528% of 2,612 records), the remaining data for testing the effect of

competitor exploration had 2,337 firm-year observations for 180 firms during 1990–2006. After

listwise deletion of 255 records with missing data (14.750% of 1,729 records), the data for testing

the effect of partner exploration had 1,474 firm-year observations for 153 firms.8

7 Increasing the reference group to seven top competitors did not materially affect our reported findings. 8 Missing data in Table 3a correspond to lack of patents for competitors (0.153%), and incomplete data in Compustat

(9.724%), CRSP (8.116%), and Corporate Affiliations (7.466%) databases; missing data in Table 3 correspond to lack

of patents for partners (7.403%) and incomplete data in Compustat (13.302%), CRSP (11.798%), and Corporate

Affiliations (11.393%) databases.

20

Measures

Dependent Variable. We measured firm exploration as a continuous variable (e.g., Greve,

2007; Lavie & Rosenkopf, 2006; Sidhu, Commandeur, & Volberda, 2007; Uotila et al., 2009),9

using the inverse of a normalized Herfindahl index that captures the diversity of unique patent

classes at year t based on patents applied for in the past five years. The measure took the form 1 −

𝐻𝐼𝑖𝑡 = N

(𝑁−1) (1 − ∑ 𝑆𝑟

2 )𝑁𝑟=1 , where Sr is the share of patent class r in firm i’s patent classes, and N

is the number of distinct patent classes. This measure captures a firm’s breadth of knowledge in a

given year, on a range between 0 and 1 (e.g., Argyres & Silverman, 2004; Trajtenberg, Henderson,

& Jaffe, 2002). Increase in the breadth of knowledge across various knowledge domains indicates

the firm’s tendency to explore (Ganzaroli, De Noni, Orsi, & Belussi, 2016; Gilsing, Nooteboom,

Vanhaverbeke, Duysters, & van den Oord, 2008; Guan & Liu, 2016). Thus, by tracking patent

applications in new patent classes (e.g., Ahuja & Lampert, 2001) while controlling for the firm’s

exploration in the previous year, we capture the firm’s tendency to explore by expanding into new

knowledge domains. To ensure that the knowledge domains to which the firm expands are indeed

new to the firm, we rely on patent classes rather than on patent subclasses, which can be potentially

related to each other.10 To avoid an inherent bias in calculating exploration (Lavie & Rosenkopf,

2006), we assumed that observations with fewer than two patents were balanced and assigned them

a value of 0.5 (Stettner & Lavie, 2014).11 Our measure was preferred to more complex measures

based on patent citations (e.g., Eggers & Kaul, 2018; Katila & Ahuja, 2002) that capture

9 The transition from exploration to exploitation is gradual, and the distinction between them is a matter of degree

rather than kind. Such transitivity and relativity call for their conceptualizing along a continuum (Lavie et al., 2010). 10 In auxiliary analysis, we replaced variables that were measured at the patent class level with measures at the patent

section, subclass, group, and subgroup levels. Classification at a higher level, e.g., section, yielded weaker support for

our hypotheses because of loss of discriminating power. Classification at a lower level, e.g., subclass (Rosenkopf &

Nerkar, 2001; Uotila et al., 2009), furnished consistent findings with the exception of Hypothesis 2a. 11 We obtained consistent findings when we dropped observations in which firms applied for fewer than two patents

per year. Consistent findings were obtained when we allowed the value for balance to range between 0.25 and 0.75.

21

exploration as knowledge that is new to the world rather than new to the firm (e.g., Eggers & Kaul,

2018; Fleming, 2001), and hence not fully in line with our theory on exploration and vicarious

learning. Explanatory variables were lagged by one year relative to our dependent variable.

Independent Variables. We applied a similar procedure for measuring the two independent

variables. Partner exploration was measured with an inverse Herfindahl index capturing the annual

diversity of patent classes of the firm’s alliance partners, considering all the patents applied for by

each partner in the past five years ending at year t-1. We measured partner exploration by averaging

this index across the firm’s partners that formed an alliance with the firm during this five-year

window. Competitor exploration was measured with an inverse Herfindahl index relating to the

average annual diversity of patent classes of the firm’s five primary competitors with the highest

Jaffe score for patent class overlap in year t-1.

Moderating Variables. We measured firm-specific uncertainty based on the volatility in the

firm’s stock price in year t-1 (Beckman et al., 2004). We calculated this measure as the difference

between the standardized monthly volatility of the firm’s stock price and the average standardized

monthly volatility in the stock prices of all sampled firms that year. By subtracting this market-

specific uncertainty component, we capture only the uncertainty that is idiosyncratic to the firm.

We divided the standard deviation in monthly closing stock price by its mean value (Gulati et al.,

2009). Hence, the measure took the form: √ ∑ (𝑝𝑖𝑇−𝑝𝑖)

212 𝑇=1

11×𝑝𝑖 2

- √ ∑ (𝑝𝑚𝑇−𝑝𝑚)

212 𝑇=1

11×𝑝𝑚 2

, where pit is firm i’s

stock closing price at the end of month T, which ranges from January to December. Similarly, pmt

is the average closing price of the sampled firms’ stocks at the end of month T.

We measured the variation in exploration levels of partners and competitors in year t-1 by

correspondingly calculating the variance in the independent variables. These measures capture the

variance in the exploration tendencies of alliance partners and the five primary competitors. We

22

measured the technological proximity of the firm to its partners and competitors using the Jaffe

(1986) proximity measure. Specifically, the technological proximity to a partner was measured

with the formula Pij = 𝐹𝑖𝐹𝑗

′

[(𝐹𝑖𝐹𝑖 ′ )(𝐹𝑗𝐹𝑗

′)]

1 2

, where Fi and Fj are vectors representing the knowledge

domains of firm i and partner j based on classes of patents applied for during a five-year window.

To compute the technological proximity to partners, we averaged this measure across the firm’s

partners in each year. Technological proximity to competitors was calculated using the average

Jaffe (1986) proximity measure corresponding to the firm’s five primary competitors.

Control variables. We included control variables characterizing the firm, its alliance portfolio,

and competitors. In particular, we controlled for the firm’s exploration level in year t-1, so that our

model estimates the firm’s exploration tendency, i.e., its inclination to change its exploration level

relative to its level of exploration in the preceding year. We also controlled for the firm’s age, size,

R&D intensity, corporate strategy function, financial solvency, and performance gap. The firm’s

size can affect its innovation output by decreasing exploration (Beckman et al., 2004). Firm size

was measured using the firm’s total revenues (Tallman & Li, 1996).12 We measured the firm’s age

as elapsed years since the firm’s incorporation. As the firm matures, it tends to decrease its

exploration level (Kang & Uhlenbruck, 2006). A firm’s R&D intensity reflects the extent to which

the firm invests in new technologies and builds its absorptive capacity (Cohen & Levinthal, 1990),

which can facilitate exploration by enabling the firm to incorporate external knowledge (Lavie &

Rosenkopf, 2006). R&D intensity was measured by dividing the firm’s R&D expenses by its total

revenue. We measured a firm’s corporate strategy function by counting its upper-echelon positions

related to strategy making, as documented in the Corporate Affiliations database. This function

12 Using alternative measures based on assets and number of employees did not affect our findings.

23

may proactively engage in developing and executing plans for exploration and exploitation (Menz

& Scheef, 2014). A firm’s solvency captures the financial resources available to support exploration

(Nohria & Gulati, 1996). We measured firm solvency with the ratio of cash to long-term debt

(Stettner & Lavie, 2014). Finally, a firm’s performance gap, i.e., the difference between the firm’s

actual performance and its performance aspiration, can affect its propensity to explore (Dothan &

Lavie, 2016; Greve, 2007). Performance aspiration was measured as a weighted linear combination

of the firm’s historical aspiration (return on assets in the preceding year) and its social aspiration

(median return on assets of publicly traded firms in the U.S. that operate in its four-digit SIC that

year),13 with weights determined using grid search. We then calculated the firm’s performance gap

as the difference between its performance and performance aspiration. We used a spline function

to model the firm’s reaction to performance feedback, splitting the performance gap into positive

(performance above aspiration) and negative (performance below aspiration) (Greve, 2003).

With respect to the alliance portfolio, we measured the alliance portfolio size by counting the

number of partners that formed alliances with the firm in the preceding five years. We also

measured the strategic significance of the alliance portfolio (Lavie, 2007) by calculating the

proportion of alliances that were classified as strategic in the SDC database out of the total number

of alliances that were formed during the past five years. Because the firm may intensify its

exploration efforts when competition becomes intensive (Jansen et al., 2005), we controlled for the

intensity of competition by counting the number of competitors that the firm encountered in the

past five years that attained a technological proximity score (Jaffe, 1986) higher than 0.25. Setting

this threshold at 0.25 generated a reasonable median competitor count of 524, with a range of 24

13 We considered alternative measures such as those based on return on sales, revenue growth, and average patent

counts, which produced consistent results, albeit less significant. This is in line with prior research that identifies ROA

as the most relevant and commonly used proxy in performance feedback studies (Greve, 2003).

24

to 1,807.14 All controls were lagged by one year relative to the dependent variable. We account for

remaining interfirm heterogeneity by including firm fixed effects. Inter-temporal trends were

controlled for with the exploration level in the previous year and the AR(1) parameter.

Analysis

We tested our hypotheses with a two-stage model specification to account for potential

selection bias in our sampling procedure and because not all firms form alliances. We used two

panel probit models to correspondingly estimate the selection to our sample and whether a firm

had formed alliances in the past five years (Heckman, 1979). In line with prior research, we

predicted the probability of being sampled based on lagged measures of the firm’s patenting

experience,15 age, size, R&D intensity, financial solvency, corporate strategy function, number of

industry peers in the same four-digit SIC, and market size proxied by the sum of industry revenues

in the firm’s primary four-digit SIC. We then estimated the probability of partnering based on

lagged measures of the firm’s partnering experience, age, size, R&D intensity, financial solvency,

corporate strategy function, number of competitors, and market size. Market size and the firm’s

experience in patenting or partnering served as the exclusion restriction variables. We calculated

the inverse Mills ratios (λ) based on the predicted values from the first-stage models and controlled

for them in the second-stage models. The λ parameter for partnering was included only in the

second-stage model estimating the effect of partners’ exploration.

