Peeking Inside the Black Box: Inventor Turnover and Patent Termination

Abstract

This study examines the consequences of inventor turnover for the termination of the patents they produced beforehand. The proponents of the knowledge-based view consider firms repositories of knowledge and inventors carriers of knowledge who create new inventions through recombination. I argue that because the knowledge about inventions resides with inventors, turnover among them may adversely affect the future use of their inventions due to the loss of tacit knowledge. As a result, firms are more likely to terminate the patents of inventors who have left than of inventors who remain. Further, the patterns of collaboration between departed inventors and others influence the aforementioned relationship. Analyses of the termination of patents, in the form of nonpayment of renewal fees, by pharmaceutical firms support the claims in the article. The findings have implications for how firms manage their knowledge portfolios in the face of inevitable inventor turnover and the resulting loss of tacit knowledge.

Keywords

turnover tacit knowledge patent termination knowledge-based view innovation portfolio

The mobility of personnel among firms is known to be a key driver of transferring knowledge, which can be a source of firm competitive advantage (Arrow, 1972; Barney, 1991; Gardner, 2002). Previous work on employee mobility suggests that it influences several organizational outcomes, including innovation, knowledge diffusion and spillover, long-term survival, and entrepreneurial spinouts (Agarwal, Ganco, & Ziedonis, 2009; Almeida & Kogut, 1999; Campbell, Ganco, Franco, & Agarwal, 2012; Ganco, 2013; Phillips, 2002; Tzabbar & Kehoe, 2014). Most of the work in employee mobility has explored the knowledge spillover and its impact on the subsequent innovation outcomes in the new firm (Møen, 2005; Song, Almeida, & Wu, 2003; Tzabbar, 2009). The premise of this work is that the arrival of an inventor to a new firm serves as a mechanism to transfer the inventor’s old firm-specific knowledge to the new firm (Hoisl, 2007; Maliranta, Mohnen, & Rouvinen, 2009). However, we have a limited understanding of what happens to the knowledge resources, such as patents, that the inventor generated in the old firm.

The study aims to peek inside this “black box” and shed light on how firms manage patents of inventors who have left their firms. In particular, I focus on the departure of inventors and its influence on firms’ decisions to terminate the patents of these inventors. I draw on the knowledge-based view of the firm for the theoretical underpinning of this article (Grant, 1996; Kogut & Zander, 1992; Nonaka, 1994). According to this view, individuals—in this case, inventors—create knowledge through recombination. Firms cannot create knowledge without inventors. However, the role of the firms is to provide appropriate context and a system of rules and procedures to facilitate the process of knowledge creation (Nonaka, 1994).

In the context of innovation, firms own the rights to patents that their inventors create and also perform an important task of managing these patents (Khanna, Guler, & Nerkar, 2016). Once a patent is created by its inventor(s) in a firm, it is the firm’s decision whether to maintain the patent for future recombinations or terminate the patent (Serrano, 2010). Knowledge workers in general, and inventors in particular, accumulate significant tacit knowledge (Almeida & Kogut, 1999; Argote & Ingram, 2000). In addition, the tacit know-how about the knowledge underlying patents is critical in using that knowledge to generate future inventions (Arora, 1996; Fey & Birkinshaw, 2005; Fleming, 2001). The key argument in this study is based on the premise that the turnover of inventors results in the loss of this tacit knowledge, which is likely to make the use of their patents as a foundation for future inventions challenging. I surmise that this difficulty leads firms to terminate the patents of departed inventors.

I also examine the conditions under which the tacit knowledge of departed inventors can be retained to some extent within firms, and therefore firms are less likely to terminate their patents. I focus on a specific mechanism for this: knowledge transfer through collaboration. Inventions often result from collaboration, as this is the best way to recombine knowledge (Nerkar & Paruchuri, 2005; Paruchuri, 2010). The informal networks resulting from collaborative ties between inventors are a source of several advantages, including mutual trust, access to resources, and knowledge transfer (T. Allen & Cohen, 1969; Reagans & McEvily, 2003; Reagans, Zuckerman, & McEvily, 2004; Tushman & Scanlan, 1981). Moreover, such collaborative ties can promote the sharing of tacit knowledge (Nahapiet & Ghoshal, 1998). Therefore, I argue that the departure of inventors does not have a uniform effect on the termination of their patents; it depends on the nature of these inventors’ collaborations before departing. The empirical analyses of inventor turnover in 102 pharmaceutical firms over 15 years indicate that firms are more likely to terminate the patents of inventors who have left them, but this relationship is weakened by the degree of knowledge overlap in the collaborative ties of these inventors with their coauthors.¹ The strength and stability of these ties also reduced the likelihood of termination of their patents.

In addition to the opening of the “black box” of inventor turnover and subsequent knowledge creation, the focus on patents that inventors generated before their turnover has two merits. First, research into the knowledge-based view has led scholars to argue that knowledge is a strategic resource for firms and that the creation of new knowledge through recombination is a key source of competitive advantage for them (Argyres, 1996; Hargadon & Sutton, 1997; Katila & Ahuja, 2002; Kogut & Zander, 1992; Rosenkopf & Nerkar, 2001). Knowledge exists in two forms: explicit and tacit. This study offers insights into what happens when the tacit source of knowledge is decoupled from the explicit source because of inventor turnover. The findings of the article indicate that separating the tacit and explicit components of knowledge changes the value of the knowledge to firms, which helps explain how firms evaluate their knowledge portfolios. Firms may retain the intellectual property rights on the patents of inventors they no longer employ, giving the firms access to the explicit or declarative knowledge of these inventors (J. R. Anderson, 1983). However, tacit or procedural knowledge is no longer available to firms following inventor turnover. The results of this study emphasize that the “know-what” contained in patents must be combined with the “know-how” held by inventors (Brown & Duguid, 1998; Ryle, 1949) for firms to be able to generate value from their knowledge-related resources and create a competitive advantage.

Second, this study responds to the call for expanding our understanding of how firms transform their resources (in this case, knowledge assets) to create value (Priem & Butler, 2001; Sirmon, Hitt, & Ireland, 2007). The mere possession of knowledge assets does not guarantee value creation, but it does require the recombination and exploitation of these assets (Barney & Arikan, 2001; Priem & Butler, 2001). This study presents a more dynamic approach to managing organizational resources, also termed “resource orchestration” by previous work (Sirmon, Hitt, Ireland, & Gilbert, 2011). This perspective stresses the need to consider firms’ value-creation capabilities as a function of their “resource bundles” and resource orchestration. A firm’s portfolio of patented inventions in the present case signifies the firm’s resource bundle and is a result of an ongoing process of search and selection by the firm. The subsequent task of the firm is to generate value from its resource bundle by deploying or orchestrating the bundle’s resources. This article underscores factors that inform firms’ decision-making in effectively managing their resource bundles, that is, their knowledge portfolios, extending our understanding of the process through which firms configure their resource bundles to create value.

This study contributes to the vast and growing literature on the knowledge-based view that underscores the role of firms in providing relevant contexts and mobilizing the tacit knowledge held in the minds of individuals (Arthur, 2007; Bartel & Garud, 2009; Gittelman & Kogut, 2003; Hall & Ziedonis, 2001; Kaplan & Vakili, 2015; Kogut & Zander, 1992). Inventors within firms create knowledge through recombination in the form of patents, and an extensive body of work has been concerned with the output of this recombination by studying its effects on various firm-level outcomes (e.g., Gavetti & Levinthal, 2000; Katila & Ahuja, 2002; Levinthal & March, 1981; Rosenkopf & Nerkar, 2001). In this article, however, I focus on inputs to the process of recombination. Due to the considerable uncertainty about research-and-development (R&D) outcomes, firms often hold a portfolio of patented inventions. However, only a fraction of these patents enjoy commercial success (Thomke & Kuemmerle, 2002), and firms must constantly assess the value of these patents as they progress through the invention process. Thus, an important decision that firms must make is whether to continue or terminate each invention. These decisions form the foundation of future knowledge generation and are fundamental to the effective management of the firms’ innovation portfolios. The findings in this article revealed that the tacit knowledge embedded in inventors’ minds affects the value of explicit knowledge contained in the patents of these inventors. The changes resulting from inventor turnover in the value of firms’ knowledge assets, such as patents, shed light on how firms make termination decisions, presenting a more balanced view of knowledge creation through recombination within firms. The study, therefore, adds to the extant work on knowledge creation by focusing on the antecedents of recombination. Moreover, this is the first study to my knowledge that examines inventor turnover as a potential source of loss of tacit knowledge and investigates how this loss translates into devaluing the explicit knowledge contained in firms’ patents.

Inventor Turnover and Patent Termination

Scholars of management have emphasized the role of inventors in improving firms’ performance (Paruchuri, 2010; Zucker, Darby, Brewer, & Peng, 1995). These employees are carriers of skills, expertise, and tacit knowledge (Kogut & Zander, 1996; Nickerson & Zenger, 2004). However, for better or worse, inventors are free to leave their respective firms at will (Coff, 1997), and studies have shown that the loss of employees has implications for firms’ overall success (Phillips, 2002; Rosenkopf & Almeida, 2003; Sørensen, 1999; Wezel, Cattani, & Pennings, 2006). The consequences of inventor turnover can be broadly divided into three categories (see Mawdsley & Somaya, 2016, for an excellent review). The first and also the most studied aspect of inventor turnover is the spillover of technological knowledge to destination firms. Inventors may carry and transfer the tacit knowledge (Hoisl, 2007; Maliranta et al., 2009; Møen, 2005), routines (Song et al., 2003; Tzabbar, 2009), and relational capital (Cohen & Levinthal, 1990) to destination firms. In the context of entrepreneurial spinouts, Gambardella, Ganco, and Honore (2015) and Ganco (2013) examine the conditions under which mobile inventors are more (or less) likely to transition to entrepreneurship. The second and relatively less explored area is the influence of this turnover on source firms. For example, Tzabbar and Kehoe (2014) highlight the effect of the departure of high-status inventors on firms’ exploitative and explorative activities. The last and also the least studied area in inventor mobility research is the examination of knowledge-related assets that inventors created in source firms before their departure. This is the gap I aim to fill in this article.

This is an important phenomenon to examine for the following reason. The creation of new knowledge is central to the existence of firms in high-tech industries. The knowledge-based view argues that firms derive a competitive advantage not only from holding a stock of knowledge but from enabling the creation of new knowledge through recombination (Argyres, 1996; Galunic & Rodan, 1998; Hargadon & Sutton, 1997; Kogut & Zander, 1992). The creation of knowledge is thus conceived of as a recombinant process involving the discovery and use of knowledge (Henderson & Clark, 1990; Kogut & Zander, 1992). This process is path dependent and incremental in nature; that is, inventors rely heavily on existing pieces of knowledge as building blocks to create new knowledge (Cohen & Levinthal 1990; March, 1991; Martin & Mitchell, 1998; Nelson & Winter, 1982; Nerkar, 2003). A natural question thus arises as to how the recombination of knowledge-related resources is impacted when the inventors who generated those resources leave their firms.

The knowledge used in recombination involves two components: explicit knowledge and tacit knowledge (Nonaka, Toyama, & Konno, 2000). Explicit knowledge can be codified, stored, and accessed by inventors easily and in a straightforward way. Tacit knowledge, on the other hand, is difficult to transfer (Polanyi & Grene, 1969) and thus is often a source of competitive advantage for firms in knowledge-intensive industries (Coff, 1997). Tacit knowledge cannot be codified or documented, but it can be acquired by people sharing experiences and working closely together. Knowledge creation requires both explicit and tacit knowledge; the possession of one without the other may not lead to desirable outcomes (Kikoski & Kikoski, 2004; Nonaka et al., 2000). Building on this notion, Ganco (2013) argued that complex knowledge, which comprises a high proportion of tacit knowledge, makes the knowledge transfer particularly challenging. Further, the study also found that inventors who possess complex knowledge are less likely to leave their firms, as their knowledge is of limited value outside its originating context due to the complications associated with the transfer of tacit knowledge.

The need for both explicit and tacit knowledge provides a substantial advantage to firms, because even if the knowledge to be recombined is observable and codified, the underlying ability to successfully combine it is often tacit and complex (Kogut & Zander, 1992; Nerkar, 2003). For instance, there are two components to each patented invention: explicit and tacit. A firm or other entity is obligated to disclose information about each invention, including specifications of its unique nature and claims, to obtain the exclusive rights of patentees. This represents the codified information available in the patent document that is available to the public, and anyone can use this information to build subsequent inventions as long as the rights in the patents are not breached. Researchers have examined the implications of knowledge externality and knowledge spillovers arising from the information contained in public patent documents (Jaffe, 1986; Jaffe, Trajtenberg, & Henderson, 1993).

The other component, tacit knowledge, is not described in the patent document, however; it resides only with the inventor (Nonaka & Takeuchi, 1995; Ravetz, 1971; Spender, 1996). The successful recombination of inventions to create knowledge requires (a) the codified information in the patent document and (b) the tacit knowledge, or know-how, of the inventor of the patent. Previous work has argued that scientists and inventors are carriers of significant tacit knowledge (Almeida & Kogut, 1999; Argote & Ingram, 2000). Firms can rely on patents to codify and protect their knowledge (Katila & Mang, 2003), but a significant portion of their knowledge is tacit and hard to codify in patents (Pisano, 1997). The tacit knowledge that inventors possess enables them to apply and recombine the codified knowledge in patent documents because of their understanding of whether, and under what conditions, the recombined knowledge is likely to lead to beneficial outcomes.

In light of the aforementioned arguments, inventor turnover should deplete a firm’s stock of tacit knowledge. The firm will still have access to the inventors’ codified knowledge but will lack their tacit knowledge. In the absence of the know-how embodied in departed inventors, it would be challenging to use the know-what embedded in their patents in future recombination. This reasoning leads to the following prediction:

Hypothesis 1: All else being equal, a firm is more likely to terminate patents of inventors who have left than of those who remain.

The Moderating Role of Departed Inventors’ Collaborative Ties

While the new knowledge is created by individuals, firms play a central role in providing appropriate context that articulates the knowledge of individuals and forms the basis of knowledge creation (Grant, 1996; Kogut & Zander, 1992). Scholars have conceptualized the process of knowledge creation as a multilevel phenomenon (Nonaka, 1994). The multilevel perspective of knowledge generations is analogous to the one argued in the literature on organizational learning as well. For example, Crossan, Lane, and White (1999), in their 4I framework, characterized organizational learning to take place at three levels: individual, group, and organization. Along the same lines, Khanna et al. (2016) conceived organizational learning a multilevel process in which scientists create patented inventions and firms manage and assess the value of these inventions for further development based on the feedback they receive.

In the present context, inventors hold knowledge in their respective research areas and collaborate in teams to create new knowledge in the form of patents. The stock of patents resulting from the collaborative efforts of numerous teams of inventors represents organizational knowledge (Nonaka, 1994). The transfer of tacit knowledge between inventors occurs when they collaborate with each other. Nonaka (1994) refers to this process as socialization. The basic premise here is that the key to acquiring tacit knowledge is experience. During the collaboration, inventors are likely to transfer tacit knowledge because they observe and follow each other. The shared experiences during collaboration allow inventors to understand the nuanced context in which the knowledge is being created, enabling them to acquire tacit knowledge more effectively.

A substantive body of empirical and theoretical work in relation to firm exit and employee mobility supports the notion that the tacit component of knowledge is largely collaboration based and embedded within the web of interactions between individuals (Fortune & Mitchell, 2012; Hoetker & Agarwal, 2007; Nelson & Winter, 1982). For instance, Hoetker and Agarwal (2007) have shown that following the exit of firms in the disk-drive industry, their patents are less likely to be cited (used) by other firms in the industry. They attributed this lack of patent use to the loss of tacit knowledge embodied in the social relations and shared cultural context (Leonard & Sensiper, 1998). Ganco (2013) has suggested that when the knowledge of inventors in the semiconductor industry is complex, that is, comprising a high degree of tacitness, the recipient firms are more inclined to hire these inventors along with their team members to minimize the loss of the tacit knowledge that is embedded in their collaborative interactions, shared experience, and coordinated routines. These empirical findings indicate that collaborative relationships between inventors may contain a portion of the tacit knowledge of individual inventors.

Thus, I focus on the collaborative ties of inventors in understanding the extent of loss of tacit knowledge following their departure. Though collaborative ties, in general, may facilitate the exchange of tacit knowledge between inventors, the extent to which this occurs may vary and depends on the nature of these ties. The heterogeneity in loss of tacit knowledge across firms following inventor turnover will likely influence the termination decisions in these firms. If the coauthors of the departed inventor acquired her or his tacit knowledge, this knowledge could be leveraged to use the inventor’s patents for future recombinations. In this case, the firm is less likely to terminate the departed inventor’s patents.

To understand under what conditions departed inventors may transfer their tacit knowledge to their collaborators, I consider three characteristics of the collaborative ties of departed inventors—knowledge overlap, strength, and stability—and argue that they affect the degree of loss of tacit knowledge and, as a result, firms’ decision to terminate patents of these inventors. The emphasis on these attributes of collaborative ties is motivated by the past research that has identified them as central in driving knowledge transfer in general and transfer of tacit knowledge in particular (Cohen & Levinthal, 1990; Ebadi & Utterback, 1984; Gargiulo, Ertug, & Galunic, 2009; Krackhardt, 1999; McEvily, Perrone, & Zaheer, 2003; Simon, 1991; Uzzi, 1997).

While various aspects of collaborative ties offer several structural benefits, they may not be as important for the transfer of tacit knowledge. For example, different types of centrality in a network indicate the importance and status of individuals with higher centrality. While higher centrality allows an individual to exert greater power and influence (see Brass, 1992, for an exhaustive review), its role in the transfer of tacit knowledge is not apparent. Also, reach and range of and structural holes in an inventor’s network may increase the diffusion of information and enable the inventor to access diverse and unique information (Burt, 1992; Hansen, Podolny, & Pfeffer, 1999; Schilling & Phelps, 2007), but they may not be as effective in the transfer of tacit knowledge, which requires common understanding, close collaboration, and shared beliefs between individuals.² Moreover, recent work has shown that collaboration networks in which individuals span structural holes are characterized by the lack of interpersonal trust and greater opportunism (Bizzi, 2013; Funk, 2015), conditions not suited for the transfer of tacit knowledge. In the next sections, I turn my attention to three aspects of departed inventors’ collaboration networks: (a) knowledge overlap, (b) strength, and (c) stability. Next, I describe prior research to underscore why these characteristics are important in determining the loss of tacit knowledge following inventor turnover. Last, I present results from the empirical analyses followed by discussion and directions for future research.

Knowledge overlap

According to the knowledge-based view, firms are social communities that enable the efficient creation and transfer of knowledge by individuals (Grant, 1996; Kogut & Zander, 1996). Furthermore, the patterns of relationships among the individuals in a firm play a central role in knowledge transfer (Nonaka, 1994). In particular, informal intraorganizational networks are thought to play an instrumental role in knowledge transfer (Uzzi, 1997). Therefore, scholars have proposed that tacit knowledge resides not only in the minds of inventors but in the nexus of relationships between them (Brown & Duguid, 2001; Reed & Defillippi, 1990). Prior work has shown that collaborative ties between individuals are central to the transfer of tacit knowledge (Reagans & McEvily, 2003). Unlike explicit knowledge, tacit knowledge cannot be recorded in manuals and documents. The best way to transfer it is thus interpersonal communication (Edmondson, Bohmer, & Pisano, 2000; Szulanski, 1996). In other words, know-how resists codification and can be passed on only through shared experience and close collaboration (Nelson & Winter, 1982).

I contend that inventors are more likely to successfully transfer tacit knowledge to each other when there is an overlap in their knowledge. Individuals assimilate new knowledge by associating it with what they already know (Cohen & Levinthal, 1990; Simon, 1991). The overlap in knowledge is likely to make it easier for an inventor to convey her or his knowledge to coauthors and enable coauthors to comprehend and apply the knowledge they have acquired for future use (Tortoriello, Reagans, & McEvily, 2012). Also, knowledge overlap between inventors leads to convergence of their knowledge and expertise over time (McFadyen, Semadeni, & Cannella, 2009). Overlapping knowledge is critical to scientific discovery, as inventors must accumulate a certain amount of basic knowledge before they can identify and reconcile it to create new knowledge (Cohen & Levinthal, 1990). In the present context, overlapping knowledge allows for better transfer of tacit knowledge by facilitating the development of consensus and group norms, expectations, and behaviors (Coleman, 1988).

These arguments suggest that the more the knowledge of an inventor overlaps with the knowledge of his or her coauthors, the transfer of tacit knowledge from the inventor to coauthors is more likely to occur. In this case, after the inventor departs, the firm does not completely lose the inventor’s tacit knowledge, as it still can access a portion of the tacit knowledge through the inventor’s coauthors. The coauthors of the departed inventor can apply whatever tacit knowledge they acquired from the inventor in future innovation activities. This should increase the likelihood that the patents of the departed inventor will be used in future recombinations within the firm and consequently decrease the likelihood that the firm will terminate these patents.

Hypothesis 2: The knowledge overlap in the collaborative ties of departed inventors negatively moderates the positive relationship between the inventors’ departure and the firm’s termination of their patents.

Strength of collaborative ties

Another factor that is considered to be important in knowledge transfer is the strength of collaborative ties. Borrowing from previous research, I argue that the strength of a tie between two inventors in the present context is characterized by the frequency of their collaboration (Granovetter, 1973; McFadyen et al., 2009). The transfer of knowledge between individuals can be more challenging than it appears. It requires individuals to not only engage in the process of knowledge exchange but also be willing to exert necessary efforts by committing time and resources to transfer the knowledge successfully (Szulanski, 1996). In addition, by sharing what they know, especially tacit know-how, with their collaborators, individuals may become redundant with them, undermining their worth and value.

Strong interpersonal ties between individuals can help overcome these obstacles in several ways. First, an inventor continues to interact with other inventors when their collaboration seems valuable and useful (Bouty, 2000). Strong ties by nature indicate frequent interactions, suggesting that the inventors sharing such ties find the collaboration useful and are therefore more likely to share exclusive or tacit knowledge. Strong ties are particularly effective in the transfer of tacit and complex knowledge that requires deeper interaction between inventors (Polanyi & Grene, 1969).

Second, strong ties between inventors are characterized by greater trust and cooperation. Individuals with strong collaborative ties exhibit greater trust (Levin & Cross, 2004) and hence are less concerned about their knowledge being misappropriated by others and are more likely to share their knowledge with each other (Granovetter, 1973; Krackhardt, 1999). The lack of opportunism and the presence of trust and cooperation in strong ties allow the transfer of exclusive and tacit knowledge between inventors (Szulanski, 1996; Uzzi, 1996). Third, individuals who collaborate frequently are more helpful to each other (Cross & Sproull, 2004; Granovetter, 1982; Krackhardt, 1992). The availability of assistance and support through strong ties further increases individuals’ willingness to exchange nuanced and tacit information (Obstfeld, 2005).

Last, the willingness to transfer knowledge is not enough for successful knowledge transfer. Individuals’ ability to share knowledge effectively is also an important factor. Strong ties promote the development of a common understanding and collective beliefs and norms, which facilitate the effective transfer of fine-grained information (Edwards, Mayernik, Batcheller, Bowker, & Borgman, 2011; Kuhn, 1962; Star & Griesemer, 1989). Individuals speaking a common language encounter fewer obstacles to transferring, integrating, and combining their tacit knowledge to create new knowledge. When an inventor shared strong ties with his or her collaborators, it is likely that the inventor transferred some of the tacit knowledge to them during their collaborative efforts. In this case, when the inventor departs, the collaborators of the inventor are better positioned to compensate for the loss of his or her tacit knowledge, which increases the likelihood that the patents of the departed inventor will be used in subsequent recombinations. Thus, I propose the following:

Hypothesis 3: The strength of the collaborative ties of departed inventors negatively moderates the positive relationship between the inventors’ departure and the firm’s termination of their patents.

Stability of collaborative ties

Collaborative efforts are at the center of achieving success in innovation (e.g., T. Allen & Cohen, 1969; Tushman, 1978, 1979; Tushman & Katz, 1980). Prior research has provided a surfeit of evidence in support of the idea that collaborative ties between various entities, such as individuals, business units, or firms, have a significant impact on innovation outcomes (Fleming, Mingo, & Chen, 2007; Guler & Nerkar, 2012; Hansen, 1999; Nerkar & Paruchuri, 2005; Reagans & McEvily, 2003). The focus of this body of research is mostly on the pattern of collaboration ties and its implications for innovation output at a point in time. On the other hand, how the composition of these ties varies over time and its effect on innovation outcomes is underresearched (Kumar & Zaheer, 2019). I add to the understanding of this neglected but important aspect of collaborative relationships by examining the stability of ties between inventors and how this may influence the transfer of knowledge between them.

The received wisdom from decades of research in social networks suggests that networks evolve and undergo changes (Baum, McEvily, & Rowley, 2012; Burt, 2001; Coleman, 1990). The stability of an inventor’s collaboration network, therefore, refers to the extent to which the ties in the inventor’s network vary (or not) from year to year (Burt & Merluzzi, 2016; Sasovova, Mehra, Borgatti, & Schippers, 2010). The sources of change in a collaboration network stem from inventors’ tendency to alter their collaborative ties. For example, an inventor can form ties with new inventors or sever existing ties with incumbent ones. The entry and exit of inventors within a firm may also introduce changes in an inventor’s network. Considering that the formation and cessation of ties are voluntary and driven by the research interests of inventors (e.g., Fleming et al., 2007; Liebeskind, Oliver, Zucker, & Brewer, 1996; Tushman & Romanelli, 1983), it can be argued that for the most part, the changes in inventors’ networks are likely deliberate. For example, an inventor can initiate new ties to gain access to diverse and fresh knowledge that lies outside of his or her current collaboration network (Ahuja, Soda, & Zaheer, 2012). On the other hand, if the inventor wants to acquire in-depth knowledge and build expertise in a particular research area, collaborating with the same set of inventors who are skilled in that research area for extended periods of time may be preferable (Gargiulo & Benassi, 1999). Sustained interactions among inventors also facilitate the smooth exchange of ideas and the development of routines that allow inventors to carry out their scientific activities more effectively.

Due to the voluntary nature of tie formation, tie-persistence mechanisms, and significant heterogeneity in the research interests of inventors, there is plausibly considerable variance in the stability of various inventors’ ties. This heterogeneity is likely to have implications for knowledge transfer and sharing in the present context. A stable collaboration network typically provides several advantages to an inventor. First, interacting regularly with the same set of inventors over time generates shared beliefs and mutual trust between them (Dyer & Singh, 1998; Gulati, 1995; Zaheer, McEvily, & Perrone, 1998). The prevalence of a common language among members with stable ties promotes the efficient transfer of knowledge between them. Second, stable ties curb the tendency of individual members to act opportunistically and provide favorable settings for members to share tacit and contextual knowledge freely (Doz, 1996; Hansen, 1999; Reagans & McEvily, 2003; Tortoriello & Krackhardt, 2010).