The second-stage models served for testing our hypotheses. Given the high proportion of

observations with no partners (36.928%), and since we sought to include these observations in the

14 Defining competitors using 0.15, 0.5, and 0.75 Jaffe scores or four-digit SIC overlap yielded consistent findings. 15 Patenting experience was modeled with a memory decay function that preserves 90% of the value from the preceding

year over a 10-year period. A similar function served for modeling partnering experience based on alliances formed.

25

analysis of competitor exploration, we split our sample16: the analysis of competitor exploration

relied on the full sample of 180 firms with 2,337 observations, whereas the analysis of partner

exploration relied on a subsample of 153 firms with alliances and their 1,474 observations. We

conducted panel data analysis with firm fixed effects, since our theory focuses on within-firm

change in the level of exploration over time. Incorporating firm fixed effects also alleviates the

need to control for industry conditions such as dynamism and resource munificence. Additionally,

we accounted for autocorrelation of errors within cross-sections with an AR(1) parameter (Baltagi

& Wu, 1999). We estimated the models using maximum likelihood and evaluated model fit with

log likelihood ratio tests comparing each model to the baseline model. Maximum VIF values

exceeded the threshold level (Hair, Black, Babin, & Anderson, 2010) but can be attributed to the

multiple instances of the main effect, with no symptoms of multicollinearity observed.

Insert tables and remaining figures here

RESULTS

Tables 1a-c report descriptive statistics.17 Table 2 reports the results of the first-stage models,

indicating that the probability of selection increases with prior patenting experience and firm age,

but declines with firm size, firm solvency, and market size. This suggests that the propensity to

patent increases with absorptive capacity (Cohen & Levinthal, 1990), but declines as the firm

accumulates assets (Hill & Rothaermel, 2003) or when the market is sufficiently large to

accommodate established technologies (Katila & Shane, 2005). Similarly, the propensity to partner

16 In auxiliary analysis, we relied on a combined dataset to test the associations with partner exploration and competitor

exploration simultaneously, obtaining consistent results despite the severe loss of degrees of freedom. 17 Correlations between variables in the first-stage model (Tables 1a-b) were low, with the exception of the correlation

of firm size with partnering experience (r = 0.724) and patenting experience (r = 0.596) (Stuart, 2000). Still, no

symptoms of multicollinearity were observed (Table 2), with the maximum VIFs reaching 1.920 and 2.320 in the

selection and partnering models, below the threshold level (Hair et al., 2010). Correlations between variables in the

second-stage model (Table 1c) were low, with the exception of the size of the alliance portfolio and the firm’s size (r

= 0.639) (Lavie, 2007; Stuart, 2000). Besides the lambda parameter for selection, which was correlated with firm age

(r = -0.900), other high correlations relate to variables that were not included in the same model.

26

increases with firm size, R&D intensity (Veugelers, 1997), and prior partnering experience (Gulati,

1999), suggesting that resource-rich firms are attractive partners (Stuart, 2000). In turn, the firm’s

propensity to partner declines with the number of competitors, and its market size, as alliance

formation decreases as markets grow (Eisenhardt & Schoonhoven, 1996).

Tables 3a-b report the results of our second-stage models for testing the effects of the

exploration levels of partners and primary competitors. The baseline model (Model 1) reveals path

dependence in a firm’s tendency to explore, indicated by the effect of the firm’s exploration in the

preceding year (β = 0.559, p < 0.001; β = 0.411, p < 0.001) (Lavie & Rosenkopf, 2006). Exploration

declines as the firm matures (β = -1. 413, p < 0.001; β = -2.143, p < 0.001) (Greve, 2007) and when

its performance exceeds aspiration (β = -0.065, p < 0.05) (Dothan & Lavie, 2016; Greve, 2003),18

but increases with the number of competitors (β = 0.210, p < 0.001) (Deeds & Hill, 1996). All

effects hold in the combined model (Table 3c).

Model 2b (Table 3a) served for testing Hypothesis 1, revealing an inverted U-shaped

association between the firm’s exploration and the exploration level of alliance partners, as

evidenced by the positive main effect (β = 0.544, p < 0.001) and the negative effect of the squared

term of partner exploration (β = -0.510, p < 0.001). This model offers better fit to the data than

Model 2a, which tests a linear function (Δ2LL = 19.320, p < 0.001). The curvilinear pattern persists

in the full model (Table 3a, Model 5) and the combined model (Table 3c, Model 5). Figure 2a

depicts this curvilinear function, demonstrating that the inflection point falls within range (Min =

0.174, Max = 1.000) at a partner exploration level of 0.748, where the firm’s exploration level

reaches a maximum of 0.734, with the 95% confidence interval ranging between 0.713 and 0.756.

18 Although the difference in coefficients of the performance gap below versus above aspiration was insignificant for

competitor exploration, it became significant in the combined model (F(1, 1293) = 8.480, p = 0.004), where the firm’s

exploration increases below aspiration (β = 0.028, p < 0.1) and declines above aspiration (β = -0.047, p < 0.05).

27

The slopes around the inflection point are different from zero (positive slope = 0.423, p < 0.001;

negative slope = -0.186, p < 0.01), in support of Hypothesis 1a. Using the utest procedure in Stata,

we confirm the presence of an inverted U-shaped association (p = 0.003), with a Fieller confidence

interval ranging between 0.704 and 0.822 (Haans, Pieters, & He, 2016; Lind & Mehlum, 2010).

Similarly, our findings grant support to Hypothesis 1b (Model 2b, Table 3b), revealing an inverted

U-shaped association between the firm’s exploration and the exploration level of its primary

competitors, as evidenced by the positive main effect (β = 0.236, p < 0.01) and the negative effect

of the squared term of competitor exploration (β = -0.209, p < 0.05). This model offers better fit

than Model 2a (Δ2LL = 11.690, p < 0.01). This curvilinear association persists in the full model

(Table 3b, Model 5) and the combined model (Table 3c, Model 5). Figure 2b depicts this function,

showing that the inflection point falls within range (Min = 0.129, Max = 1.000) at a competitor

exploration level of 0.820, where the firm’s exploration level reaches a maximum of 0.722, with

the 95% confidence interval ranging between 0.704 and 0.779. The slopes around the inflection

point are different from zero (positive slope = 0.257, p < 0.01; negative slope = -0.067, p < 0.1).

The Stata utest procedure confirmed the inverted U-shaped association (p = 0.088), with a Fieller

confidence interval ranging between 0.729 and 1.106 (Haans et al., 2016; Lind & Mehlum, 2010),

thus offering marginal support for Hypothesis 1b. To ensure that the effects of partner and

competitor exploration are inverted U-shaped, we added their cubic terms in auxiliary analysis,

which revealed no additional inflection point within range, thus ruling out an S-shaped association.

As predicted by Hypothesis 2a, Model 3 (Table 3a, Figure 3a) reveals that the positive

association between a firm’s exploration and the exploration level of its partners becomes stronger

when firm-specific uncertainty increases (β = 0.221, p < 0.01). This effect persists in the full model

(β = 0.218, p < 0.01) (Table 3a, Model 5) and the combined model (Table 3c, Model 5). Per Model

3 (Table 3b, Figure 3b), firm-specific uncertainty reinforces the positive association between the

28

firm’s exploration and its competitors’ exploration (β = 0.142, p < 0.1). This effect persists in the

full model (β = 0.170, p < 0.01) (Table 3b, Model 5) and the combined model, in line with

Hypothesis 2b. Model 4 (Table 3a, Figure 4a) furnishes support for Hypothesis 3a, revealing that

variation in partner exploration levels weakens the positive association between firm exploration

and partner exploration (β = -0.351, p < 0.001). This effect persists in the full model (β = -0.343, p

< 0.001) (Table 3a, Model 5) and the combined model (Table 3c, Model 5). Similarly, Model 4

(Table 3b, Figure 4b) offers support for Hypothesis 3b, indicating that variation in competitors’

exploration levels weakens the positive association between the firm’s exploration and its

competitors’ exploration (β = -0.210, p < 0.01). This effect persists in the full model (β = -0.185, p

< 0.01) (Table 3b, Model 5) and the combined model (Table 3c, Model 5).

Model 5 (Table 3a) offers no support for Hypothesis 4a about the moderating effect of

technological proximity to partners. Counter to Hypothesis 4b, Model 5 (Table 3b, Figure 5b)

reveals that technological proximity to competitors weakens the positive association between the

firm’s exploration and that of its competitors (β = -0.261, p < 0.01). These findings indicate that

technological proximity does not motivate a firm to more closely follow the exploration levels of

its partners and competitors. One explanation is that we restricted our sample to the five most

proximate competitors, which limits the range of the moderator. Another possibility is that since

the firm operates in knowledge domains similar to those of its competitors, it is already aware of

opportunities in these domains, and thus can rely less on its competitors for cues on the desirable

exploration level. Although this explanation may apply also for partners, in the case of competitors,

the firm’s divergence may also be tied to its effort to differentiate itself from proximate competitors

and maintain a distinctive industry position vis-à-vis these competitors. Differentiation enables the

firm to avoid competitive pressure (Deephouse, 1999) and delineate uncontested markets.

Robustness Tests

29

We considered alternative model specifications and measures. First, we ran a Tobit model,

which produced consistent findings, with the exception of Hypothesis 2b. We retained our reported

models because they account for firm fixed effects and correct for autocorrelation. Second, in line

with the notion of trait imitation (Haunschild & Miner, 1997), we considered whether the firm

follows industry leaders rather than primary competitors and partners (Massini et al., 2005). We

replaced our independent variable with the lagged exploration level of the top ten percent

performers in the firm’s four-digit SIC, but found no support for this hypothesis. Third, by

controlling for both the linear and quadratic terms of the lagged exploration variable, we ruled out

the possibility that our findings are driven by firms’ independent efforts to strive toward an

intermediate level of exploration. The desirable balance point varies across firms and is difficult to

discern, so a firm is more likely to follow the exploration levels of alters. Fourth, we created a

matched sample of hypothetical partners using a Mahalanobis distance calculation (e.g., Aguinis,

Gottfredson, & Joo, 2013). Results of t-tests revealed higher differences in the exploration levels

of the firm and its partner relative to the differences with the alternative matched partners, thus

ruling out selection of partners with similar exploration levels. When we control for the absolute

difference in firm/partner exploration levels in the prior year, its effect was insignificant, with our

findings remaining intact, suggesting no selection bias.