Due to the presence of mutual trust and collegial relationships between the members in networks characterized by stable ties, these networks offer an efficient mechanism for transferring tacit and complex knowledge (Ebadi & Utterback, 1984; Moran, 2005). The dynamic aspect of stable ties reveals an additional layer of the role of collaborative ties in knowledge transfer. To illustrate this point, consider two inventors who have the same collaboration networks in time t: that is, both inventors collaborated with n other inventors on k patents with the same collaborative patterns (e.g., the same density, number of ties, etc.). One may conclude in this case that since the pattern of ties in the networks of the two inventors is the same, the extent of knowledge transfer in their networks will be comparable. However, if one inventor had maintained ties with n inventors for the previous 5 years (t − 5 to t), whereas the other inventor had acquired ties with n inventors in just that same year (t), the stability of ties for the first inventor would be higher than that of the second inventor; this difference is likely to have implications regarding knowledge transfer for the reasons I discussed earlier. Essentially, in addition to the frequency of ties, the sustainability of ties over time between inventors is an important source of developing trust and cooperation (Moran, 2005).

If the departed inventors had stable collaborative ties, they sustained long-term collaborative relationships, which likely led to the exchange of tacit knowledge with their coauthors. I can thus expect the stable collaborative ties of departed inventors to partly offset the tendency of their departure to lead to a loss of tacit knowledge and the termination of their patents. The discussion leads to the following hypothesis:

Hypothesis 4: The stability of the collaborative ties of departed inventors negatively moderates the positive relationship between the inventors’ departure and the firm’s termination of their patents.

Data and Methods

I exploit patent data provided by the U.S. Patent and Trademark Office (USPTO) and use USPTO Classes 424 and 514 to identify patents in the pharmaceutical industry. The patents in these classes belonged to more than 200 firms. Because one aspect I examine involves the collaboration networks of inventors, I consider only firms that were active in patenting, as a smaller number of inventors could lead to an erroneous interpretation of their networks. Thus, I removed firms that did not produce at least one patent for 10 successive years between 1985 and 2000, reducing the sample size to 123 firms. After accounting for missing values for the variables used in the empirical analysis, the final sample contained 108,616 patents granted to 76,515 inventors in 102 firms. I restricted the year of analysis to 2000 to allow the patents enough time to accumulate forward citations, reducing concerns about right censoring on the dependent variable.

There were 15,498 turnover events, representing a turnover rate of 20.25%. The average employee mobility rate in the United States is above 20% but varies by industry (Bureau of Labor Statistics, 2016). For example, in 2006, the annual turnover rate in federal jobs was 9.3%, but in retail trade, it was much higher, at 34.7%. Although the turnover rate in the pharmaceutical industry falls near the middle part of the spectrum, its cost of turnover is relatively high. Because scientists have unique skills and expertise accumulated over time, it is difficult for firms to replace them quickly, leading to a loss of tacit knowledge and delays in product development.

Outcome Variable

The outcome variable is the likelihood that a patent will be terminated. In line with prior work (Khanna, Guler, & Nerkar, 2018; Serrano, 2010), I consider the nonrenewal of a patent after 4 years of its grant date to be termination. The USPTO provides information on all patents and their renewal status through maintenance fee events. A firm’s decision not to pay the maintenance fee results automatically in the termination of the patent. Past research argued that such decisions are voluntary and deliberate and represent firms’ attempts to streamline their portfolios to identify promising patents and cull worthless ones (Khanna et al., 2018; Moore, 2005). For a detailed review of the pharmaceutical industry and patent termination within the industry, please refer to Online Appendix A.

The dependent variable, patent termination, is a binary outcome with value 1 if the patent was terminated after 4 years (i.e., the firm did not renew it by paying maintenance fees) and 0 otherwise. There are three points at which a patent can be terminated: at 4, 8, and 12 years. I focus on the 4-year window to rule out factors in termination that may be specific to different periods (Khanna et al., 2018). Firms in the sample terminated 17% of the patents after 4 years, 19% after 8 years, and 15% after 12 years, leading to the total termination of 51% of the patents granted to them over the course of 12 years. I explore the timing of termination and its implications in some detail in the supplementary analyses reported in Online Appendix B (S5).

Independent and Moderating Variables

The independent variable is inventor turnover between 1985 and 2000. I use patent data to determine the movement of inventors between firms. To preclude erroneous cases of turnover, I set several conditions for the identification of turnover events. First, I consider the entire universe of patents and firms available in the patent data between 1985 and 2010. To identify inventor mobility events, I rely on a methodology suggested in earlier work (Agarwal et al., 2009; Corredoira & Rosenkopf, 2010) and use matching between the names of inventors and the names of firms. Please refer to Online Appendix C for more details on the matching process. I consider only cases in which the application year for a patent by an inventor at a new firm is before the year in which the termination decision is made for the inventor’s patents at the old firm. Thus, for each patent observation p, the independent variable, inventor turnover, is binary and equal to 1 if at least one inventor of p has moved, having applied for a patent at another firm before the year t in which the old firm takes termination decision about p, and 0 otherwise.³

Second, firms allow their inventors to collaborate with others outside the firm. Therefore, it is possible for an inventor to have collaborated with people at a firm without having moved to that firm. To account for this, I consider only cases in which the inventor who received the patent at the new firm never had any more patents with the old firm.

It is also plausible that patent data misrepresent inventor turnover for other reasons. To address this concern, I manually checked 150 turnover events using a search engine (Google), inventors’ pages, and a professional networking platform (LinkedIn). Of the 150 cases, only six of the cases had been misidentified. I repeated this procedure with another randomly selected 150 turnover events and found only four errors. The small number of errors in both samples indicates that misidentified cases do not seem to be a cause for concern and are not likely to contaminate the empirical analysis in any systematic or significant way.⁴

I measure the moderating variables using the collaborative ties of inventors who have left their firms. I consider only the collaborative ties of departed inventors with other inventors in the same firm who did not leave. For example, if inventor i departs firm j in year t, I consider i’s collaborative ties with other inventors at j in the previous 5 years who did not leave j until the patents of i faced termination decision. Although departed inventors may have collaborated with inventors outside their firms, I do not consider those ties because outside inventors are not likely to contribute to the firm’s subsequent knowledge creation activities. In other words, a firm is unlikely to rely on the external collaborators of its departed inventors to compensate for the loss of their tacit knowledge.⁵

All moderating variables are calculated from the first-degree collaborative ties of the departed inventors because I believe that these ties are the most critical in knowledge transfer, especially for tacit knowledge. Thus, if inventor i departed firm j in year t, and collaborated with n other inventors in j between t − 5 and t − 1, I focused on the characteristics of the network comprising the collaborative ties between i and these n coauthors.⁶

The first moderating variable is knowledge overlap in the collaborative ties of departed inventors. Following tradition and staying true to the idea of recombination in the article, I measure knowledge overlap between inventors using the extent to which inventors draw from the same pools of technological knowledge, that is, citations to patents in their patents (Jaffe et al., 1993; Sears & Hoetker, 2014). The approach to measure the knowledge overlap is multifold (Mowery, Oxley, & Silverman, 1996). Consider inventor i who departed in year t and collaborated with n inventors in 5 years before t, that is, between t − 5 and t − 1. First, I identify the patents that are cited in the patents of both i and i’s coauthor j between t − 5 and t − 1.⁷ I refer to them as common patents. Second, I estimate the proportion of the number of times these common patents are cited by i to the total number of citations in i’s patents; I repeat this step for coauthor j. Third, I add these two ratios to estimate the knowledge overlap between i and j. I reiterate these three steps for all n coauthors of i and take the average of these pairwise knowledge overlaps between i and her or his coauthors to estimate the moderating variable, knowledge overlap. Formally,

\begin{array}{l} K n o w l e d g e o v e r l a p_{i, j, t} = \frac{C i t a t i o n s i n i, s p a t e n t s t o c o m m o n p a t e n t s_{t - 5, t - 1}}{T o t a l c i t a t i o n s i n i, s p a t e n t s_{t - 5, t - 1}} \\ + \frac{C i t a t i o n s i n j, s p a t e n t s t o c o m m o n p a t e n t s}{T o t a l c i t a t i o n s i n j, s p a t e n t s} . \end{array}

(1)

K n o w l e d g e o v e r l a p_{i, t} = \frac{\sum_{j \in n} K n o w l e d g e o v e r l a p_{i, j, t}}{n} .

(2)

Equation (1) measures the knowledge overlap between i and i’s coauthor j, and Equation (2) calculates the knowledge overlap between i and i’s coauthors by calculating the mean value of i’s knowledge overlap with her or his n coauthors. The knowledge overlap measure in Equation (2) captures the degree to which inventor i and her or his coauthors draw from the same technological knowledge pool, and a higher value of the measure indicates a greater degree of overlap between technological knowledge of i and i’s coauthors.

The second moderating variable, strength of ties, captures the strength of the departed inventor’s collaborative ties. This measure derives from the intensity of the interactions among the individuals in various relationships (Granovetter, 1973; Marsden & Campbell, 1984). I borrow from previous research and measure the strength of ties among coauthors by the frequency of their collaborations (Hansen, 1999; McFadyen & Cannella, 2004; Uzzi, 1996). I count the number of patents the departed inventor i had with others in firm j and average this number with the number of unique inventors with whom i collaborated. As before, I consider i’s collaborations in the 5 years before the year of departure,

S t r e n g t h o f t i e s_{i, j, t} = \frac{n u m b e r o f i, s p a t e n t s_{t - 5, t - 1}}{\begin{array}{l} n u m b e r o f u n i q u e i n v e n t o r s \\ w i t h w h o m i c o l l a b o r a t e d_{t - 5, t - 1} \end{array}} .

(3)

Equation (3) estimates the mean number of patents per inventor that i was granted in firm j during the 5 years before departing in year t. It takes into account the importance of inventors’ working together repeatedly, returning higher values when i collaborated with the same inventor on multiple patents than when i collaborated with multiple inventors on the same number of patents.

The last moderating variable, stability of ties, is measured using changes in the composition of inventors’ ties over time. This approach is similar to the one used in previous research (Burt & Merluzzi, 2016; Kumar & Zaheer, 2019; Sasovova et al., 2010). The operationalization of the measure involves two steps. In the first, I calculate the churn in the collaborative ties of departed inventor i by adding the number of collaborations with new inventors and the number of discontinued collaborations with old inventors and dividing the sum by the number of unique collaborators i had. This approximates the number of changes in i’s collaborative ties, with high values representing more changes.

In the next step, I subtract the measure of churn from (1) to calculate the stability of ties. This is computed annually but using a moving average of the previous 5 years to allow for variation across time periods. Thus, stability of ties measures the percentage of ties that stay the same from t1 to t2. As before, it is calculated for inventor i who departed firm j in year t:

C h u r n_{i, j, t} = \frac{\begin{array}{l} t i e s a d d e d w i t h n e w i n v e n t o r s_{t - 2, t - 1} \\ + t i e s d i s s o l v e d w i t h o l d i n v e n t o r s_{t - 2, t - 1} \end{array}}{n u m b e r o f i, s u n i q u e c o - i n v e n t o r s_{t - 1}}

(4)

S t a b i l i t y o f t i e s_{i, j, t} = \frac{\sum_{y = t - 4 t o t} (1 - C h u r n_{i, j, y})}{5}

(5)

These equations estimate i’s stability of ties annually and average them over the past 5 years. For illustration purposes, I provide examples of the collaborative ties of three inventors who left their firms in 2001. Figures 1 to 3 depict the collaborative ties and the values of all moderating variables. For comparison purposes, I selected inventors who collaborated with the same number of inventors (n = 5) in the 5-year period before their departure (1995–2000). The squares represent inventors who departed in 2001, the circles their coauthors, and numbers inside circles the number of patents between departed inventors and their respective coauthors. As Figures 1 and 2 show, the strengths of the collaborative ties of i1 and i2 are the same, but their knowledge overlap and stability of ties differ considerably. Similarly, i2 and i3 (Figures 2 and 3) have the same levels of knowledge overlap but significantly different in strengths and stabilities of ties. Last, i1 and i3 (Figures 1 and 3) have comparable stability but vary greatly in knowledge overlap and strength. These sample networks illustrate that the three characteristics of collaborative ties I examine are distinct from each other. This is further evident by the low correlation values among these variables, as shown in Table 2.

Figure 1

Collaborative Ties of i1 Between 1995 and 2000

Figure 2

Collaborative Ties of i2 Between 1995 and 2000

Figure 3

Collaborative Ties of i3 Between 1995 and 2000

In addition, I conduct pairwise discriminant validity tests among the three moderating variables used in this study. The approach to establish discriminant validity involves three steps (J. C. Anderson & Gerbing, 1988; Bagozzi & Phillips, 1982; Bagozzi, Yi, & Phillips, 1991; Hoskisson, Hitt, Johnson, & Moesel, 1993). First, for a pair, I calculate the χ² statistic from a confirmatory factor analysis (CFA) model in which the covariance between the variables in the pair is allowed to vary. Second, I estimate the χ² statistic from the CFA model in which the covariance between the variables is fixed at 1. In the last step, I calculate the difference between the χ² statistics (i.e., Δχ²) obtained from the two CFA models. I repeat these steps for all three pairs of the moderating variables, and the results show that the Δχ² for each pair is significantly larger (>45) than the χ² value of 6.64 at p < .01 for 1 degree of freedom. This analysis further establishes that the measures in this study are capturing distinct constructs. The 95% confidence interval using Fisher’s transformation for each pairwise correlation between the moderating variables contained 0, supporting the results from the discriminant analysis.⁸

Control Variables

I include several control variables that may act as confounders in regression analyses in this study. First, the number of citations to a patent can be a valid indicator of its value, and prior work suggests that patents with relatively large numbers of citations are less likely to be terminated (Serrano, 2010). Next, the number of claims to a patent can be another proxy for its future value (Lanjouw & Schankerman, 2004; Tong & Frame, 1994). Thus, I control for number of citations and claims to each patent in the analysis.

Pharmaceutical firms often collaborate with foreign firms for access to knowledge and resources that are unavailable in their home countries (Kobrin, 1991). Therefore, the presence of inventors from foreign countries can influence termination decisions, as those involve the interests of such individuals and firms. To account for this, for each patent-year observation, I construct a binary variable, foreign, that takes the value 1 for a given patent if there is at least one inventor on the patent who was not from the United States and 0 otherwise.