Fifth, we considered an exploration measure based on knowledge search scope (Dothan &

Lavie, 2016; Katila & Ahuja, 2002) that captures the proportion of new patent citations that the

firm did not cite in the previous five years. The corresponding findings offer no support for our

hypotheses. This suggests that firms follow the exploratory behavior of their partners and

competitors when entering new knowledge domains rather than when incorporating particular

knowledge elements, which in turn are more difficult to observe, and thus do not support vicarious

learning. We next considered exploration measures based on the inverse of Fleming’s (2001)

30

measures of component, combination, and cumulative familiarity, which capture the extent to

which the firm relies on recent and frequently used patent classes. Although we find support for

Hypotheses 1 and 2 in the competitor exploration models, these alternative measures center on

knowledge that is new to the world rather than to the firm, and their complexity does not support

vicarious learning. For similar reasons, an alternative measure based on Eggers and Kaul (2018)

yielded no significant findings. The firm is unlikely to observe class-to-class citation patterns.

Sixth, we tested whether firms adjust their exploration level based on the number of partners

and competitors with high-level exploration rather than based on their exploration level, but these

variables had no significant effect on firm exploration. Seventh, we verified that convergence of

exploration levels cannot be ascribed to the pursuit of similar technological opportunities, finding

only 2.95% overlap in the new patent classes entered by the firm and its partners. A control for this

overlap was insignificant, while our reported findings remained intact. Similar results were

obtained for primary competitors (9.36% overlap) and when measuring overlap with a one-year

lag. Eighth, we considered an alternative definition of competitors based on firms’ competitor lists

in annual reports, and found consistent results despite severe loss of degrees of freedom ascribed

to 73.341% missing values, which prompted us to retain our original definition of competitors.19

Ninth, we tested a dynamic panel model using the Arellano-Bond model specification, which

produced consistent findings with the exception of Hypothesis 2a. Nevertheless, our reported

findings were insensitive to the exclusion of the lagged dependent variable, thus rendering the

Arellano-Bond model estimates redundant. We conclude that our reported model is preferred

19 The missing values occur because the EDGAR database includes SEC filings only since 1994 and the SEC does not

require firms to list competitors, while their voluntary statements can be unsystematic, incomplete, and less reliable

(e.g., Botosan & Stanford, 2005; Elshandidy, Fraser, & Hussainey, 2013). Still, the alternative measure of competitor

exploration was correlated (r = 0.246, p < 0.001) with our reported measure, which offers a more complete list of

competitors, including those that have yet to introduce competing products but operate in relevant knowledge domains.

31

because of the large number of time points per firm (12.983 observations on average), which

restricts potential dynamic panel bias (Roodman, 2009). The number of moment conditions

required by the Arellano-Bond GMM estimator yields weak instruments (Blundell & Bond, 1998),

and the estimates generated by this alternative model can be unstable (Greene, 2012: 448).

Tenth, we tested whether our moderators affect the quadratic functions of partner and

competitor exploration; however, corresponding models exhibited symptoms of multicollinearity,

so we could not interpret their findings. We also tested the effects of our moderators on the positive

slope of the spline function relating to the inverted U-shaped effect of partner and competitor

exploration. We found consistent results, with the exception of Hypothesis 4b, although per our

theory, the moderators apply at any level of exploration. Eleventh, we tested for the moderating

effects of firm size, corporate strategy function, and market size, which turned out insignificant

without affecting our findings. For competitor exploration, we also tested the moderating effect of

the number of competitors, which weakens the effect of competitor exploration without affecting

our reported findings. For partner exploration, we tested the moderating effect of the strategic

significance of the alliance portfolio, which turned out negative but left our reported findings

virtually intact. Twelfth, we considered a forward-looking measure of uncertainty (Toh & Kim,

2013) capturing the implied volatility of the firm’s one-month expiration of a European-style, at-

the-money call option on the first trading day of the year. However, we encountered 78.862%

missing values in the OptionMetrics database, which provides data only from 1996 and with many

publicly traded firms not meeting the requirements for options trading.

Thirteenth, we tested for reversed causality, but the corresponding effects were insignificant

and the model fit was significantly lower for both partners and competitors. Indeed, it is unlikely

that all of the firm’s partners and competitors follow its behavior unless it is the undisputable

market leader. Fourteenth, we split our sample into subsamples that include or exclude R&D

32

alliances, strategic alliances, and coopetitors, finding support for our hypotheses in most

subsamples for competitor exploration, but weaker support for partner exploration, probably

because of the smaller subsample sizes. Fifteenth, we replaced the five-year window with three-

and seven-year windows for identifying competitors and partners. The analysis using the seven-

year window yielded support for Hypotheses 1a-b and 3a, with consistent findings for Hypothesis

4b. The analysis relying on the three-year window granted support for Hypotheses 1a-b and 3b.

These analyses reaffirm our five-year window specification, which is most suitable for the

electronics industry. Sixteenth, we considered endogeneity in the choice of exploration alliances

versus exploitation alliances by predicting whether the firm had at least one upstream alliance in

the first-stage model, but our results remained virtually unchanged. Finally, our findings were

insensitive to outliers.20 Overall, our tests reaffirm our measures and model specification.

DISCUSSION

Our study promotes research on the antecedents of exploration by suggesting that firms’

tendencies to explore are interdependent and subject to various boundary conditions that restrict

convergence with the exploration levels of alliance partners and competitors. Hence, our study goes

beyond prior research that showed how firms’ exploration tendencies are uniformly shaped by

exogenous industry conditions or independently driven by firms’ organizational characteristics.

Unlike some prior research that relates exploration to the sheer number of competitors (e.g., Skilton

& Bernardes, 2015) or alliance partners (e.g., Lavie & Drori, 2012; Rothaermel & Deeds, 2004),

we consider these alters as reference groups for firms’ exploration efforts. We conjecture that firms

seek to learn not only from their partners’ and competitors’ knowledge (Mowery et al., 1996), but

20 We tested sensitivity to outliers using various approaches (Aguinis et al., 2013; Billor, Hadi, & Velleman, 2000;

Upton & Cook, 1996), and when we dropped outliers, the results remained virtually unchanged, and even improved.

33

also from these alters’ exploratory behaviors. Acknowledging this interdependence vis-à-vis

alliance partners and competitors is essential for understanding the antecedents of exploration and

for explaining heterogeneity in firms’ exploration tendencies.

In this study, we advance a vicarious learning theory and identify boundary conditions that

explain how partners and competitors shape a firm’s exploration tendency. At low exploration

levels, the firm increases its tendency to explore when either its partners or competitors increase

their exploration levels. Convergence at that level is ascribed to imitation and legitimation (e.g.,

Haunschild & Miner, 1997; Lieberman & Asaba, 2006). However, as these alters’ exploration

tendencies become excessive, the firm diverges from these tendencies and reverts to exploitation.

This divergence is ascribed to the perceived risk of excessive exploration and to the firm’s efforts

to leverage external exploration efforts of partners, while restricting its own exploration tendency

(Stettner & Lavie, 2014). In turn, its divergence from competitors’ exploration is attributed to its

differentiation efforts. We contend that specialization in a relatively narrow set of knowledge

domains provides the impetus for both division of labor with partners and differentiation vis-à-vis

competitors. Nevertheless, the decline in the firm’s exploration is stronger for excessive partner

exploration than for excessive competitor exploration. This is in line with research suggesting that

recurrent cycles of imitation and innovation reinforce status quo with rivals (Giachetti, Lampel, &

Pira, 2017). Still, our findings stand in contrast to optimal distinctiveness theory that implies that

firms would strive to reconcile the tension between conformity (convergence) and differentiation

(divergence) by reaching an intermediate level of novelty (Zhao, Fisher, Lounsbury, & Miller,

2017) irrespective of the observed level of exploration. Instead, we find that firms either converge

or diverge, depending on the level of exploration exhibited by their partners and competitors.

Our study further contributes by showing how convergence with the exploration tendencies of

partners and competitors is subject to boundary conditions that influence the firm’s motivation and

34

ability to learn and follow the typical exploration levels in its reference groups. In particular, we

show that as firm-specific uncertainty increases, the firm tends to better align its exploration

tendency with the exploration levels of its partners and competitors. However, the firm’s abilities

to learn the typical exploration pattern, imitate it, and gain legitimacy depend on the coherence of

that pattern (Rhee et al., 2006). When alters pursue diverse exploration levels, this inconsistent

pattern limits the firm’s ability to systematically react to increases in their exploration levels.

Finally, counter to expectations, we reveal that the motivation and ability to converge with the

exploration exhibited by partners and competitors do not increase with their proximity to the firm

(e.g., Rosenkopf & Almeida, 2003). The fact that partners and competitors develop expertise in

knowledge domains similar to those of the firm suggests that the firm has already learned about

opportunities in related domains, so need not rely on these alters’ exploratory behavior as a cue for

its desirable exploration tendency. Rather, the firm’s efforts to differentiate itself from proximate

competitors outweigh the ease of convergence, and thus lead to divergence. By revealing several

boundary conditions for convergence with the exploration levels of partners and competitors, we

complement research that has shown how environmental conditions such as market concentration

reinforce interdependence in firms’ innovation strategies (Turner, Mitchell, & Bettis, 2010). We

claim that firms do not respond merely to uniform industry conditions, but to idiosyncratic

exploration tendencies in their particular cooperation and competition networks.

Our main contribution is in enhancing understanding of the antecedents of exploration (Lavie

et al., 2010). We reveal how a firm’s exploration tendencies converge with the typical exploration

level of alters in its main reference groups, namely alliance partners and competitors, depending

on the nature of the firm’s relations with them. Convergence is explained by vicarious learning that

is driven by imitation and legitimation. However, the perceived risk of excessive exploration

restricts convergence. Finally, we claim that specialization reinforces divergence, as the firm

35

divides labor with partners and improves its position vis-à-vis competitors. This enables the firm

to leverage its partners’ complementary skills while maintaining competitive parity with its rivals.

Our study also contributes to the literature on exploration and exploitation by identifying

conditions that shape firms’ interdependent exploration efforts. When considering the desirable

balance between exploration and exploitation, firms observe alters and consider adopting their

behavior. A firm’s partners and competitors serve as relevant reference groups, but the extent of

convergence with their exploration tendencies depends on firm-specific uncertainty, the coherence

of their behavior, and their technological proximity to the firm. Thus, we extend research on

environmental antecedents (e.g., Sidhu et al., 2004) by showing that exploration is shaped by

conditions that vary across firms with unique portfolios of interfirm relations. We also complement

research on learning from performance feedback, which shows how a firm intensifies exploration

when its performance falls below aspiration (Chen, 2008; Dothan & Lavie, 2016; Greve, 2007), by

revealing that the firm’s reference groups play a more profound role, not only in shaping the firm’s

performance aspiration but also in offering a benchmark for the desirable level of exploration.

Moreover, our study contributes to research on vicarious learning, imitation, and legitimation

(e.g., Lieberman & Asaba, 2006; Suddaby et al., 2017; Terlaak & Gong, 2008) by demonstrating

that when a behavior is risky and its outcomes are unforeseen, firms deviate from the paradigm of

convergence. Specifically, they follow the population average rather than a small group of leaders

(Massini et al., 2005) and engage in frequency imitation rather than in trait imitation or outcome

imitation (Haunschild & Miner, 1997). Lastly, as alters’ exploration further increases, we expect

perceived risk to mitigate imitation and legitimation, while specialization offsets them and fosters

divergence of behaviors. Hence, whereas prior research suggested that perceived risk can lead to

convergence of behaviors (e.g., Lieberman & Asaba, 2006), we show that it results in divergence

when the risk is inherent to the imitated behavior as opposed to the targeted market or technology

36

(Srinivasan et al., 2007). We further identify boundary conditions that restrict convergence of

behaviors, namely firm-specific uncertainty, variance in alters’ behavior, and proximity. We expect

these conditions to influence vicarious learning irrespective of the type of observed behavior.

Finally, we offer managerial implications for firms seeking to balance exploration and

exploitation. We advise firms to consider departing from industry conventions and adjusting their

exploration tendencies in line with the typical exploratory behavior of their specific set of partners

and competitors. In particular, under uncertainty, a firm’s decision to enter new knowledge

domains should take into account the idiosyncratic competitive position and the unique

configuration of its alliance portfolio, rather than simply be based on industry trends dictated by

environmental conditions. We found that learning from primary competitors can complement

learning from partners, irrespective of performance feedback that is limited in the case of

exploration. Given uncertainty about the prospects of exploration, managers often opt for

mimicking the behavior of the reference group, expecting to gain legitimacy if not enhanced

performance. However, managers should not blindly adopt the observed exploration level prevalent

in the firm’s reference groups, and even depart from it when the firm’s close competitors engage

in excessive exploration or their exploration pattern is incoherent. In turn, this enables increased

specialization in the firm’s knowledge domain. Understanding the pathways of convergence and

divergence can help managers foresee the exploration tendencies of their partners and competitors,

which in turn can influence the firm’s own exploration tendencies. Still, given the disparity between

observed behavioral patterns and prescriptive advice, future research should study the performance

implications of convergence with the exploration levels exhibited by partners and competitors.

Although our study advances research on the antecedents of exploration, it faces several

limitations that pave directions for future research. Conceptually, one may study the performance

implications of convergence with the exploration tendencies of alliance partners and competitors.

37

Additionally, future research may identify additional reference groups based on various corporate

relations such as corporate venture capital investors, customers, and suppliers that may influence a

firm’s tendency to explore (Sidhu et al., 2007). Furthermore, given the implications of variation in

competitors’ exploration levels, scholars should study the firm’s ability to learn from a reference

group that exhibits incoherent tendencies and decide which partner or competitor to follow. It is

possible that firms pay more attention to partners with whom they engage in more substantial

collaborative relations or to certain types of alliances, such as joint ventures versus non-equity

alliances. Thus, more insights can be gained by studying when, why, how, and whom a firm follows

or benchmarks against when converging with a typical pattern of exploration. Additionally, as we

furnish no direct evidence of our proposed mechanisms, future research may operationalize and

test the effects of imitation, legitimacy, perceived risk, balance across modes, and differentiation.

Such research can indicate, for instance, whether convergence is driven mostly by imitation or

legitimacy (Zajac & Westphal, 2004). Moreover, since we study exploration in the knowledge

domain, future research may generalize our findings to other domains, such as business

diversification and internationalization (Wilden et al., 2018). Along the same lines, we identified a

firm’s competitors based on their knowledge similarity, so future research may define competitors

based on other types of resource similarity, market communality, or perceived rivalry (Chen, 1996).

Furthermore, our measure of exploration captures the increase in the diversity of the firm’s patent

classes. Future research may consider more complex measures of exploration (e.g., Eggers and

Kaul 2018; Fleming, 2001; Katila & Ahuja, 2002) that capture novelty to the world rather than to

the firm, and that pose challenges for vicarious learning. Finally, given our focus on the electronics

industry, it is worth testing the generalizability of our findings to other industries. Despite its

limitations, our study sheds new light on previously overlooked antecedents of exploration and

enhances our understanding of this important organizational phenomenon.

38

REFERENCES

Abrahamson, E. 1996. Management fashion. Academy of Management Review, 21: 254–285.

Aguinis, H., Gottfredson, R. K., & Joo, H. 2013. Best-practice recommendations for defining,

identifying, and handling outliers. Organizational Research Methods, 16(2): 270–301.

Ahuja, G., & Lampert, C. M. 2001. Entrepreneurship in the large corporation: A longitudinal study

of how established firms create breakthrough inventions. Strategic Management Journal, 22:

521–543.

Alexiev, A. S., Jansen, J. J., Van den Bosch, F. A., & Volberda, H. W. 2010. Top management

team advice seeking and exploratory innovation: The moderating role of TMT heterogeneity.

Journal of Management Studies, 47: 1343–1364.

Anand, J., Mesquita, L. F., & Vassolo, R. S. 2009. The dynamics of multimarket competition in

exploration and exploitation activities. Academy of Management Journal, 52: 802–821.

Argyres, N. S., & Silverman, B. S. 2004. R&D, organization structure, and the development of

corporate technological knowledge. Strategic Management Journal, 25: 929–958.

Baltagi, B. H., & Wu, P. X. 1999. Unequally spaced panel data regressions with AR(1)

disturbances. Econometric Theory, 15: 814–823.

Barreto, I., & Baden-Fuller, C. 2006. To conform or to perform? Mimetic behaviour, legitimacy-

based groups and performance consequences. Journal of Management Studies, 43(7): 1559–

1581.

Baum, J. A. C., Li, S. X., & Usher, J. M. 2000. Making the next move: How experiential and

vicarious learning shape the locations of chains’ acquisitions. Administrative Science Quarterly,

45: 766–801.

Beckman, C. M., Haunschild, P. R., & Phillips, D. J. 2004. Friends or strangers? Firm-specific

uncertainty, market uncertainty, and network partner selection. Organization Science, 15: 259–

275.

Billor, N., Hadi, A. S., & Velleman, P. F. 2000. BACON: Blocked adaptive computationally

efficient outlier nominators. Computational Statistics & Data Analysis, 34: 279–298.

Blundell, R., & Bond, S. 1998. Initial conditions and moment restrictions in dynamic panel data

models. Journal of Econometrics, 87: 115–143.

Botosan, C., & Stanford, M. 2005. Managers’ motives to withhold segment disclosures and the

effect of SFAS no. 131 on analysts’ information environment. The Accounting Review, 80(3):

751–771.

Brix, J. 2017. Exploring knowledge creation processes as a source of organizational learning: A

longitudinal case study of a public innovation project. Scandinavian Journal of Management,

33: 113–127.

Chen, M.-J. 1996. Competitor analysis and interfirm rivalry: Toward a theoretical integration.

Academy of Management Review, 21: 100–134.

Chen, W. R. 2008. Determinants of firms’ backward- and forward-looking R&D search behavior.

Organization Science, 19(4): 609–622.

Clark, B., & Montgomery, D. 1999. Managerial identification of competitors. Journal of

Marketing, 63(3): 67–83.

Cockburn, I. M., & Griliches, Z. 1987. Industry effects and appropriability measures in the stock

market’s valuation of R&D and patents. NBER Working Paper No. 2465.

Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: A new perspective on learning and

innovation. Administrative Science Quarterly, 35: 128–152.

39

Conti, R., 2014. Do non-competition agreements lead firms to pursue risky R&D

projects? Strategic Management Journal, 35(8): 1230–1248.

Cuypers, I. R., & Martin, X. 2010. What makes and what does not make a real option? A study of

equity shares in international joint ventures. Journal of International Business Studies, 41: 47–

69.

Deeds, D. L., & Hill, C. W. L. 1996. Strategic alliances and the rate of new product development:

An empirical study of entrepreneurial biotechnology firms. Journal of Business Venturing, 11:

41–55.

Deephouse, D. L. 1996. Does isomorphism legitimate? Academy of Management Journal, 39:

1024–1039.

Deephouse, D. L. 1999. To be different, or to be the same? It’s a question (and theory) of strategic

balance. Strategic Management Journal, 20: 147–166.

DiMaggio, P., & Powell, W. W. 1983. The iron cage revisited: Collective rationality and

institutional isomorphism in organizational fields. American Sociological Review, 48: 147–160.

Dobrev, S. D. 2007. Competing in the looking-glass market: Imitation, resources, and crowding.

Strategic Management Journal, 28: 1267–1289.

Dothan, A., & Lavie, D. 2016. Resource reconfiguration: Learning from performance feedback.

In T. B. Folta, C. E. Helfat, & S. Karim (Eds.), Resource redeployment and corporate strategy

(Advances in Strategic Management 35): 319–369. Bingley, UK: Emerald Group Publishing

Limited.

Eggers, J. P., & Kaul, A. 2018. Motivation and ability? A behavioral perspective on the pursuit of

radical invention in multi-technology incumbents. Academy of Management Journal, 61(1): 67–

93.

Eisenhardt, K. M., & Schoonhoven, C. B. 1996. Resource-based view of strategic alliance

formation: Strategic and social effects in entrepreneurial firms. Organization Science, 7: 136–

150.

Elshandidy, T., Fraser, I., & Hussainey, K. 2013. Aggregated, voluntary, and mandatory risk

disclosure incentives: Evidence from UK FTSE all-share companies. International Review of

Financial Analysis, 30: 320–333.

Fiegenbaum, A., & Thomas, H. 1995. Strategic groups as reference groups: Theory, modeling and

empirical examination of industry and competitive strategy. Strategic Management Journal, 16:

461–476.