The number of inventors on a patent can also influence a firm’s termination decision, as inventors who stayed behind can compensate for the knowledge lost in turnover. To account for this, I include number of inventors for each patent in the model. The way firms conduct R&D can also be endogenous to the decision to terminate a patent. Some firms generate high R&D output and also terminate more patents, and vice versa. To account for firm-level heterogeneity in the production and termination of patented inventions, for each firm, I include the stock of patents granted to it (R&D productivity) and the number of patents terminated by it (total terminations) each year.

Prior work has argued that a termination decision may also be associated with the scope of its search efforts (Li & Chi, 2013). Whether a firm explores diverse technologies or focuses on a narrow set of technologies may affect its decisions to terminate these technologies. And technological diversity has a significant influence on a firm’s knowledge-building activities (Tzabbar, 2009) and may affect its decisions to terminate patents in a certain area. To account for this, I calculate technological diversity (Herfindahl) at the firm level as 1 minus the sum of the squares of the proportions of a firm’s patents across subclasses.

The most prevalent mode of collaboration between pharmaceutical firms is alliances (Sampson, 2007). The alliances firms create to drive their R&D will influence their termination decisions, too, as some inventions could be the result of collaborations between firms. Thus, I include the number of new alliances (alliances) made by each firm each year in the model.

The competition in an area of technology can also affect termination decisions, as firms could hold on to some patents longer or terminate them sooner due to competitive pressure (Khanna et al., 2018). To account for these dynamics, for each patent, I calculate the number of firms that have patents in the same technologies. The variable degree of competition captures the average number of firms with patents in the subclasses underlying the focal patent. I also control for the interdependency of patents, as prior work has shown that the interdependency of the subclasses underlying a patent influences its likelihood of termination (Khanna et al., 2018). I follow past work and measure interdependency of a patent by counting the times the subclasses of a patent were combined with other subclasses and dividing this by the total number of subclasses underlying the patent (Fleming, 2001; Fleming & Sorenson, 2004).

Next, I control for the recombinant potential of the technologies underlying each patent. This is the subclass’s potential for being combined with other subclasses to generate useful inventions (Fleming & Sorenson, 2004). Because a patent in subclasses characterized by high recombinant potential is more likely to be used in future inventions, a firm is less likely to terminate it. I calculate a patent’s recombinant potential by estimating the average of the recombinant potentials of its subclasses. The recombinant potential of a subclass is measured by the proportion of the number of other subclasses combined with it to the total number of patents in the subclass (Fleming & Sorenson, 2001).

Prior work has shown that the presence of a star inventor on a patent significantly reduces its likelihood of termination (Liu, 2014). For each patent, I create a dummy variable, star, that is coded 1 if a patent has a “star” among its inventors and 0 otherwise. I borrow from previous work (Rothaermel & Hess, 2007; Zucker & Darby, 2001) to identify stars. First, I calculate the citation-weighted patents for each inventor annually. Second, I sum the citation-weighted patents for the previous 5 years for all inventors. I categorize as a star any inventor who is above the 90th percentile with respect to the sum calculated in the second step.

Last, I control for network characteristics, such as reach, range, and structural holes of departed inventors. While I argued earlier that these properties might not be important in the transfer of tacit knowledge and, therefore, may not influence the dependent variable, patent termination, as a precaution, I included them as controls in all models. Network reach refers to the average lengths of paths between individuals in a network. An inventor with higher reach in her or his network will be able to reach other inventors using shorter paths. I measure the reach of a departed inventor’s collaboration network as the sum of the inverse of geodesic distances to all inventors in her or his firm in 5 years prior to her or his departure (Watts, 1999). Following tradition, I measure the range of departed inventors’ collaboration networks using their information centrality (Brandes & Fleischer, 2005; Stephenson & Zelen, 1989). The measure of the range is identical to reach in that it also considers geodesic distances of departed inventors from other inventors but differs from reach as it gives higher weights to shorter paths. To measure the span of structural holes of departed inventors, I use Burt’s (1992) effective size measure, which calculates the number of nonredundant ties in inventors’ networks.⁹ Specifically, the measure counts the number of ties of a departed inventor with her or his coauthors in 5 years prior to her or his departure and subtracts the average number of ties between her or his coauthors in the same period.

Estimation Strategy and Potential Endogeneity

The goal of the empirical work is to understand the influence of inventor turnover on the likelihood of patent termination. One could argue that an inventor’s decision to leave and a firm’s decision to terminate the patents of the inventor are related; for instance, the departure may be due to the belief that the patents have little value and will be terminated. A firm might thus have decided to terminate a patent for reasons other than its inventor’s departure, and that decision may have led to the departure. Using the lagged values of the independent variable is one way to resolve the problem of reverse causality in this case, but due to the possibility of measurement error associated with variables, it may not be a sound approach (Reed, 2015). To gauge the endogeneity in the model, I conducted the Hausman specification test, and the p value for the independent variable, inventor turnover, was .076 (<.10), providing weak evidence of endogeneity in the sample.

To overcome the endogeneity in the data, I estimate models using the Heckman correction procedure (Heckman, 1979). The method involves two steps. The first is the estimation of the likelihood of inventor turnover using instruments and other covariates. In the second step, the predicted or instrumentally measured values generated in the first step are added, along with the covariates used in the first model.

I use multiple instrumental variables in the two-stage model, as they provide more accurate estimates of the coefficients (Angrist & Keueger, 1991). I rely on prior work to select the instruments (Carnabuci & Operti, 2013; Tzabbar & Kehoe, 2014). First, I use changes in top management teams as a predictor of inventor turnover. Previous work has shown that CEO turnover can disrupt routines and set a new direction for a firm’s R&D activities, potentially causing the departure of inventors. To measure the first instrument, I track whether firms appointed a new CEO during the study period. I collected these data from press releases by the firms, as listed in LexisNexis. In line with previous work, I use a lag of 3 years for the variable new CEO to account for the time the CEO takes to influence the research agendas and routines in the firm (Shen & Cannella, 2002).

Second, the entry of new inventors to a firm could disrupt collaborative activities and lead to a lengthy and uncomfortable adjustment of routines for incumbent inventors (D. Allen, 2006). Thus, for each year, I use the number of inventors in a firm (number of newcomers) who have not previously filed patents with that firm as the second instrument (Carnabuci & Operti, 2013).

Last, a firm’s interactions with external entities in the form of alliances and mergers and acquisitions (M&As) can create negative experiences for its inventors due to incompatibilities in R&D routines and organizational culture (Kapoor & Lim, 2007; Paruchuri, Nerkar, & Hambrick, 2006). Thus, I use the numbers of alliances and M&A events for each firm as two additional instruments. I use the Securities Data Company database to obtain this information. The evidence from Sargan statistics (Sargan, 1958) and Hansen’s J statistic (Hansen, 1982) support the exogeneity of the instruments (both p values > .40), conferring more confidence in the regression model. In addition, the minimum eigenvalue statistic from the Anderson canonical correlation test is larger than the critical value of 16.38 for all instruments, letting me reject the null hypothesis that the instruments are weak.

There could be two more estimation issues with the model. First, three or four instruments are recommended for each endogenous variable, and in addition to being exogenous, as investigated in the previous analysis, these instruments must also be dissimilar—that is, they should not be multicollinear (Wooldridge, 2015). To test the multicollinearity between the instruments, I used the difference-in-Sargan test (Hayashi, 2000). The test found no evidence of multicollinearity, as the p values for all the instruments were above .10. Second, not only the independent variable but other covariates in the model could be endogenous. To test whether this is a problem, I ran a difference-in-Sargan test again on each covariate in the first stage instrumental regression model. The p values for all the covariates were above .10, so the possibility of endogeneity can be ruled out with reasonable confidence (Bascle, 2008; Hayashi, 2000).

Because both the outcome variable (patent termination) and the endogenous independent variable (inventor turnover) are limited and binary in nature, I use the maximum-likelihood probit model with sample selection (Van de Ven & Van Pragg, 1981). The goal of the procedure is to correct for selection bias, generating consistent and asymptotically efficient estimates for parameters in the model. The equations that test the hypotheses take the following form:

Selection equation:

\begin{array}{l} I n v e n t o r t u r n o v e r_{i, t} = γ_{0} + γ_{1} n e w C E O_{i, t - 3} + γ_{2} n e w c o m e r s_{i, t - 3} \\ + γ_{3} n u m b e r o f a l l i a n c e s_{i, t - 3} \\ + γ_{4} n u m b e r o f M & A s_{i, t - 3} + \sum_{k = 1}^{n} γ_{k} Z_{i, t - 5, t - 1} + ε_{i, t} . \end{array}

(6)

Probit equation:

P a t e n t t e r m i n a t i o n_{i, t} = G (β_{0} + i n v e n t o r t u r n o v e r_{i, t} + \sum_{k = 1}^{n} β_{k} Z_{i, t - 5, t - 1} + μ_{i, t}) .

(7)

where $Σ_{k = 1}^{n} γ_{k} Z_{i t}$ is a vector of moderating and exogenous variables.

The analysis is conducted at the patent level. The dependent variable, patent termination, takes the value 1 if the patent was terminated after 4 years of its grant date and 0 otherwise. The independent variable, inventor turnover, takes the value 1 if at least one inventor on the patent has left the firm and 0 otherwise. The moderating variables measure the characteristics of the collaborative ties of inventors who have left a firm. In cases where multiple inventors on a given patent have left, I used the mean values of all moderating variables. The panel of patent-year data contains 108,616 observations for 102 pharmaceutical firms for 1985 to 2000.

Results

Tables 1 and 2 present the descriptive statistics and correlation matrix of the variables used in the empirical analysis. The correlations between variables do not seem high enough to raise concerns, but as a precaution, I checked the variation inflation factor of the full model, including the interaction terms, and it was under 5.

Table 1

Descriptive Statistics

Variable	n	M	SD	Min.	Max.
Patent termination	108,616	0.17	0.36	0	1
Inventor turnover	108,616	0.19	0.26	0	1
Knowledge overlap	108,616	0.28	0.40	0	2
Strength of ties	108,616	1.48	0.66	1	13.6
Stability of ties	108,616	0.66	0.18	0	1
Inventor Turnover × Knowledge Overlap	108,616	0.06	0.15	0	1.84
Inventor Turnover × Strength of Ties	108,616	0.29	0.26	0	13.6
Inventor Turnover × Stability of Ties	108,616	0.14	0.45	0	1
Number of citations	108,616	6.16	9.67	0	686
Number of claims	108,616	14.43	14.61	1	868
Foreign	108,616	0.57	0.48	0	1
Number of inventors	108,616	3.75	2.30	1	33
R&D productivity	108,616	494	443	9	2380
Total terminations	108,616	15.06	26.42	1	171
Herfindahl	108,616	0.13	0.23	0.04	1
Alliances	108,616	1.58	1.70	0	19
Degree of competition	108,616	18.77	23.07	1	41
Patent interdependency	108,616	0.02	0.04	0	1
Recombinant potential	108,616	0.28	0.14	0	1
Star	108,616	0.22	0.39	0	1
Reach	108,616	19.12	34.76	0	253.7
Range	108,616	2.49	0.53	0	53.2
Structural holes	108,616	0.23	0.34	0	1
New CEO	76,515	0.08	0.12	0	1
Number of newcomers	76,515	1.93	2.27	0	33
Number of M&As	76,515	0.44	0.79	0	7

Note: R&D = research and development; M&A = mergers and acquisitions.

Table 2

Partial Correlation Table

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25
1. Patent termination	—
2. Inventor turnover	.24	—
3. Knowledge overlap	(.13)	.08	—
4. Strength of ties	(.40)	(.30)	.24	—
5. Stability of ties	(.05)	(.14)	.36	.15	—
6. Inventor Turnover × Knowledge Overlap	(.17)	.41	.39	(.08)	(.06)	—
7. Inventor Turnover × Strength of Ties	(.09)	.29	.11	.47	.12	.43	—
8. Inventor Turnover × Stability of Ties	(.04)	.36	(.03)	.07	.44	.46	.39	—
9. Number of citations	(.21)	.24	.07	.22	(.02)	.04	.24	(.01)	—
10. Number of claims	(.07)	.27	.09	.31	.10	.10	.15	.19	.17	—
11. Foreign	.07	(.01)	.21	.08	(.05)	.15	.01	(.04)	.19	.11	—
12. Number of inventors	(.24)	.03	(.09)	.21	.17	(.11)	.30	.25	.05	(.03)	.04	—
13. R&D productivity	.05	(.02)	.01	.10	(.09)	.01	.18	(.15)	(.37)	.06	(.03)	.08	—
14. Total terminations	.02	.05	(.02)	.10	(.17)	(.01)	.18	(.11)	.21	.10	(.18)	.09	.10	—
15. Herfindahl	(.03)	(.05)	.01	(.02)	.04	.00	(.01)	.02	(.27)	(.05)	.04	.14	(.32)	(.21)	—
16. Alliances	.03	.07	.01	.03	(.12)	.00	.05	(.03)	(.17)	(.11)	.05	(.13)	(.17)	.21	.07	—
17. Degree of competition	.01	.10	.03	.05	(.04)	.01	.08	(.06)	(.02)	(.09)	.07	(.23)	(.05)	.21	.14	.19	—
18. Patent interdependency	.01	.19	.04	.02	(.03)	.02	.03	(.03)	.01	(.05)	.09	(.11)	(.02)	.26	.10	.37	.14	—
19. Recombinant potential	(.00)	.19	.10	.00	(.02)	.11	.04	(.00)	.00	(.00)	.29	(.29)	(.03)	.07	.03	.39	.05	.29	—
20. Star	(.45)	.01	(.02)	.27	.14	(.01)	.04	.09	.17	.05	(.05)	.11	(.07)	(.09)	.27	.21	.02	.06	.15	—
21. Reach	.01	.12	.11	.06	(.04)	.01	.08	(.07)	(.03)	.10	.09	(.19)	.04	.21	(.03)	.04	.15	.21	.15	.05	—
22. Range	(.09)	.27	.09	.33	.13	.15	.37	.18	.20	.09	.01	.05	(.04)	(.02)	.09	.08	(.18)	.06	.27	.09	.33	—
23. Structural holes	(.08)	.29	.06	.30	.11	.07	.14	.15	.12	(.20)	.04	.11	.18	.13	.05	(.03)	.03	.08	(.21)	.24	.12	.24	—
24. New CEO	.06	.11	(.24)	.04	.30	.00	.08	(.07)	(.03)	.01	(.04)	.21	(.01)	(.02)	.29	(.01)	(.03)	.07	.34	(.05)	(.03)	.20	.03	—
25. Number of newcomers	.13	.04	.02	(.05)	(.02)	.11	.01	.24	.18	.02	.01	.08	.08	(.07)	(.04)	(.12)	(.06)	(.10)	.09	(.28)	.01	(.05)	(.07)	(.00)	—
26. Number of M&As	.17	.09	.11	.21	.17	.07	.27	.11	(.08)	.16	(.13)	(.03)	.20	(.00)	(.01)	.11	(.05)	(.01)	(.11)	.24	.00	(.09)	(.00)	(.04)	.28

Note: R&D = research and development; M&A = mergers and acquisitions.