Fleming, L. 2001. Recombinant uncertainty in technological search. Management Science, 47(1):

117–132.

Gallini, N. T. 2002. The economics of patents: Lessons from recent U.S. patent reform. Journal

of Economic Perspectives, 16(2): 131–154.

Ganzaroli, A., De Noni, I., Orsi, L., & Belussi, F. 2016. The combined effect of technological

relatedness and knowledge utilization on explorative and exploitative invention performance

post-M&A. European Journal of Innovation Management, 19(2): 167–188.

Giachetti, C., Lampel, J. & Pira, S.L., 2017. Red queen competitive imitation in the UK mobile

phone industry. Academy of Management Journal, 60(5): 1882–1914.

Giachetti, C., & Lanzolla, G. 2016. Product technology imitation over the product diffusion cycle:

Which companies and product innovations do competitors imitate more quickly? Long Range

Planning, 49: 250–264.

Gilsing, V., Nooteboom, B., Vanhaverbeke, W., Duysters, G., & van den Oord, A. 2008. Network

embeddedness and the exploration of novel technologies: Technological distance, betweenness

centrality and density. Research Policy, 37(10): 1717–1731.

40

Gimeno, J., Hoskisson, R. E., Beal, B. D., & Wan, W. P. 2005. Explaining the clustering of

international expansion moves: A critical test in the US telecommunications industry. Academy

of Management Journal, 48: 297–319.

Gioia, D. A., & Manz, C. C. 1985. Linking cognition and behavior: A script processing

interpretation of vicarious learning. Academy of Management Review, 10: 527–539.

Greene, W. H. 2012. Econometric analysis (7th ed., international ed.). Boston: Pearson.

Greve, H. R. 1996. Patterns of competition: The diffusion of a market position in radio

broadcasting. Administrative Science Quarterly, 41: 29–60.

Greve, H. R. 1998. Managerial cognition and the mimetic adoption of market positions: What you

see is what you do. Strategic Management Journal, 19: 967–988.

Greve, H. R. 2003. Organizational learning from performance feedback: A behavioral

perspective on innovation and change. Cambridge: Cambridge University Press.

Greve, H. R. 2005. Interorganizational learning and heterogeneous social structure. Organization

Studies, 26: 1025–1047.

Greve, H. R. 2007. Exploration and exploitation in product innovation. Industrial and Corporate

Change, 16: 945–975.

Greve, H. R. 2011. Fast and expensive: The diffusion of a disappointing innovation. Strategic

Management Journal, 32(9): 949–968.

Griliches, Z. 1998. Patent statistics as economic indicators: A survey. In Z. Griliches (Ed.), R&D

and productivity: The econometric evidence: 287–343. Chicago: University of Chicago Press.

Grimpe, C., & Hussinger, K. 2014. Resource complementarity and value capture in firm

acquisitions: The role of intellectual property rights. Strategic Management Journal, 35(12):

1762–1780.

Guan, J., & Liu, N. 2016. Exploitative and exploratory innovations in knowledge network and

collaboration network: A patent analysis in the technological field of nano-energy. Research

Policy, 45(1): 97–112.

Gulati, R. 1999. Network location and learning: The influence of network resources and firm

capabilities on alliance formation. Strategic Management Journal, 20: 397–420.

Gulati, R., Lavie, D., & Singh, H. 2009. The nature of partnering experience and the gains from

alliances. Strategic Management Journal, 30: 1213–1233.

Gupta, A., & Misangyi, V. F. 2018. Follow the leader (or not): The influence of peer CEOs’

characteristics on interorganizational imitation. Strategic Management Journal, 39(5): 1437–

1472.

Hair, J., Jr., Black, W., Babin, B., & Anderson, R. 2010. Multivariate data analysis (7th ed.).

Upper Saddle River, NJ: Prentice Hall.

Hall, B. H., Jaffe, A. B., & Trajtenberg, M. 2001. The NBER patent citation data file: Lessons,

insights and methodological tools. NBER Working Paper 8498.

Hall, B. H., & Ziedonis, R. H. 2001. The patent paradox revisited: An empirical study of patenting

in the US semiconductor industry, 1979–1995. RAND Journal of Economics, 32: 101–128.

Haans, R. F., Pieters, C., & He, Z.-L. 2016. Thinking about U: Theorizing and testing U- and

inverted U-shaped relationships in strategy research. Strategic Management Journal, 37(7):

1177–1195.

Haunschild, P. R., & Miner, A. S. 1997. Modes of interorganizational imitation: The effects of

outcome salience and uncertainty. Administrative Science Quarterly, 42: 472–500.

Haveman, H. A. 1993. Follow the leader: Mimetic isomorphism and entry into new markets.

Administrative Science Quarterly, 38: 593–627.

Heckman, J. 1979. Sample selection bias as a specification error. Econometrica, 47: 153–161.

41

Henisz, W. J., & Delios, A. 2001. Uncertainty, imitation, and plant location: Japanese multinational

corporations, 1990–1996. Administrative Science Quarterly, 46: 443–475.

Hess, A. M., & Rothaermel, F. T. 2011. When are assets complementary? Star scientists, strategic

alliances, and innovation in the pharmaceutical industry. Strategic Management Journal, 32:

895–909.

Hill, C. W. L., & Rothaermel, F. T. 2003. The performance of incumbent firms in the face of radical

technological innovation. Academy of Management Review, 28: 257–274.

Hsieh, K.-Y., Tsai, W., & Chen, M.-J. 2015. If they can do it, why not us? Competitors as reference

points for justifying escalation of commitment. Academy of Management Journal, 58: 38–58.

Hu, S., Blettner, D., & Bettis, R. A. 2011. Adaptive aspirations: Performance consequences of risk

preferences at extremes and alternative reference groups. Strategic Management

Journal, 32(13): 1426–1436.

Huber, G. P. 1991. Organizational learning: The contributing processes and the

literatures. Organization Science, 2(1): 88–115.

Jaffe, A. B. 1986. Technological opportunity and spillovers of R&D: Evidence from firms’ patents,

profits and market value. American Economic Review, 76: 984–999.

Jaffe, A. B., Trajtenberg, M., & Henderson, R. 1993. Geographic localization of knowledge

spillovers as evidenced by patent citations. Quarterly Journal of Economics, 108: 577–598.

Jansen, J. J., Kostopoulos, K. C., Mihalache, O. R., & Papalexandris, A. 2016. A socio-

psychological perspective on team ambidexterity: The contingency role of supportive leadership

behaviours. Journal of Management Studies, 53: 939–965.

Jansen, J. J., Van Den Bosch, F. A., & Volberda, H. W. 2006. Exploratory innovation, exploitative

innovation, and performance: Effects of organizational antecedents and environmental

moderators. Management Science, 52: 1661–1674.

Jansen, J. J., Vera, D., & Crossan, M. 2009. Strategic leadership for exploration and exploitation:

The moderating role of environmental dynamism. Leadership Quarterly, 20: 5–18.

Jansen, J. J., Volberda, H. W., & Van Den Bosch, F. A. 2005. Exploratory innovation, exploitative

innovation, and ambidexterity: The impact of environmental and organizational antecedents.

Schmalenbach Business Review, 57: 351–363.

Kang, E., & Uhlenbruck, K. 2006. A process framework of entrepreneurship: From exploration, to

exploitation, to exit. Academy of Entrepreneurship Journal, 12: 47–71.

Katila, R., & Ahuja, G. 2002. Something old, something new: A longitudinal study of search

behavior and new product introduction. Academy of Management Journal, 45: 1183–1194.

Katila, R., & Shane, S. 2005. When does lack of resources make new firms innovative? Academy

of Management Journal, 48: 814–829.

Khanagha, S., Volberda, H., & Oshri, I. 2017. Customer co-creation and exploration of emerging

technologies: The mediating role of managerial attention and initiatives. Long Range Planning,

50: 221–242.

Khanna, T., Gulati, R., & Nohria, N. 1998. The dynamics of learning alliances: Competition,

cooperation, and relative scope. Strategic Management Journal, 19: 193–210.

Kilduff, G. J., Elfenbein, H. A., & Staw, B. M. 2010. The psychology of rivalry: A relationally

dependent analysis of competition. Academy of Management Journal, 53: 943–969.

Kim, J.-Y. J., & Miner, A. S. 2007. Vicarious learning from the failures and near-failures of others:

Evidence from the US commercial banking industry. Academy of Management Journal, 50:

687–714.

42

Kim, T., & Rhee, M. 2009. Exploration and exploitation: Internal variety and environmental

dynamism. Strategic Organization, 7: 11–41.

Kondra, A. Z., & Hinings, C. R. 1998. Organizational diversity and change in institutional theory.

Organization Studies, 19: 743–767.

Kraatz, M. S. 1998. Learning by association? Interorganizational networks and adaptation to

environmental change. Academy of Management Journal, 41: 621–643.

Lavie, D. 2007. Alliance portfolios and firm performance: A study of value creation and

appropriation in the US software industry. Strategic Management Journal, 28: 1187–1212.

Lavie, D. 2009. Capturing value from alliance portfolios. Organizational Dynamics, 38: 26–36.

Lavie, D., & Drori, I. 2012. Collaborating for knowledge creation and application: The case of

nanotechnology research programs. Organization Science, 23: 704–724.

Lavie, D., & Rosenkopf, L. 2006. Balancing exploration and exploitation in alliance formation.

Academy of Management Journal, 49: 797–818.

Lavie, D., Stettner, U., & Tushman, M. L. 2010. Exploration and exploitation within and across

organizations. Academy of Management Annals, 4: 109–155.

Levinthal, D. A., & March, J. G. 1993. The myopia of learning. Strategic Management Journal,

14: 95–112.

Li, Q., Maggitti, P. G., Smith, K. G., Tesluk, P. E., & Katila, R. 2013. Top management attention

to innovation: The role of search selection and intensity in new product introductions. Academy

of Management Journal, 56: 893–916.

Lieberman, M. B., & Asaba, S. 2006. Why do firms imitate each other? Academy of Management

Review, 31: 366–385.

Lind, J., & Mehlum, H, 2010. With or without U? The appropriate test for a U shaped

relationship. Oxford Bulletin of Economics and Statistics, 72(1): 109–118.

Maitlis, S., & Christianson, M. 2014. Sensemaking in organizations: Taking stock and moving

forward. Academy of Management Annals, 8: 57–125.