Table 3 reports the results for the second-stage probit model that was used to test the hypotheses. Model 1 is the baseline model and includes only control variables. The influence of the controls on the dependent variable is largely as expected. The number of citations is negatively associated with the likelihood of termination (Serrano, 2010). High R&D productivity is also associated with a high likelihood of termination of a patent, meaning firms that produce more patents also tend to terminate more patents. Finally, patents produced by star inventors are less likely to be terminated, a finding consistent with prior research (Liu, 2014).

Table 3

Second Stage Probit Estimates (Dependent Variable = Likelihood of Patent Termination)

Variable	Model 1		Model 2		Model 3		Model 4		Model 5
Variable	Coeff. (SE)	p	Coeff. (SE)	p	Coeff. (SE)	p	Coeff. (SE)	p	Coeff. (SE)	p
Inventor turnover			0.55 (0.018)	.002	0.59 (0.31)	.057	0.58 (0.12)	.000	0.50 (0.12)	.000
Inventor Turnover × Knowledge Overlap					−0.16 (0.015)	.000	−0.04 (0.02)	.012	−0.10 (0.05)	.045
Inventor Turnover × Strength of Ties							−0.05 (0.01)	.000	−0.08 (0.03)	.007
Inventor Turnover × Stability of Ties									−0.04 (0.00)	.008
Knowledge overlap	0.17 (0.24)	.483	0.09 (0.30)	.763	0.13 (0.1)	.198	0.15 (0.16)	.342	0.04 (0.09)	.678
Strength of ties	−0.07 (0.03)	.009	−0.09 (0.04)	.051	−0.09 (0.06)	.137	−0.04 (0.08)	.612	−0.02 (0.1)	.848
Stability of ties	−0.14 (0.25)	.565	−0.04 (0.42)	.923	−0.04 (0.62)	.943	−0.06 (0.38)	.871	−0.04 (0.45)	.930
Number of citations	−0.02 (0.01)	.098	−0.03 (0.01)	.009	−0.01 (0.00)	.013	−0.01 (0.00)	.000	−0.01 (0.00)	.000
Number of claims	−0.06 (0.00)	.000	−0.05 (0.00)	.000	−0.02 (0.00)	.000	−0.03 (0.00)	.000	−0.04 (0.00)	.000
Foreign	−0.08 (0.11)	.504	−0.08 (0.14)	.578	−0.07 (0.11)	.526	−0.04 (0.06)	.518	−0.07 (0.02)	.001
Number of inventors	−0.02 (0.00)	.000	−0.02 (0.00)	.000	−0.03 (0.00)	.000	−0.02 (0.00)	.000	−0.01 (0.00)	.000
R&D productivity	0.00 (0.01)	.001	0.00 (0.02)	.004	0.00 (0.02)	.000	0.00 (0.01)	.000	0.00 (0.02)	.000
Total terminations	0.00 (0.01)	.887	0.00 (0.01)	.786	0.00 (0.00)	.992	0.00 (0.00)	.881	0.00 (0.00)	.786
Herfindahl	0.11 (0.01)	.000	0.10 (0.01)	.000	0.15 (0.00)	.000	0.05 (0.00)	.000	0.03 (0.00)	.000
Alliances	−0.09 (0.37)	.804	−0.12 (0.63)	.856	−0.19 (0.8)	.809	−0.06 (0.49)	.902	−0.02 (0.41)	.969
Degree of competition	−0.03 (0.11)	.824	−0.04 (0.16)	.809	−0.01 (0.15)	.927	−0.01 (0.09)	.886	−0.02 (0.12)	.889
Patent interdependency	−2.86 (1.62)	.079	−3.03 (0.62)	.000	−3.89 (0.87)	.000	−1.65 (1.47)	.260	−1.45 (1.79)	.418
Recombinant potential	−0.09 (0.01)	.000	−0.14 (0.01)	.000	−0.18 (0.00)	.000	−0.30 (0.00)	.000	−0.13 (0.00)	.000
Star	−0.19 (0.06)	.001	−0.26 (0.05)	.000	−0.24 (0.07)	.001	−0.11 (0.11)	.326	−0.12 (0.19)	.519
Reach	−0.18 (0.12)	.128	−0.08 (0.08)	.316	−0.02 (0.04)	.514	−0.10 (0.13)	.477	−0.03 (0.17)	.859
Range	0.29 (0.19)	.124	0.08 (0.18)	.670	0.06 (0.06)	.324	0.07 (0.19)	.716	0.05 (0.11)	.641
Structural holes	−0.08 (0.04)	.061	−0.13 (0.06)	.035	−0.08 (0.05)	.170	−0.06 (0.06)	.339	−0.07 (0.08)	.383
Log likelihood	−262.41		−234.07		−220.03		−179.54		−168.05
χ_p²	0.000		0.000		0.000		0.000		0.000

Note: N = 108,616 for 102 firms. Robust standard errors clustered at the firm level are in parentheses. All models include year and firm fixed effects. Coeff. = coefficient; R&D = research and development.

Model 2 in Table 3 tests the first hypothesis and finds support for it. The coefficient on the variable inventor turnover is significant at a p value of .002. The result indicates that on average, firms were 36% more likely to terminate the patents of departed inventors than of inventors who stayed. This result provides a strong indication of the effect of inventor turnover on patent termination from both statistical and practical perspectives.

Next, I incorporate the interaction of inventor turnover and knowledge overlap into the regression model. The results are reported in Model 3 in Table 3. The interaction term is negative at a p value <.000, providing support for the second hypothesis. Calculations based on the marginal effects of parameters reveal that for a one-standard-deviation increase above the mean in the knowledge overlap in the collaborative ties of departed inventors, the likelihood of their patents’ being terminated declines by 8%. The results point to a considerable drop in the likelihood of termination of patents whose inventors had a high degree of knowledge overlap with their coauthors. The drop of 8% in termination likelihood varies between 6% and 11% for 99% confidence interval (t value = 2.576).

To further explore the findings, I plot the relationship between inventors’ departures and the likelihood of termination of their patents at different levels of knowledge overlap in the inventor’s collaborative ties: low and high. I consider knowledge overlap low if the knowledge overlap in an inventor’s collaborative ties is 0 and high if it is a standard deviation above the mean (= 0.68). As Figure 4 shows, the likelihood of termination of an inventor’s patents increases from 18% to 28% following the inventor’s departure when knowledge overlap is low. By contrast, the inventor’s departure leads to an increase in the likelihood of termination, from 17.5% to 21%, when knowledge overlap is high. These results together indicate that high knowledge overlap in collaborative ties enables the retention of tacit knowledge of departed inventors in their coauthors, resulting in a smaller increase in the termination likelihood of their patents.

Figure 4

Relationship Between an Inventor’s Turnover and the Likelihood of Termination of Patent(s) of the Inventor at Different Levels of Knowledge Overlap in the Inventor’s Collaboration Network

The third hypothesis is that the strength of ties in departed inventors’ collaborative networks moderates the relationship between their departure and the termination of their patents. To test this, I interact variables strength of ties and inventor turnover. I report the results in Model 4 in Table 3. The coefficient on the interaction term is negative, with a p value <.000, providing strong support for Hypothesis 3. As before, I graph the interaction at conditional values of the strength of ties: low and high. As shown in Figure 5, the likelihood that firms will terminate patents of departed inventors decreases significantly when the inventors had strong ties with their coauthors. Specifically, inventors’ departure increased the likelihood of termination of their patents from 17% to 27% when the strength of ties was low, that is, at the minimum (= 1). By contrast, when the strength of ties was high—a standard deviation above the mean (= 2.14)—the likelihood of termination increased from 16% to 20%.

Figure 5

Relationship Between an Inventor’s Turnover and the Likelihood of Termination of Patent(s) of the Inventor at Different Levels of Strength of Ties in the Inventor’s Collaboration Network

Finally, I argue that the stability of ties in departed inventors’ collaboration networks moderates the relationship between inventor turnover and termination of patents. To examine this effect, I interact stability of ties with the independent variable, inventor turnover. Model 5 in Table 3 reports the results of the analysis. The coefficient on the interaction term is negative, at a p value of .008, conferring support for the fourth hypothesis. Figure 6 presents the interactive effect of stability and inventor turnover on the probability of termination of inventors’ patents. I consider the stability of ties to be low when the stability is at the minimum (= 0) and high when it is a standard deviation above the mean (= 1.84). Firms are less likely to terminate the patents of departed inventors when those inventors had more stable ties with their coauthors.¹⁰ Overall, the results support the hypotheses and arguments and suggest that inventors’ departures and collaborative ties play important roles in firms’ termination decisions. I perform several robustness tests, presented in the online appendix, to validate the findings of the main analysis.

Figure 6

Relationship Between an Inventor’s Turnover and the Likelihood of Termination of Patent(s) of the Inventor at Different Levels of Stability of Ties in the Inventor’s Collaboration Network

Discussion and Conclusion

Due to the uncertainty inherent in the process of R&D in knowledge-intensive industries, many firms hold a portfolio of inventions. However, it is not practical for firms to develop all of these inventions due to constraints related to time and resources (Fleming & Sorenson, 2004; Stuart & Podolny, 1996). The process intrinsically requires firms to make choices with respect to which inventions to select for recombination and which ones to terminate. Prior work has argued that firms use patent renewal as a tool to streamline their portfolios of inventions and guide their scientific efforts in a fruitful direction (Moore, 2005). The interviews with attorneys at pharmaceutical firms alluded to similar motives behind these decisions.

The process of recombination, however, is not straightforward, and its success depends on the tacit know-how and deep experience of inventors (Fleming, 2001). I argue that the turnover of inventors is likely to result in the loss of these inventors’ tacit knowledge, and the results indicate that this leads to decisions within firms to terminate their patents. While the departed inventor no longer produces knowledge in the form of patents for the source firm, the patents that the inventor generated before departing are also more likely to be terminated. This is an interesting result, as it provides direct evidence of how the explicit knowledge contained in patents and the tacit knowledge of their inventors are complementary to each other; it also shows how the absence of one of these may adversely affect the value that firms can create from their knowledge assets, such as patents. The results from the supplementary analyses in Online Appendix B (S4) support this finding and reveal the lack of use of departed inventors’ patents in the creation of future patents within firms. Overall, the findings from this research illustrate that in knowledge-intensive industries, the consequences of turnover of employees for source firms can be far-reaching and long-lasting. Given that firms hold a portfolio of inventions, the management or deployment of these knowledge assets to generate value is at the center of creating competitive advantage (e.g., Boisot, 1998; Bowonder & Miyake, 2000; Chen & Edgington, 2005). While there is value in owning the rights to patents (resource bundles), what is perhaps even more valuable is the ability to use these patents as inputs to subsequent knowledge-generation processes (resource orchestration) (Sirmon et al., 2011). The insights from this study inform us about what factors determine a firm’s ability to effectively manage its knowledge portfolio under conditions of uncertainty.

The article makes an important contribution to the current literature in this field by advancing our understanding of how firms manage their knowledge-related resources, such as patents. The questions of termination and exit are central to strategy and have been gaining increasing scholarly attention in recent years (e.g., Elfenbein & Knott, 2015; Guler, 2018; Sleesman, Lennard, McNamara, & Conlon, 2018). Researchers have proposed several explanations of why firms terminate certain projects (Guler, 2007; Joseph, Klingebiel, & Wilson, 2016; Serrano, 2010). Most of these explanations revolve around the quality of the projects, as indicated by their technical aspects and future economic returns. This study offers an alternative outlook to inform this literature and argues that a firm’s decision to terminate an invention may depend on its ability to effectively combine the explicit knowledge contained in the invention and the tacit knowledge held by the inventor. The results from the supplementary analyses (presented in Online Appendix B [S2]) support this thesis and show that while firms are less likely to terminate high-quality patents, that is, patents with more citations, the likelihood of their termination increases significantly when the inventors of these patents no longer remain employed by the firms. Thus, by highlighting how firms approach termination decisions to shape their portfolio of inventions, this study contributes to a more nuanced understanding of the process of value creation from knowledge assets.

The findings in the article generate several additional insights that have important implications for existing research. A long line of management scholars has argued about the effects of human and relational capital on organizational outcomes (Adler & Kwon, 2002; Becker, 1975; Coff, 2002; Hatch & Dyer, 2004; Nahapiet & Ghoshal, 1998). Recent work in employee mobility has called for disentangling the organizational impacts of human and relational capital (Mawdsley & Somaya, 2016). This article takes steps in this direction and presents an interesting interplay between the human and relational capital of mobile inventors. I show that while the human capital loss in the form of tacit knowledge decreases the value of patents, the relational capital embedded in collaborative ties helps to compensate for that loss. By doing so, the article is able to systematically make a distinction between these two dimensions and underscore their heterogeneous effects on an important organizational outcome: termination decisions.