Makri, M., Hitt, M. A., & Lane, P. J. 2010. Complementary technologies, knowledge relatedness,

and invention outcomes in high technology mergers and acquisitions. Strategic Management

Journal, 31: 602–628.

March, J. G. 1991. Exploration and exploitation in organizational learning. Organization Science,

2: 71–87.

Massini, S., Lewin, A. Y., & Greve, H. R. 2005. Innovators and imitators: Organizational reference

groups and adoption of organizational routines. Research Policy, 34: 1550–1569.

Menz, M., & Scheef, C. 2014. Chief strategy officers: Contingency analysis of their presence in

top management teams. Strategic Management Journal, 35(3): 461–471.

McMullen, J. S., & Shepherd, D. A. 2006. Entrepreneurial action and the role of uncertainty in the

theory of the entrepreneur. Academy of Management Review, 31: 132–152.

McPherson, M., Smith-Lovin, L., & Cook, J. M. 2001. Birds of a feather: Homophily in social

networks. Annual Review of Sociology, 27(1): 415–444.

Miner, A. S., & Mezias, S. J. 1996. Ugly duckling no more: Pasts and futures of organizational

learning research. Organization Science, 7: 88–99.

Mowery, D. C., Oxley, J. E., & Silverman, B. S. 1996. Strategic alliances and interfirm knowledge

transfer. Strategic Management Journal, 17: 77–91.

Nelson R. R., & Winter, S. 1982. An evolutionary theory of economic change. Cambridge:

Belknap/Harvard University Press.

Nohria, N., & Gulati, R. 1996. Is slack good or bad for innovation? Academy of Management

Journal, 39: 1245–1264.

43

O’Neill, H. M., Pouder, R. W., & Buchholtz, A. K. 1998. Patterns in the diffusion of strategies

across organizations: Insights from the innovation diffusion literature. Academy of Management

Review, 23: 98–114.

Phene, A., Fladmoe-Lindquist, K., & Marsh, L. 2006. Breakthrough innovations in the U.S.

biotechnology industry: The effects of technological space and geographic origin. Strategic

Management Journal, 27: 369–388.

Polidoro, F., Jr., 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(1): 203–223.

Porac, J. F., & Thomas, H. 1990. Taxonomic mental models in competitor definition. Academy of

Management Review, 15: 224–240.

Porac, J. F., Thomas, H., & Baden-Fuller, C. 1989. Competitive groups as cognitive communities:

The case of Scottish knitwear manufacturers. Journal of Management Studies, 26: 397–416.

Powell, W. W., Koput, K. W., & Smith-Doerr, L. 1996. Interorganizational collaboration and the

locus of innovation: Networks of learning in biotechnology. Administrative Science Quarterly,

41: 116–145.

Puranam, P., & Srikanth, K. 2007. What they know vs. what they do: How acquirers leverage

technology acquisitions. Strategic Management Journal, 28(8): 805–825.

Reed, R., & DeFilippi, R. J. 1990. Causal ambiguity, barriers to imitation and sustainable

competitive advantage. Academy of Management Review, 15(1): 88–102.

Rhee, M., Kim, Y.-C., & Han, J. 2006. Confidence in imitation: Niche-width strategy in the UK

automobile industry. Management Science, 52(4): 501–513.

Roodman, D. 2009. How to do xtabond2: An introduction to difference and system GMM in Stata.

Stata Journal, 9: 86–136.

Rosenkopf, L., & Abrahamson, E. 1999. Modeling reputational and informational influences in

threshold models of bandwagon innovation diffusion. Computational & Mathematical

Organization Theory, 5: 361–384.

Rosenkopf, L., & Almeida, P. 2003. Overcoming local search through alliances and mobility.

Management Science, 49: 751–766.

Rosenkopf, L., & Nerkar, A. 2001. Beyond local search: boundary-spanning, exploration, and

impact in the optical disk industry. Strategic Management Journal, 22: 287–306.

Rothaermel, F. T., & Deeds, D. L. 2004. Exploration and exploitation alliances in biotechnology:

A system of new product development. Strategic Management Journal, 25: 201–221.

Schilling, M. A. 2015. Technology shocks, technological collaboration, and innovation

outcomes. Organization Science, 26(3): 668–686.

Schimmer, M., & Brauer, M. 2012. Firm performance and aspiration levels as determinants of a

firm’s strategic repositioning within strategic group structures. Strategic Organization, 10: 406–

435.

Sears, J., & Hoetker, G. 2014. Technological overlap, technological capabilities, and resource

recombination in technological acquisitions. Strategic Management Journal, 35(1): 48–67.

Semadeni, M., & Anderson, B. S. 2010. The follower’s dilemma: Innovation and imitation in the

professional services industry. Academy of Management Journal, 53: 1175–1193.

Sidhu, J. S., Commandeur, H. R., & Volberda, H. W. 2007. The multifaceted nature of exploration

and exploitation: Value of supply, demand, and spatial search for innovation. Organization

Science, 18: 20–38.

Sidhu, J. S., Volberda, H. W., & Commandeur, H. R. 2004. Exploring exploration orientation and

its determinants: Some empirical evidence. Journal of Management Studies, 41: 913–932.

44

Skilton, P. F., & Bernardes, E. 2015. Competition network structure and product market entry.

Strategic Management Journal, 36: 1688–1696.

Sorensen, J. B., & Stuart, T. E. 2000. Aging, obsolescence, and organizational innovation.

Administrative Science Quarterly, 45: 81–112.

Srinivasan, R., Haunschild, P., & Grewal, R. 2007. Vicarious learning in new product introductions

in the early years of a converging market. Management Science, 53(1): 16–28.

Stettner, U., & Lavie, D. 2014. Ambidexterity under scrutiny: Exploration and exploitation via

internal organization, alliances, and acquisitions. Strategic Management Journal, 35: 1903–

1929.

Stuart, T. E. 2000. Interorganizational alliances and the performance of firms: A study of growth

and innovation rates in a high-technology industry. Strategic Management Journal, 21: 791–

811.

Suchman, M. C. 1995. Managing legitimacy: Strategic and institutional approaches. Academy of

Management Review, 20: 571–610.

Suddaby, R., Bitektine, A., & Haack, P. 2017. Legitimacy. Academy of Management

Annals, 11(1): 451–478.

Tallman, S., & Li, J. 1996. Effects of international diversity and product diversity on the

performance of multinational firms. Academy of Management Journal, 39: 179–196.

Terlaak, A., & Gong, Y. 2008. Vicarious learning and inferential accuracy in adoption

processes. Academy of Management Review, 33(4): 846–868.

Toh, P. K., & Kim, T. 2013. Why put all your eggs in one basket? A competition-based view of

how technological uncertainty affects a firm’s technological specialization. Organization

Science, 24(4): 1214–1236.

Trajtenberg, M., Henderson, R., & Jaffe, A. B. 2002. University versus corporate patents: A

window on the basicness of invention. In A. B. Jaffe & M. Trajtenberg (Eds.), Patents, citations

& innovation: A window on the knowledge economy: 51–87. Cambridge, MA: MIT Press.

Turner, S. F., Mitchell, W., & Bettis, R. A. 2010. Responding to rivals and complements: How

market concentration shapes generational product innovation strategy. Organization Science, 21:

854–872.

Uotila, J., Maula, M., Keil, T., & Zahra, S. A. 2009. Exploration, exploitation, and financial

performance: Analysis of S&P 500 corporations. Strategic Management Journal, 30: 221–231.

Upton, G., & Cook, I. 1996. Understanding statistics. Oxford: Oxford University Press.

Veugelers, R. 1997. Internal R & D expenditures and external technology sourcing. Research

Policy, 26: 303–315.

Voss, G. B., Sirdeshmukh, D., & Voss, Z. G. 2008. The effects of slack resources and

environmental threat on product exploration and exploitation. Academy of Management

Journal, 51: 147–164.

Wilden, R., Hohberger, J., Devinney, T., & Lavie, D. 2018. Revisiting James March (1991):

Whither exploration and exploitation? Strategic Organization, 16(3), 352–369.

Zajac, E., & Westphal, J. 2004. The social construction of market value: Institutionalization and

learning perspectives on stock market reactions. American Sociological Review, 69: 433–457.

Zhao, E.Y., Fisher, G., Lounsbury, M., & Miller, D. 2017. Optimal distinctiveness: Broadening the

interface between institutional theory and strategic management. Strategic Management

Journal, 38: 93–113.

45

Tables and Figures Table 1a. 1st-Stage Descriptive Statistics and Correlations for Sample Selection, 1990–2006 (N = 30,976)

Table 1b. 1st-Stage Descriptive Statistics and Correlations for Partnering, 1990–2006 (N = 2,612)

Table 1c. 2nd-Stage Descriptive Statistics and Correlations for Sample, 1990–2006 (N = 2,337) Variables Mean S.D. Min Max 1 2 3 4 5 6 7 8 9 10 11

1. Firm Exploration t 0.727 0.172 0.091 0.986

2. Firm Exploration t-1 0.757 0.165 0.091 1 0.817 ***

3. Firm Age t-1 50.651 37.575 2 168 0.204 *** 0.279***

4. Firm Size t-1 5.645 13.101 0.003 151.802 0.161 *** 0.164*** 0.236***

5. Firm R&D Intensity t-1 0.108 0.219 0.003 5.402 -0.039 -0.065 ** -0.232*** -0.092***

6. Firm Solvency t-1 0.328 0.478 0 6.323 -0.158 *** -0.183*** -0.317*** -0.129*** 0.138***

7. Firm Uncertainty t-1 0.055 0.189 -0.144 2.359 -0.010 -0.038 -0.224 *** -0.079** 0.060* 0.047†

8. Performance Gap (Below Aspiration) -0.037 0.091 -1.176 0 0.030 0.037 0.126*** 0.078** -0.141*** -0.019 -0.204***

9. Performance Gap (Above Aspiration) 0.035 0.084 0 1.391 -0.066** -0.056* -0.118*** -0.079** 0.071** 0.119*** 0.187*** 0.170***

10. Alliance Portfolio Size t-1 20.046 45.56 1 470 0.046 † 0.013 0.000 0.639*** -0.026 -0.045† -0.010*** 0.019 -0.042

11. Strategic Alliance Portfolio t-1 0.8 0.288 0 1 -0.049 † -0.102 -0.252 -0.082† 0.107*** 0.105*** 0.106*** -0.078** 0.024 0.013