The attempt to separate the effects of the human and relational capital of inventors on termination decisions has important normative implications for organizations. Following the departure of inventors, firms lose their human capital in the form of inventors’ skills, knowledge, and expertise. However, the relation capital of these inventors may not be lost entirely. Based on the findings in the article, firms can put certain measures in place to limit the negative aftermaths of turnover events. For example, if certain patents are important to a firm in advancing its future R&D endeavors, the firm can facilitate collaborations between inventors whose knowledge overlaps with the knowledge of the inventors of these patents. In addition, the firm may also try to ensure that there are sustained (higher-stability) and frequent (higher-strength) collaborations between inventors of these important patents and their coauthors. The results in this study suggest that it is not the number of coauthors of departed inventors that is likely to play a bigger role in termination decisions but, rather, the knowledge of these coauthors and their patterns of ties with the departed inventors. These findings, therefore, uncover a set of tools that firms can deploy to secure the tacit know-how of their most valuable inventors so that in a scenario in which these inventors leave, their patents can still contribute to the firms’ subsequent R&D activities, as opposed to being terminated.

This study further contributes to the emerging star-employee research. Recent years have seen a surge in studies aiming to understand the implications of star mobility for individuals and firms (Azoulay et al., 2019; Groysberg & Lee, 2009; Khanna, 2020; Oettl, 2012; Oldroyd & Morris, 2012; Tzabbar & Kehoe, 2014). I extend and complement the existing work in this area by taking a distinct approach and focusing on all inventors within firms. Both star and nonstar inventors are important participants in knowledge production. A systematic and inclusive investigation of their roles in knowledge creation is necessary to generate a deeper understanding of their functional impacts. This study joins recent attempts to consider both stars and nonstars in explaining individual and organizational outcomes (Chen & Garg, 2018; Liu, Mihm, & Sosa, 2018). The results highlight a significant effect of inventor turnover on patent termination after controlling for stars’ presence on patents.

The findings in this article also inform the broader literature investigating the link between employee mobility and entrepreneurial behavior (Chatterji, 2009; Elfenbein, Hamilton, & Zenger, 2010; Gambardella et al., 2015; Ganco, 2013). While we know that characteristics of inventors’ knowledge (such as its breadth and complexity) have an effect on the entrepreneurial tendency of these inventors, we lack an understanding of the role of their collaborative ties in such outcomes. For instance, Ganco (2013) showed that if mobile inventors hold complex knowledge, they are more likely to move in teams due to the tacit knowledge embedded in the ties with their collaborators. But does the likelihood of inventors’ creating an entrepreneurial venture depend on the nature of their ties? For instance, do certain features of collaborative ties support movement in teams more than others, controlling for the type of knowledge embedded in these ties? Further, this research shows that under certain conditions, firms are able to partly retain the tacit knowledge of departed inventors through their coauthors. In this case, will mobile inventors in their entrepreneurial careers have less of a competitive edge against their former employers? This article suggests that answering such questions by examining the collaborative ties of mobile inventors provides an exciting extension to the existing research in this area.

The study suffers from several limitations that also offer opportunities for future research. A key point to note is that termination decisions studied in the article entail considerable strategic risks for firms, as terminating valuable patents and continuing investing in unworthy patents can severely challenge firms’ competitive advantage in an industry. I show that inventor turnover can be one of the inputs in termination decisions. Yet, an important question that remains unanswered is whether firms make Type I or Type II errors in informing their termination decisions based on inventor turnover. Prior research has noted that while firms generate hundreds of patents each year, the number of patents that lead to the development of new drugs is extremely small (Thomke & Kuemmerle, 2002). Also, the cost of bringing a new drug to market is estimated to be more than $1 billion (DiMasi & Grabowski, 2007). Considering the uncertainty and risk inherent in the innovation process, it becomes imperative to understand the extent of errors in termination decisions as these errors can be costly to firms in terms of draining financial resources and challenging their competitive positions. Thus, an important way to extend the current work is to examine the performance implications of termination decisions. A greater understanding of the implications of firm heterogeneity in termination decisions and, eventually, performance will be instrumental to scholars in this field and managers within firms.

Another interesting research inquiry could focus on the reasons for inventor turnover. I acknowledged the possibility that some of the turnovers I study may not be voluntary in nature, that is, inventors could be laid off or even fired due to poor performance. In this case, firms may have decided to terminate their patents ex ante. I used an alternative approach, that is, CEM, Coarsened Exact Matching (CEM) in supplementary analyses (Online Appendix B [S1]) to verify the concerns for the endogeneity and found results to be robust to this approach. However, future work focusing only on voluntary turnover can confer more confidence in the findings here.

In addition, future work can examine the nature of turnovers and focus on how turnovers differ characteristically and what are the implications of this heterogeneity for termination decisions. I explored some of these possibilities in the supplementary analyses. For example, I found that of all turnover events, around 2.7% involved turnover of more than one inventor on a patent, and treatment of turnover as a continuous variable yielded similar results. In addition, the turnover of an inventor who was the first author on a patent had a stronger impact on the termination likelihood of the patent (Online Appendix B [S7]). However, a more nuanced examination of how these turnover events may differ along various technological and social dimensions and their influence on termination decisions and perhaps firm performance may yield additional useful findings.

Finally, the process of innovation and its basic details may vary across industries. For instance, the complexities inherent in drug discovery requires scientists to collaborate extensively within firms and also across firm boundaries (Hunter, 2014). Additionally, scientists engaged in R&D develop their unique set of skills and expertise from years of commitment to a scientific inquiry. The persistence in R&D activities is commonly referred to as “path dependence,” typically explained by the tendency of scientists to build on the previous knowledge in an incremental manner (Helfat, 1994; Hess & Rothaermel, 2011). By contrast, even though industries such as semiconductors and information technology are also high-tech, they are characterized by rapid technological change and short technological life cycles (Hall & Ziedonis, 2001). Inventors in these industries must keep pace with new technological innovations, and therefore, it is likely that the focus of inventors is more on learning new skills in a fast-paced environment rather than developing skills and expertise collaboratively in an incremental manner. The raw calculations based on the patent data from the National Bureau of Economic Research provide some indication of this notion, as the number of inventors per patent in the pharmaceutical industry was 35% more (= 3.5) than in the semiconductor industry (= 2.6). Also, due to the fast-paced nature of some of these high-tech industries, inventors may be better off replacing their ties more frequently (lower stability), as opposed to maintaining them, to access fresh know-how. This direction of future research is in line with the call to identify contingencies to explain when tie stability may (or not) be necessary (Levin & Walter, 2018). Thus, the fundamental differences between the pharmaceutical industry studied here and other high-tech industries could unlock some interesting insights into how firms in different sectors manage their knowledge portfolios in the face of employee turnover. Overall, this study offers several exciting opportunities for future research.

Supplemental Material

sj-docx-1-jom-10.1177_0149206321997910 – Supplemental material for Peeking Inside the Black Box: Inventor Turnover and Patent Termination

Supplemental material, sj-docx-1-jom-10.1177_0149206321997910 for Peeking Inside the Black Box: Inventor Turnover and Patent Termination by Rajat Khanna in Journal of Management

Footnotes

Acknowledgements

I am grateful to professor Jorge Walter and two anonymous reviewers for their insightful suggestions and guidance. I also thank Isin Guler, Jeff Edwards, and Kiran Awate for their helpful comments while I was working on this paper.

ORCID iD

Rajat Khanna

Supplemental material for this article is available with the manuscript on the JOM website.

Notes

References

Adler

P. S.

Kwon

S.-W.

2002. Social capital: Prospects for a new concept. Academy of Management Review, 27: 17-40.

Agarwal

Ganco

Ziedonis

R. H.

2009. Reputations for toughness in patent enforcement: Implications for knowledge spillovers via inventor mobility. Strategic Management Journal, 30: 1349-1374.

Ahuja

Soda

Zaheer

2012. Introduction to the special issue: The genesis and dynamics of organizational networks. Organization Science, 23: 434-448.

*Alcacer

Gittelman

2006. Patent citations as a measure of knowledge flows: The influence of examiner citations. Review of Economics and Statistics, 88: 774-779.

Allen

D. G.

2006. Do organizational socialization tactics influence newcomer embeddedness and turnover? Journal of Management, 32: 237-256.

Allen

T. J.

Cohen

S. I.

1969. Information flow in research and development laboratories. Administrative Science Quarterly, 14: 12-19.

*Allison

P. D.

Waterman

R. P.

2002. Fixed-effects negative binomial regression models. In Stolzenberg

R. M.

(Ed.), Sociological methodology 2002, Vol. 32: 247-265. Oxford: Basil Blackwell.

Almeida

Kogut

1999. Localization of knowledge and the mobility of engineers in regional networks. Management Science, 45: 905-917.

Anderson

J. C.

Gerbing

D. W.

1988. Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103: 411.

10.

Anderson

J. R.

1983. A spreading activation theory of memory. Journal of Verbal Learning and Verbal Behavior, 22: 261-295.

11.

Angrist

J. D.

Keueger

A. B.

1991. Does compulsory school attendance affect schooling and earnings? Quarterly Journal of Economics, 106: 979-1014.

12.

Argote

Beckman

S. L.

Epple

1990. The persistence and transfer of learning in industrial settings. Management Science, 36: 140-154.

13.

Argote

Ingram

2000. Knowledge transfer: A basis for competitive advantage in firms. Organizational Behavior and Human Decision Processes, 82: 150-169.

14.

Argyres

1996. Evidence on the role of firm capabilities in vertical integration decisions. Strategic Management Journal, 17: 129-150.

15.

Arora

1996. Contracting for tacit knowledge: The provision of technical services in technology licensing contracts. Journal of Development Economics, 50: 233-256.

16.

Arrow

K. J.

1972. Economic welfare and the allocation of resources for invention. In EDITOR NAME(S) (Ed.), Readings in industrial economics: 219-236. Berlin: Springer.

17.

Arrow, Kenneth Joseph. “Economic welfare and the allocation of resources for invention.” In Readings in industrial economics, pp. 219-236. Palgrave, London, 1972.

18.

Arthur

W. B.

2007. The structure of invention. Research Policy, 36: 274-287.

19.

Arundel

Kabla

1998. What percentage of innovations are patented? Empirical estimates for European firms. Research Policy, 27: 127-141.

20.

Azoulay

Fons-Rosen

Zivin

J. S. G.

2019. Does science advance one funeral at a time? American Economic Review, 109: 2889-2920.

21.

Bagozzi

R. P.

Phillips

L. W.

1991. Assessing construct-validity in organizational research. Administrative Science Quarterly, 36: 421-458.

22.

Bagozzi

R. P.

Phillips

L. W.

1982. Representing and testing organizational theories: A holistic construal. Administrative Science Quarterly, 27: 459-489.

23.

Barney

1991. Firm resources and sustained competitive advantage. Journal of Management, 17: 99–120.

24.

Barney

J. B.

Arikan

A. M.

2001. The resource-based view: Origins and implications. In Hitt

M. A.

Freeman

R. E.

Harrison

J. S.

(Eds.), The Blackwell handbook of strategic management: 124-188. Oxford: Blackwell.

25.

Bartel

C. A.

Garud

2009. The role of narratives in sustaining organizational innovation. Organization Science, 20: 107-117.

26.

Bascle

2008. Controlling for endogeneity with instrumental variables in strategic management research. Strategic Organization, 6: 285-327.

27.

Baum

J. A. C.

McEvily

Rowley

T. J.

2012. Better with age? Tie longevity and the performance implications of bridging and closure. Organization Science, 23: 529-546.

28.

Becker

G. S.

1975. Human capital: A theoretical and empirical analysis, with special reference to education (2nd ed.). New York: National Bureau of Economic Research/Columbia University Press.

29.

Bizzi

2013. The dark side of structural holes: A multilevel investigation. Journal of Management, 39: 1554-1578.

30.

*Blackwell

Iacus

King

Porro

2009. CEM: Coarsened exact matching in Stata. Stata Journal, 9: 524-546.

31.

Boisot

M. H.

1998. Knowledge assets: Securing competitive advantage in the information economy. Oxford: Oxford University Press.

32.

Bouty

2000. Interpersonal and interaction influences on informal resource exchanges between R&D researchers across organizational boundaries. Academy of Management Journal, 43: 50-65.

33.

Bowonder

Miyake

2000. Technology management: A knowledge ecology perspective. International Journal of Technology Management, 19: 662-684.

34.

Brandes

Fleischer

2005. Centrality measures based on current flow. In Annual symposium on theoretical aspects of computer science: 533-544. Berlin: Springer.

35.

Brass

D. J.

1992. Power in organizations: A social network perspective. In Moore

Whitt

J. A.

(Eds.), Research in politics and society, Vol. 4: 295-323. Greenwich, CT: JAI Press.

36.

*Brown

J. S.

Duguid

1991. Organizational learning and communities-of-practice: Toward a unified view of working, learning, and innovation. Organization Science, 2: 40-57.

37.

Brown

J. S.

Duguid

1998. Organizing knowledge. California Management Review, 40: 90-111.

38.

Brown

J. S.

Duguid

2001. Knowledge and organization: A social-practice perspective. Organization Science, 12: 198-213.

39.

Bureau of Labor Statistics, U.S. Department of Labor, 2016, Economic News Release, Annual total separations rates by industry and region on the Internet at https://www.bls.gov/news.release/jolts.t16.htm (visited October 12, 2020).

40.

Burt

R. S.

1992. Structural holes. Cambridge, MA: Harvard University Press.

41.

Burt

R. S.

2001. Closure as social capital. In Lin

Cook

Burt

R. S.

(Eds.), Social capital: Theory and research: 31-55. New York: Aldine de Gruyter.

42.

Burt

R. S.

Merluzzi

2016. Network oscillation. Academy of Management Discoveries, 2: 368-391.

43.

Campbell

B. A.

Ganco

Franco

A. M.

Agarwal

2012. Who leaves, where to, and why worry? Employee mobility, entrepreneurship and effects on source firm performance. Strategic Management Journal, 33: 65-87.

44.

Carnabuci

Operti

2013. Where do firms’ recombinant capabilities come from? Intraorganizational networks, knowledge, and firms’ ability to innovate through technological recombination. Strategic Management Journal, 34: 1591-1613.

45.

Chatterji

A. K.

2009. Spawned with a silver spoon? Entrepreneurial performance and innovation in the medical device industry. Strategic Management Journal, 30: 185-206.

46.

Chen

A. N.

Edgington

T. M.

2005. Assessing value in organizational knowledge creation: Considerations for knowledge workers. MIS Quarterly, 29: 279-309.

47.

Chen

J. S.

Garg

2018. Dancing with the stars: Benefits of a star employee’s temporary absence for organizational performance. Strategic Management Journal, 39(5): 1239-1267.

48.

Coff

R. W.

1997. Human assets and management dilemmas: Coping with hazards on the road to resource-based theory. Academy of Management Review, 22: 374-402.

49.

Coff

R. W.

2002. Human capital, shared expertise, and the likelihood of impasse in corporate acquisitions. Journal of Management, 28: 107-128.

50.

Cohen

W. M.

Levinthal

D. A.

1990. Absorptive-capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35: 128-152.

51.

Coleman

J. S.

1988. Social capital in the creation of human-capital. American Journal of Sociology, 94: S95-S120.

52.

Coleman

J. S.

1990. Foundations of social theory. Cambridge, MA: Harvard University Press.

53.

*Cook

S. D. N.

Brown

J. S.

1999. Bridging epistemologies: The generative dance between organizational knowledge and organizational knowing. Organization Science, 10: 381-400.

54.

Corredoira

R. A.

Rosenkopf

2010. Should auld acquaintance be forgot? The reverse transfer of knowledge through mobility ties. Strategic Management Journal, 31: 159-181.

55.

Cross

Sproull

2004. More than an answer: Information relationships for actionable knowledge. Organization Science, 15: 446-462.

56.

Crossan

M. M.

Lane

H. W.

White

R. E.

1999. An organizational learning framework: From intuition to institution. Academy of Management Review, 24: 522-537.

57.

DiMasi

J. A.

Grabowski

H. G.

2007. The cost of biopharmaceutical R&D: Is biotech different? Managerial and Decision Economics, 28: 469-479.

58.

Doz

Y. L.

1996. The evolution of cooperation in strategic alliances: Initial conditions or learning processes? Strategic Management Journal, 17: 55-83.

59.

Dyer

J. H.

Singh

1998. The relational view: Cooperative strategy and sources of interorganizational competitive advantage. Academy of Management Review, 23: 660-679.

60.

Ebadi

Y. M.

Utterback

J. M.

1984. The effects of communication on technological innovation. Management Science, 30: 572-585.

61.

Edmondson

A. C.

Bohmer

R. M.

Pisano

2000. Disrupted routines: Effects of team learning on new technology adaptation. Boston: Harvard Business School.

62.

Edwards

P. N.

Mayernik

M. S.

Batcheller

A. L.

Bowker

G. C.

Borgman

C. L.

2011. Science friction: Data, metadata, and collaboration. Social Studies of Science, 41: 667-690.

63.

Elfenbein

D. W.

Hamilton

B. H.

Zenger

T. R.

2010. The small firm effect and the entrepreneurial spawning of scientists and engineers. Management Science, 56: 659-681.

64.

Elfenbein

D. W.

Knott

A. M.

2015. Time to exit: Rational, behavioral, and organizational delays. Strategic Management Journal, 36: 957-975.

65.

Epple

Argote

Devadas

1991. Organizational learning curves: A method for investigating intra-plant transfer of knowledge acquired through learning by doing. Organization Science, 2: 58-70.

66.

Fey

C. F.

Birkinshaw

2005. External sources of knowledge, governance mode, and R&D performance. Journal of Management, 31: 597-621.

67.

Fleming

2001. Recombinant uncertainty in technological search. Management Science, 47: 117-132.

68.

Fleming

Mingo

Chen

2007. Collaborative brokerage, generative creativity, and creative success. Administrative Science Quarterly, 52: 443-475.

69.

Fleming

Sorenson

2001. Technology as a complex adaptive system: Evidence from patent data. Research Policy, 30: 1019-1039.

70.

Fleming

Sorenson

2004. Science as a map in technological search. Strategic Management Journal, 25: 909-928.

71.

Fortune

Mitchell

2012. Unpacking firm exit at the firm and industry levels: The adaptation and selection of firm capabilities. Strategic Management Journal, 33: 794-819.

72.

Freeman

L. C.

1977. A set of measures of centrality based on betweenness. Sociometry, 40: 35-41.

73.

Funk

R. J.

2015. The dark side of brokerage: Conflicts between individual and collective pursuits of innovation. Academy of Management Annual Meeting Proceedings, 2015: 11819.

74.

Galunic

D. C.

Rodan

1998. Resource recombinations in the firm: Knowledge structures and the potential for Schumpeterian innovation. Strategic Management Journal, 19: 1193-1201.

75.

Gambardella

Ganco

Honore

2015. Using what you know: Patented knowledge in incumbent firms and employee entrepreneurship. Organization Science, 26: 456-474.

76.

Ganco

2013. Cutting the Gordian knot: The effect of knowledge complexity on employee mobility and entrepreneurship. Strategic Management Journal, 34: 666-686.

77.

Gardner

T. M.

2002. In the trenches at the talent wars: Competitive interaction for scarce human resources. Human Resource Management, 41: 225-237.

78.

Gargiulo

Benassi

1999. The dark side of social capital. In Leenders

R. T. A. J.

Gabbay

S. M.

(Eds.), Corporate social capital and liability: 298-322. Boston: Springer.

79.

Gargiulo

Ertug

Galunic

2009. The two faces of control: Network closure and individual performance among knowledge workers. Administrative Science Quarterly, 54: 299-333.

80.

Gavetti

Levinthal

2000. Looking forward and looking backward: Cognitive and experiential search. Administrative Science Quarterly, 45: 113-137.

81.

C. M.

Huang

K. W.

Png

I. P. L.

2016. Engineer/scientist careers: Patents, online profiles, and misclassification bias. Strategic Management Journal, 37: 232-253.

82.

*Gilbert

Shapiro

1990. Optimal patent length and breadth. Rand Journal of Economics, 21: 106-112.

83.

Gittelman

Kogut

2003. Does good science lead to valuable knowledge? Biotechnology firms and the evolutionary logic of citation patterns. Management Science, 49: 366-382.

84.

*Grabowski

2002. Patents, innovation and access to new pharmaceuticals. Journal of International Economic Law, 5: 849-860.

85.

Granovetter

1973. The strength of weak ties. American Journal of Sociology, 78: 1360-1380.

86.

Granovetter

1982. The strength of weak ties: A network theory revisited. In Marsden

P. V.

Lin

(Eds.), Social structure and network analysis: 105-130. Beverly Hills, CA: Sage.

87.

Grant

R. M.

1996. Toward a knowledge-based theory of the firm. Strategic Management Journal, 17: 109-122.

88.

*Grigoriou

Rothaermel

F. T.

2014. Structural microfoundations of innovation: The role of relational stars. Journal of Management, 40: 586-615.

89.

Groysberg

Lee

L. E.

2009. Hiring stars and their colleagues: Exploration and exploitation in professional service firms. Organization Science, 20: 740-758.

90.

Gulati

1995. Does familiarity breed trust: The implications of repeated ties for contractual choice in alliances. Academy of Management Journal, 38: 85-112.

91.

Guler

2007. Throwing good money after bad? Political and institutional influences on sequential decision making in the venture capital industry. Administrative Science Quarterly, 52: 248-285.

92.

Guler

2018. Pulling the plug: The capability to terminate unsuccessful projects and firm performance. Strategy Science, 3: 481-497.

93.

Guler

Nerkar

2012. The impact of global and local cohesion on innovation in the pharmaceutical industry. Strategic Management Journal, 33: 535-549.

94.

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.

95.

Hansen

L. P.

1982. Large sample properties of generalized method of moments estimators. Econometrica, 50: 1029-1054.

96.

Hansen

M. T.

1999. The search-transfer problem: The role of weak ties in sharing knowledge across organization subunits. Administrative Science Quarterly, 44: 82-111.

97.

Hansen

Morten T.

Podolny

Joel M.

Pfeffer

Jeffrey

. “So many ties, so little time: A task contingency perspective on the value of social capital in organization.” Administrative Science Quarterly 42 (1999): 1-50.

98.

Hansen

M. T.

Podolny

J. M.

Pfeffer

1999. So many ties, so little time: A task contingency perspective on the value of social capital in organizations. Citeseer.

99.

Hargadon

Sutton

R. I.

1997. Technology brokering and innovation in a product development firm. Administrative Science Quarterly, 42: 716-749.

100.

*Harhoff

Narin

Scherer

F. M.

Vopel

1999. Citation frequency and the value of patented inventions. Review of Economics and Statistics, 81: 511-515.

101.

Hatch

N. W.

Dyer

J. H.

2004. Human capital and learning as a source of sustainable competitive advantage. Strategic Management Journal, 25: 1155-1178.

102.

Hausman

J. A.

Abrevaya

Scott-Morton

F. M.

1998. Misclassification of the dependent variable in a discrete-response setting. Journal of Econometrics, 87: 239-269.

103.

Hayashi

2000. Econometrics. Princeton, NJ: Princeton University Press.

104.

Heckman

J. J.

1979. Sample selection bias as a specification error. Econometrica, 47: 153-161.

105.

Helfat

C. E.

1994. Evolutionary trajectories in petroleum firm R&D. Management Science, 40: 1720-1747.

106.

*Helfat

C. E.

Peteraf

M. A.

2003. The dynamic resource-based view: Capability lifecycles. Strategic Management Journal, 24: 997-1010.

107.

Henderson

R. M.

Clark

K. B.

1990. Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative Science Quarterly, 35: 9-30.

108.

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.

109.

Hoetker

Agarwal

2007. Death hurts, but it isn’t fatal: The postexit diffusion of knowledge created by innovative companies. Academy of Management Journal, 50: 446-467.

110.

Hoisl

2007. Tracing mobile inventors: The causality between inventor mobility and inventor productivity. In A study of inventors: 65-119. CITY: Springer.

111.

Hoisl

2007. Tracing mobile inventors: The causality between inventor mobility and inventor productivity. Research Policy, 36: 619-636.

112.

Hoskisson

R. E.

Hitt

M. A.

Johnson

R. A.

Moesel

D. D.

1993. Construct-validity of an objective (entropy) categorical measure of diversification strategy. Strategic Management Journal, 14: 215-235.

113.

Hunter

2014. Collaboration for innovation is the new mantra for the pharmaceutical industry. Drug Discovery, 9: 9-15.

114.

*Iacus

S. M.

King

Porro

2011. Multivariate matching methods that are monotonic imbalance bounding. Journal of the American Statistical Association, 106: 345-361.

115.

Jaffe

A. B.

1986. Technological opportunity and spillovers of R&D: Evidence from firms’ patents, profits and market value. Cambridge, MA: National Bureau of Economic Research.

116.

*Jaffe

A. B.

Lerner

2011. Innovation and its discontents: How our broken patent system is endangering innovation and progress, and what to do about it. Princeton, NJ: Princeton University Press.

117.

*Jaffe

A. B.

Trajtenberg

1999. International knowledge flows: Evidence from patent citations. Economics of Innovation and New Technology, 8: 105-136.

118.

*Jaffe

A. B.

Trajtenberg

Fogarty

M. S.

2000. Knowledge spillovers and patent citations: Evidence from a survey of inventors. American Economic Review, 90: 215-218.

119.

Jaffe

A. B.

Trajtenberg

Henderson

1993. Geographic localization of knowledge spillovers as evidenced by patent citations. Quarterly Journal of Economics, 108: 577-598.

120.

Joseph

Klingebiel

Wilson

A. J.

2016. Organizational structure and performance feedback: Centralization, aspirations, and termination decisions. Organization Science, 27: 1065-1083.

121.

Kaplan

Vakili

2015. The double-edged sword of recombination in breakthrough innovation. Strategic Management Journal, 36: 1435-1457.

122.

Kapoor

Lim

2007. The impact of acquisitions on the productivity of inventors at semiconductor firms: A synthesis of knowledge-based and incentive-based perspectives. Academy of Management Journal, 50: 1133-1155.

123.

Katila

Ahuja

2002. Something old, something new: A longitudinal study of search behavior and new product introduction. Academy of Management Journal, 45: 1183-1194.

124.

Katila

Mang

P. Y.

2003. Exploiting technological opportunities: The timing of collaborations. Research Policy, 32: 317-332.

125.

Khanna

2020. Aftermath of a tragedy: A star’s death and coauthors’ subsequent productivity. Research Policy, 50: 104-159.

126.

Khanna

Guler

Nerkar

2016. Fail often, fail big, and fail fast? Learning from small failures and R&D performance in the pharmaceutical industry. Academy of Management Journal, 59: 436-459.

127.

Khanna

Guler

Nerkar

2018. Entangled decisions: Knowledge interdependencies and terminations of patented inventions in the pharmaceutical industry. Strategic Management Journal, 39: 2439-2465.

128.

Kikoski

C. K.

Kikoski

J. F.

2004. The inquiring organization: Tacit knowledge, conversation, and knowledge creation skills for 21st-century organizations. London: Greenwood.

129.