12. Number of Competitors t-1 621.527 358.1 24 1807 0.202 *** 0.180*** -0.136*** 0.228*** 0.119*** 0.064* 0.144*** -0.033 -0.014 0.238 0.151+

13. Corporate Strategy Function t-1 0.072 0.331 0 4 0.098 *** 0.117*** 0.112*** 0.171*** -0.052* -0.096*** -0.022 0.042+ -0.048† 0.155*** 0.155†

14. Lambda Selection t-1 4.503 3.51 0 12.959 -0.210 *** -0.264*** -0.900*** -0.253*** 0.329*** 0.381*** 0.223*** -0.136*** 0.138*** 0.088*** -0.142***

15. Lambda Partnering t-1 0.246 0.338 0 3.046 -0.010 0.049 † 0.277*** -0.220*** -0.094*** 0.119*** -0.074** 0.064* 0.002 -0.287*** -0.063**

16. Partner Exploration t-1 0.747 0.164 0.174 1 -0.015 -0.033 0.050 † 0.076† -0.060† -0.047† -0.026 0.043 -0.086** 0.101*** -0.054*

17. Competitor Exploration t-1 0.771 0.139 0.129 1 0.322 *** 0.405*** 0.138*** -0.067** -0.053* -0.050* -0.082*** 0.056† -0.003 -0.173*** -0.039

18. Variation in Partner Exploration t-1 0.021 0.028 0 0.156 -0.066 * -0.108*** -0.125*** 0.235*** 0.102*** 0.034 0.019 0.016 -0.054* 0.334*** 0.053**

19. Variation in Competitor Exploration t-1 0.029 0.038 0 0.261 -0.268 *** -0.268*** -0.009 -0.052* -0.041 -0.004 0.004 -0.010 -0.022 -0.042* 0.056†

20. Technological Proximity to Partners t-1 0.221 0.108 0.023 0.525 -0.063 * -0.092*** -0.429*** 0.153*** 0.169*** 0.174*** 0.142*** -0.071** -0.094*** 0.349*** 0.181***

21. Technological Proximity to Competitors t-1 0.185 0.078 0.030 0.553 -0.124 *** -0.100*** -0.320*** -0.023 0.298*** 0.186*** 0.080** 0.001 0.013 0.080*** 0.180***

Variables Sample Selection Mean S.D. Min Max 1 2 3 4 5 6 7

1. Patenting Experience t-1 13.321 72.324 0 1877.522

2. Firm Age t-1 15.345 19.221 0 169 0.188 ***

3. Firm Size t-1 1.680 7.973 0 192.319 0.596 *** 0.215***

4. Firm R&D Intensity t-1 2.281 22.651 0 889.500 -0.016 ** -0.051*** -0.021***

5. Firm Solvency t-1 1.055 6.160 0 659.027 -0.022 *** -0.070*** -0.030*** 0.039***

6. Corporate Strategy Function t-1 0.005 0.087 0 4 0.090 *** 0.148*** 0.089*** -0.005 -0.008

7. Number of Industry Peers t-1 125.844 153.369 1 673 -0.038 *** -0.230*** -0.070*** 0.043*** 0.049*** -0.025***

8. Market Size t-1 65.844 119.166 0 1482.922 0.101 *** -0.024*** 0.384*** 0.033*** 0.001 0.005 0.285***

Variables for Partnering Mean S.D. Min Max 1 2 3 4 5 6 7 8

1. Partnering Experience t-1 5.920 17.482 0 209.875

2. Firm Age t-1 51.093 36.762 1 168 0.014

3. Firm Size t-1 3.959 10.783 0.001 151.802 0.724 *** 0.208***

4. Firm R&D Intensity t-1 0.100 0.259 0 7.939 0.001 -0.207 *** -0.059**

5. Firm Solvency t-1 0.383 1.237 0 28.163 -0.019 -0.158 *** -0.060** 0.076***

6. Corporate Strategy Function t-1 0.056 0.294 0 4 0.135 *** 0.089*** 0.192*** -0.038* -0.042*

7. Number of Competitors t-1 556.474 344.921 0 1821 0.316 *** -0.135*** 0.238*** 0.095† -0.011 0.131***

8. Market Size t-1 45.059 60.72 0.098 434.844 0.339 *** -0.070*** 0.321*** 0.234*** 0.144*** 0.098*** 0.345***

9. Lambda Selection t-1 4.375 3.480 0 12.959 -0.040 * -0.910*** -0.210*** 0.282*** 0.234*** -0.104*** 0.112*** 0.345

46

Significance levels: † p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001

Table 2. 1st-Stage Probit Panel Model for Probabilities of Selection and Partnering Probability of Selection Probability of Partnering

Partnering Experience t-1 24.374 ** (0.042)

Patenting Experience t-1 1.844 *** (0.002)

Firm Age t-1 9.031 *** (0.006) 0.429 (0.003)

Firm Size t-1 -1.030 * (0.014) 2.064*** (0.028)

Firm R&D Intensity t-1 -11.899 (0.139) 0.691 * (0.587)

Firm Solvency t-1 -2.475 *** (0.033) -0. 313 (0.076)

Corporate Strategy Function t-1 0.066 (0.363) 0.161 (0.224)

Number of Industry Peers t-1 0.260 (0.001)

Number of Competitors t-1 -0.248 † (0.002)

Market Size t-1 -6.113 *** (0.002) -1.028*** (0.000)

Probability of Selection 0.965† (0.074)

N firms 3,624 184

N firm-years 30,976 2,612

N firms (Selected) 184 162

N firm-years (Selected) 2,612 (8.432%) 1,729 (66.195%)

Pseudo R2 0.864 0.136

-2 log likelihood 1,620.500 1794.742

Wald χ2 810.250*** 262.612*** Standardized beta coefficients; standard errors in parentheses; significance levels: † p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001

Variables Mean S.D. Min Max 12 13 14 15 16 17 18 19 20

13. Corporate Strategy Function t-1 0.072 0.331 0 4 0.096 ***

14. Lambda Selection t-1 4.503 3.51 0 12.959 0.088 *** -0.142***

15. Lambda Partnering t-1 0.246 0.338 0 3.046 -0.148 *** -0.112*** -0.184***

16. Partner Exploration t-1 0.747 0.164 0.174 1 -0.081 ** 0.054* -0.092*** -0.278

17. Competitor Exploration t-1 0.771 0.139 0.129 1 -0.100 *** -0.003 -0.136*** 0.135*** 0.029

18. Variation in Partner Exploration t-1 0.021 0.028 0 0.156 0.274 *** 0.020 0.125*** -0.395*** -0.103*** -0.173***

19. Variation in Competitor Exploration t-1 0.029 0.038 0 0.261 0.020 * -0.026* 0.005 0.001 -0.056* -0.392*** 0.037

20. Technological Proximity to Partners t-1 0.221 0.108 0.023 0.525 0.692 *** 0.050† 0.422*** -0.297*** -0.011 -0.199*** 0.375*** 0.692***

21. Technological Proximity to Competitors t-1 0.185 0.078 0.03 0.553 0.662 *** 0.015 0.344*** -0.052* -0.069** -0.125*** 0.217*** 0.153*** 0.824***

47

Table 3a. 2nd-Stage Panel Models for Partners with Fixed Effects and AR(1) Process DV: Firm exploration t Model 1 Model 2a Model 2b Model 3 Model 4 Model 5

Firm Fixed Effects Included Included Included Included Included Included

Firm Exploration t-1 0.559 *** (0.037) 0.554*** (0.037) 0.541*** (0.037) 0.546*** (0.036) 0.543*** (0.036) 0.540*** (0.036)

Firm Age t-1 -1.413 *** (0.001) -1.378*** (0.001) -1.545*** (0.001) -1.635*** (0.001) -1.692*** (0.001) -1.731*** (0.001)

Firm Size t-1 0.064 (0.001) 0.068 (0.001) 0.065 (0.001) 0.062 (0.001) 0.080 (0.001) 0.074 (0.001)

Firm R&D Intensity t-1 -0.009 (0.018) -0.007 (0.018) -0.003 (0.018) -0.002 (0.017) 0.000 (0.017) -0.001 (0.017)

Firm Solvency t-1 0.004 (0.007) 0.002 (0.007) -0.001 (0.007) 0.004 (0.007) 0.001 (0.007) 0.000 (0.007)

Performance Gap (Below Aspiration) t-1 0.013 (0.025) 0.012 (0.025) 0.011 (0.025) 0.017 (0.025) 0.018 (0.025) 0.018 (0.025)

Performance Gap (Above Aspiration) t-1 -0.065 * (0.036) -0.063* (0.036) -0.060* (0.036) -0.062* (0.036) -0.061* (0.036) -0.060* (0.036)

Corporate Strategy Function t-1 0.020 (0.008) 0.023 (0.008) 0.020 (0.008) 0.021 (0.008) 0.022 (0.008) 0.022 (0.008)

Alliance Portfolio Size t-1 -0.022 (0.000) -0.025 (0.000) -0.034 (0.000) -0.037 (0.000) -0.026 (0.000) -0.030 (0.000)

Strategic Alliance Portfolio t-1 -0.038 † (0.012) -0.037† (0.012) -0.037† (0.012) -0.035 (0.012) -0.041† (0.012) -0.040† (0.012)

Lambda Sample Selection t-1 0.023 (0.006) 0.007 (0.006) -0.051 (0.006) -0.109 (0.006) -0.113 (0.006) -0.126 (0.006)

Lambda Partnering t-1 -0.028 (0.017) -0.022 (0.017) 0.011 (0.018) 0.010 (0.018) 0.016 (0.018) 0.013 (0.018)

Partner Exploration t-1 0.041 * (0.020) 0.544*** (0.143) 0.485*** (0.142) 0.623*** (0.159) 0.624*** (0.164)

Partner Exploration t-1 2 -0.510*** (0.103) -0.463** (0.102) -0.588*** (0.114) -0.587*** (0.114)

Firm Uncertainty t-1 -0.137 † (0.065) -0.129 (0.065) -0.134† (0.065)

Firm Uncertainty t-1 × Partner Exploration t-1 0.221 ** (0.085) 0.212** (0.084) 0.218** (0.085)

Variation in Partner Exploration t-1 0.320 ** (0.601) 0.311** (0.614)

Variation in Partner Exploration t-1 × Partner Exploration t-1 -0.351 *** (0.891) -0.343*** (0.907)

Technological Proximity to Partners t-1 0.051 (0.147)

Technological Proximity to Partners t-1 × Partner Exploration t-1 -0.007 (0.180)

AR(1) 0.18 0.18 0.18 0.17 0.16 0.16

N firm-years 1474 1474 1474 1474 1474 1474

N firms 153 153 153 153 153 153

F 36.3 33.9 32.7 31.4 29.1 26.3

Degrees of freedom 1309 1308 1307 1305 1303 1301

Log likelihood 1746.026 1748.466 1755.688 1770.004 1777.431 1778.098

χ2 (∆2LL) 4.880* 19.320*** 47.960*** 62.810*** 64.140*** Standardized beta coefficients; standard errors in parentheses; significance levels: † p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001

48

Table 3b. 2nd-Stage Panel Models for Competitors with Fixed Effects and AR(1) Process DV: Firm exploration t Model 1 Model 2a Model 2b Model 3 Model 4 Model 5

Firm Fixed Effects Included Included Included Included Included Included

Firm Exploration t-1 0.411 *** (0.030) 0.401*** (0.030) 0.399*** (0.030) 0.402*** (0.030) 0.402*** (0.030) 0.416*** (0.030)

Firm Age t-1 -2.143 *** (0.001) -2.153*** (0.001) -2.193*** (0.001) -2.222*** (0.001) -2.248*** (0.001) -2.277*** (0.001)

Firm Size t-1 0.031 (0.001) 0.036 (0.001) 0.039 (0.001) 0.039 (0.001) 0.041 (0.001) 0.044 (0.001)

Firm R&D Intensity t-1 -0.011 (0.019) -0.010 (0.019) -0.011 (0.019) -0.009 (0.019) -0.010 (0.019) -0.013 (0.019)

Firm Solvency t-1 -0.004 (0.003) -0.003 (0.003) -0.002 (0.003) -0.002 (0.003) -0.000 (0.003) -0.006 (0.003)

Performance Gap (Below Aspiration) t-1 -0.004 (0.024) -0.004 (0.024) -0.004 (0.024) 0.004 (0.025) 0.005 (0.025) 0.007 (0.025)

Performance Gap (Above Aspiration) t-1 -0.009 (0.029) -0.010 (0.029) -0.013 (0.029) -0.018 (0.029) -0.019 (0.029) -0.017 (0.029)

Corporate Strategy Function t-1 0.017 (0.009) 0.017 (0.009) 0.017 (0.009) 0.019 (0.009) 0.019 (0.009) 0.015 (0.009)

Number of Competitors t-1 0.210 *** (0.000) 0.215*** (0.000) 0.217*** (0.000) 0.204*** (0.000) 0.204*** (0.000) 0.264*** (0.000)

Lambda Sample Selection t-1 -0.032 (0.007) -0.048 (0.007) -0.071 (0.007) -0.091 (0.006) -0.117 (0.006) -0.079 (0.007)

Competitor Exploration t-1 0.034 * (0.020) 0.236** (0.109) 0.228** (0.109) 0.362*** (0.125) 0.485*** (0.138)

Competitor Exploration t-1 2 -0.209* (0.076) -0.205* (0.076) -0.329*** (0.089) -0.343*** (0.089)

Firm Uncertainty t-1 -0.080 (0.070) -0.091 (0.070) -0.116 (0.069)

Firm Uncertainty t-1 × Competitor Exploration t-1 0.142 † (0.091) 0.152* (0.090) 0.170* (0.090)

Variation in Competitor Exploration t-1 0.190 ** (0.314) 0.170* (0.314)

Variation in Competitor Exploration t-1 × Competitor Exploration t-1 -0.210 ** (0.464) -0.185** (0.465)

Technological Proximity to Competitors t-1 0.092 (0.190)

Technological Proximity to Competitors t-1 × Competitor Exploration t-1 -0.261 ** (0.227)

AR(1) 0.26 0.26 0.26 0.26 0.25 0.24

N firm-years 2337 2337 2337 2337 2337 2337

N firms 180 180 180 180 180 180

F 59.5 54.4 50.5 45.3 40.6 39.4

Degrees of freedom 2147 2146 2145 2143 2141 2139

Log likelihood 2420.087 2422.690 2425.932 2435.092 2441.421 2456.993

χ2 (∆2LL) 5.210* 11.690** 30.010*** 42.670*** 73.810*** Standardized beta coefficients; standard errors in parentheses; significance levels: † p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001

49

Table 3c. 2nd-Stage Panel Models for Partners and Competitors with Fixed Effects and AR(1) Process DV: Firm exploration t Model 1 Model 2a Model 2b Model 3 Model 4 Model 5

Firm Fixed Effects Included Included Included Included Included Included

Firm Exploration t-1 0.537 *** (0.036) 0.531*** (0.036) 0.521*** (0.036) 0.526*** (0.036) 0.529*** (0.036) 0.538*** (0.036)

Firm Age t-1 -1.776 *** (0.001) -1.761*** (0.001) -1.910*** (0.001) -1.952*** (0.001) -2.013*** (0.001) -2.042*** (0.001)

Firm Size t-1 0.067 (0.001) 0.073 (0.001) 0.071 (0.001) 0.064 (0.001) 0.083 + (0.001) 0.106* (0.001)

Firm R&D Intensity t-1 -0.018 (0.017) -0.016 (0.017) -0.013 (0.017) -0.011 (0.017) -0.010 (0.017) -0.009 (0.016)

Firm Solvency t-1 0.002 (0.007) 0.001 (0.007) -0.001 (0.007) 0.003 (0.007) 0.002 (0.007) 0.004 (0.007)

Performance Gap (Below Aspiration) t-1 0.024 † (0.026) 0.023 (0.026) 0.022 (0.026) 0.026† (0.026) 0.027+ (0.026) 0.028† (0.026)

Performance Gap (Above Aspiration) t-1 -0.054 ** (0.036) -0.052** (0.036) -0.049** (0.036) -0.046* (0.036) -0.047* (0.036) -0.047* (0.036)

Corporate Strategy Function t-1 0.012 (0.008) 0.014 (0.008) 0.013 (0.008) 0.015 (0.008) 0.016 (0.008) 0.013 (0.008)

Alliance Portfolio Size t-1 -0.042 (0.000) -0.046 (0.000) -0.055 † (0.000) -0.058* (0.000) -0.041 (0.000) -0.034 (0.000)

Strategic Alliance Portfolio t-1 -0.029 (0.012) -0.028 (0.012) -0.028 (0.012) -0.029 (0.012) -0.036 † (0.012) -0.043* (0.012)

Number of Competitors t-1 0.184 *** (0.000) 0.188*** (0.000) 0.188*** (0.000) 0.175*** (0.000) 0.165*** (0.000) 0.239*** (0.000)

Lambda Sample Selection t-1 0.016 (0.006) -0.006 (0.006) -0.056 (0.006) -0.106 (0.006) -0.141 (0.006) -0.122 (0.006)

Lambda Partnering t-1 -0.036 (0.017) -0.030 (0.017) -0.008 (0.017) -0.009 (0.017) -0.006 (0.017) -0.002 (0.017)

Partner Exploration t-1 0.041 * (0.019) 0.413** (0.142) 0.371** (0.141) 0.476** (0.157) 0.377* (0.163)

Competitor Exploration t-1 0.001 (0.022) 0.164 + (0.118) 0.147 (0.118) 0.278* (0.136) 0.354** (0.151)

Partner Exploration t-1 2 -0.377** (0.101) -0.347* (0.101) -0.442** (0.113) -0.379* (0.113)

Competitor Exploration t-1 2 -0.171+ (0.085) -0.161+ (0.084) -0.271* (0.098) -0.266* (0.099)

Firm Uncertainty t-1 -0.264 * (0.087) -0.258* (0.087) -0.232* (0.087)

Firm Uncertainty t-1 × Partner Exploration t-1 0.215 ** (0.083) 0.198* (0.083) 0.168* (0.083)

Firm Uncertainty t-1 × Competitor Exploration t-1 0.125 (0.086) 0.137 † (0.086) 0.133 (0.086)

Variation in Partner Exploration t-1 0.236 * (0.589) 0.238* (0.594)

Variation in Partner Exploration t-1 × Partner Exploration t-1 -0.259 ** (0.872) -0.252* (0.879)

Variation in Competitor Exploration t-1 0.250 *** (0.339) 0.224** (0.345)

Variation in Competitor Exploration t-1 × Competitor Exploration t- -0.263 *** (0.506) -0.230** (0.516)

Technological Proximity to Partners t-1 -0.262 * (0.185)

Technological Proximity to Partners t-1 × Partner Exploration t-1 0.091 (0.176)

Technological Proximity to Competitors t-1 0.178 † (0.228)

Technological Proximity to Competitors t-1 × Competitor Exploration t-1 -0.185 * (0.244)

AR(1) 0.16 0.15 0.15 0.14 0.14 0.13

N firm-years 1474 1474 1474 1474 1474 1474

N firms 153 153 153 153 153 153

F 40.7 35.9 32.7 29.8 26.4 23.9

Degrees of freedom 1308 1306 1304 1301 1297 1293

Log likelihood 1775.337 1778.646 1785.296 1798.201 1811.188 1821.138

χ2 (∆2LL) 6.620* 19.920** 45.730*** 71.700*** 91.600*** Standardized beta coefficients; standard errors in parentheses; significance levels: † p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001

50

Figure 1. Conceptual Model for Convergence and Divergence of Exploration Levels

Figure 2a. The Effect of Partner Exploration

on Firm Exploration

Figure 2b. The Effect of Competitor Exploration

on Firm Exploration

.5 5

.6 .6

5 .7

.7 5

F ir

m E

x p

lo ra

ti o

n (

t)

.2 .4 .6 .8 1

Partner Exploration (t-1)

Mean Mean

51

Figure 3a. Firm Exploration by Partner

Exploration and Firm-Specific Uncertainty

Figure 3b. Firm Exploration by Competitor

Exploration and Firm-Specific Uncertainty

Figure 4a. Firm Exploration by Partner

Exploration and Variation in Partner Exploration

Figure 4b. Firm Exploration by Competitor

Exploration and Variation in Competitor Exploration

Figure 5b. Firm Exploration by Competitor

Exploration and Technological Proximity

View publication statsView publication stats