*Klemperer

1990. How broad should the scope of patent protection be. RAND Journal of Economics, 21: 113-130.

130.

Kobrin

S. J.

1991. An empirical-analysis of the determinants of global integration. Strategic Management Journal, 12: 17-31.

131.

Kogut

Zander

1992. Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science, 3: 383-397.

132.

Kogut

Zander

1996. What firms do? Coordination, identity, and learning. Organization Science, 7: 502-518.

133.

Krackhardt

1999. The ties that torture: Simmelian tie analysis in organizations. Research in the Sociology of Organizations, 16: 183-210.

134.

Krackhardt

Nohria

Eccles

2003. The strength of strong ties. Networks in the Knowledge Economy, 82.

135.

Krackhardt

1992. The strength of strong ties: The importance of philos in organizations. In Nohria

Eccles

R. G.

(Eds.), Networks and organizations: Structure, form and action: 216-239. Cambridge, MA: Harvard Business School Press.

136.

Kuhn

T. S.

1962. The structure of scientiﬁc revolutions. Chicago: University of Chicago Press.

137.

Kumar

Zaheer

2019. Ego-network stability and innovation in alliances. Academy of Management Journal, 62: 691-716.

138.

*Lanjouw

J. O.

1998. Patent protection in the shadow of infringement: Simulation estimations of patent value. Review of Economic Studies, 65: 671-710.

139.

Lanjouw

J. O.

Schankerman

2004. Patent quality and research productivity: Measuring innovation with multiple indicators. Economic Journal, 114: 441-465.

140.

Leonard

Sensiper

1998. The role of tacit knowledge in group innovation. California Management Review, 40: 112-132.

141.

Levin

D. Z.

Cross

2004. The strength of weak ties you can trust: The mediating role of trust in effective knowledge transfer. Management Science, 50: 1477-1490.

142.

Levin

D. Z.

Walter

2018. Is tie maintenance necessary? Academy of Management Discoveries, 4: 497-500.

143.

Levinthal

March

J. G.

1981. A model of adaptive organizational search. Journal of Economic Behavior & Organization, 2: 307-333.

144.

*Levitas

Chi

2002. Rethinking Rouse and Daellenbach’s rethinking: Isolating vs. testing for sources of sustainable competitive advantage. Strategic Management Journal, 23: 957-962.

145.

*Levitas

Chi

2010. A look at the value creation effects of patenting and capital investment through a real options lens: The moderating role of uncertainty. Strategic Entrepreneurship Journal, 4: 212-233.

146.

Chi

T. L.

2013. Venture capitalists’ decision to withdraw: The role of portfolio configuration from a real options lens. Strategic Management Journal, 34: 1351-1366.

147.

Liebeskind

J. P.

Oliver

A. L.

Zucker

Brewer

1996. Social networks, learning, and flexibility: Sourcing scientific knowledge in new biotechnology firms. Organization Science, 7: 428-443.

148.

Liu

2014. Human capital, social collaboration, and patent renewal within US pharmaceutical firms. Journal of Management, 40: 616-636.

149.

*Liu

Arthurs

Cullen

Alexander

2008. Internal sequential innovations: How does interrelatedness affect patent renewal? Research Policy, 37: 946-953.

150.

Liu

Mihm

Sosa

M.E.

2018. Where do stars come from? The role of star vs. nonstar collaborators in creative settings. Organization Science, 29(6), pp.1149-1169.

151.

Maliranta

Mohnen

Rouvinen

2009. Is inter-firm labor mobility a channel of knowledge spillovers? Evidence from a linked employer–employee panel. Industrial and Corporate Change, 18: 1161-1191.

152.

March

J. G.

1991. Exploration and exploitation in organizational learning. Organization Science, 2: 71-87.

153.

*March

J. G.

Simon

H. A.

1958. Organizations. New York: Wiley.

154.

Marsden

P. V.

Campbell

K. E.

1984. Measuring tie strength. Social Forces, 63: 482-501.

155.

Martin

Mitchell

1998. The influence of local search and performance heuristics on new design introduction in a new product market. Research Policy, 26: 753-771.

156.

Mawdsley

J. K.

Somaya

2016. Employee mobility and organizational outcomes: An integrative conceptual framework and research agenda. Journal of Management, 42: 85-113.

157.

McEvily

Perrone

Zaheer

2003. Introduction to the special issue on trust in an organizational context. Organization Science, 14: 1-4.

158.

McFadyen

M. A.

Cannella

A. A.

2004. Social capital and knowledge creation: Diminishing returns of the number and strength of exchange relationships. Academy of Management Journal, 47: 735-746.

159.

McFadyen

M. A.

Semadeni

Cannella

A. A.

Jr.

2009. Value of strong ties to disconnected others: Examining knowledge creation in biomedicine. Organization Science, 20: 552-564.

160.

Møen

2005. Is mobility of technical personnel a source of R&D spillovers? Journal of Labor Economics, 23: 81-114.

161.

Moore

K. A.

2005. Worthless patents. Berkeley Technology Law Journal, 20: 1521.

162.

Moran

2005. Structural vs. relational embeddedness: Social capital and managerial performance. Strategic Management Journal, 26: 1129-1151.

163.

Mowery

D. C.

Oxley

J. E.

Silverman

B. S.

1996. Strategic alliances and interfirm knowledge transfer. Strategic Management Journal, 17: 77-91.

164.

Nahapiet

Ghoshal

1998. Social capital, intellectual capital, and the organizational advantage. Academy of Management Review, 23: 242-266.

165.

Nelson

R. R.

Winter

S. G.

1982. In An evolutionary theory of economic change: Cambridge, MA: Harvard. University Press.

166.

Nerkar

2003. Old is gold? The value of temporal exploration in the creation of new knowledge. Management Science, 49: 211-229.

167.

Nerkar

Paruchuri

2005. Evolution of R&D capabilities: The role of knowledge networks within a firm. Management Science, 51: 771-785.

168.

Nickerson

J. A.

Zenger

T. R.

2004. A knowledge-based theory of the firm: The problem-solving perspective. Organization Science, 15: 617-632.

169.

Nonaka

1994. A dynamic theory of organizational knowledge creation. Organization Science, 5: 14-37.

170.

Nonaka

Takeuchi

1995. The knowledge-creating company: How Japanese companies create the dynamics of innovation. New York: Oxford University Press.

171.

Nonaka

Toyama

Konno

2000. SECI, ba and leadership: A unified model of dynamic knowledge creation. Long Range Planning, 33: 5-34.

172.

Obstfeld

2005. Social networks, the tertius lungens orientation, and involvement in innovation. Administrative Science Quarterly, 50: 100-130.

173.

Oettl

2012. Reconceptualizing stars: Scientist helpfulness and peer performance. Management Science, 58: 1122-1140.

174.

Oldroyd

J. B.

Morris

S. S.

2012. Catching falling stars: A human resource response to social capital’s detrimental effect of information overload on star employees. Academy of Management Review, 37: 396-418.

175.

*Pakes

1986. Patents as options: Some estimates of the value of holding European patent stocks. Econometrica, 54: 755-784.

176.

Paruchuri

2010. Intraorganizational networks, interorganizational networks, and the impact of central inventors: A longitudinal study of pharmaceutical firms. Organization Science, 21: 63-80.

177.

Paruchuri

Nerkar

Hambrick

D. C.

2006. Acquisition integration and productivity losses in the technical core: Disruption of inventors in acquired companies. Organization Science, 17: 545-562.

178.

Phillips

D. J.

2002. A genealogical approach to organizational life or chances: The parent-progeny transfer among Silicon Valley law firms, 1946-1996. Administrative Science Quarterly, 47: 474-506.

179.

Pisano

G. P.

1997. R&D performance, collaborative arrangements and the market for know-how: A test of the “lemons” hypothesis in biotechnology. Working paper no. 97-105, Harvard Business School, Cambridge, MA.

180.

Polanyi

Grene

1969. Knowing and being: Essays. Chicago: University of Chicago Press.

181.

Priem

R. L.

Butler

J. E.

2001. Is the resource-based “view” a useful perspective for strategic management research? Academy of Management Review, 26: 22-40.

182.

Ravetz

J. R.

1973. Scientific knowledge and its social problems. CITY: Transaction.

183.

Ravetz

J. R.

1971. Scientific knowledge and its social problems. London: Oxford University Press.

184.

Reagans

McEvily

2003. Network structure and knowledge transfer: The effects of cohesion and range. Administrative Science Quarterly, 48: 240-267.

185.

*Reagans

Singh

P. V.

Krishnan

2015. Forgotten third parties: Analyzing the contingent association between unshared third parties, knowledge overlap, and knowledge transfer relationships with outsiders. Organization Science, 26: 1400-1414.

186.

Reagans

Zuckerman

McEvily

2004. How to make the team: Social networks vs. demography as criteria for designing effective teams. Administrative Science Quarterly, 49: 101-133.

187.

Reed

Defillippi

R. J.

1990. Causal ambiguity, barriers to imitation, and sustainable competitive advantage. Academy of Management Review, 15: 88-102.

188.

Reed

W. R.

2015. On the practice of lagging variables to avoid simultaneity. Oxford Bulletin of Economics and Statistics, 77(6), pp.897-905.

189.

*Reitzig

2004. Improving patent valuations for management purposes: Validating new indicators by analyzing application rationales. Research Policy, 33: 939-957.

190.

*Rogan

Sorenson

2014. Picking a (poor) partner: A relational perspective on acquisitions. Administrative Science Quarterly, 59: 301-329.

191.

Rosenkopf

Almeida

2003. Overcoming local search through alliances and mobility. Management Science, 49: 751-766.

192.

Rosenkopf

Nerkar

2001. Beyond local search: Boundary-spanning, exploration, and impact in the optical disk industry. Strategic Management Journal, 22: 287-306.

193.

*Rothaermel

F. T.

2001. Incumbent’s advantage through exploiting complementary assets via interfirm cooperation. Strategic Management Journal, 22: 687-699.

194.

Rothaermel

F. T.

Hess

A. M.

2007. Building dynamic capabilities: Innovation driven by individual-, firm-, and network-level effects. Organization Science, 18: 898-921.

195.

Ryle

1949. The concept of mind. London: Huchinson.

196.

Sampson

R. C.

2007. R&D alliances and firm performance: The impact of technological diversity and alliance organization on innovation. (2): 364-386.

197.

Sargan

J. D.

1958. The estimation of economic relationships using instrumental variables. Econometrica, 26: 393-415.

198.

Sasovova

Mehra

Borgatti

S. P.

Schippers

M. C.

2010. Network churn: The effects of self-monitoring personality on brokerage dynamics. Administrative Science Quarterly, 55: 639-668.

199.

*Schankerman

1998. How valuable is patent protection? Estimates by technology field. RAND Journal of Economics, 29: 77-107.

200.

Schilling

M. A.

Phelps

C. C.

2007. Interfirm collaboration networks: The impact of large-scale network structure on firm innovation. Management Science, 53: 1113-1126.

201.

*Scotchmer

1999. On the optimality of the patent renewal system. RAND Journal of Economics, 30: 181-196.

202.

Sears

Hoetker

2014. Technological overlap, technological capabilities, and resource recombination in technological acquisitions. Strategic Management Journal, 35: 48-67.

203.

*Semadeni

Withers

M. C.

Trevis Certo

2014. The perils of endogeneity and instrumental variables in strategy research: Understanding through simulations. Strategic Management Journal, 35: 1070-1079.

204.

Serrano

C. J.

2010. The dynamics of the transfer and renewal of patents. RAND Journal of Economics, 41: 686-708.

205.

*Shapiro

2000. Navigating the patent thicket: Cross licenses, patent pools, and standard setting. Innovation Policy and the Economy, 1: 119-150.

206.

Shen

Cannella

A. A.

2002. Power dynamics within top management and their impacts on CEO dismissal followed by inside succession. Academy of Management Journal, 45: 1195-1206.

207.

Simon

H. A.

1991. Bounded rationality and organizational learning. Organization Science, 2: 125-134.

208.

*Singh

Agrawal

2011. Recruiting for ideas: How firms exploit the prior inventions of new hires. Management Science, 57: 129-150.

209.

Sirmon

D. G.

Hitt

M. A.

Ireland

R. D.

2007. Managing firm resources in dynamic environments to create value: Looking inside the black box. Academy of Management Review, 32: 273-292.

210.

Sirmon

D. G.

Hitt

M. A.

Ireland

R. D.

Gilbert

B. A.

2011. Resource orchestration to create competitive advantage: Breadth, depth, and life cycle effects. Journal of Management, 37: 1390-1412.

211.

Sleesman

D. J.

Lennard

A. C.

McNamara

Conlon

D. E.

2018. Putting escalation of commitment in context: A multilevel review and analysis. Academy of Management Annals, 12: 178-207.

212.

*Somaya

2012. Patent strategy and management: An integrative review and research agenda. Journal of Management, 38: 1084-1114.

213.

Song

Almeida

2003. Learning-by-hiring: When is mobility more likely to facilitate interfirm knowledge transfer? Management Science, 49: 351-365.

214.

Sørensen

J. B.

1999. Executive migration and interorganizational competition. Social Science Research, 28: 289-315.

215.

Spender

J. C.

1996. Making knowledge the basis of a dynamic theory of the firm. Strategic Management Journal, 17: 45-62.

216.

Star

S. L.

Griesemer

J. R.

1989. Institutional ecology, “translations” and boundary objects: Amateurs and professionals in Berkeley’s Museum of Vertebrate Zoology, 1907-39. Social Studies of Science, 19: 387-420.

217.

Stephenson

Zelen

1989. Rethinking centrality: Methods and examples. Social Networks, 11: 1-37.

218.

*Stock

J. H.

Wright

J. H.

Yogo

2002. A survey of weak instruments and weak identification in generalized method of moments. Journal of Business & Economic Statistics, 20: 518-529.

219.

Stuart

T. E.

Podolny

J. M.

1996. Local search and the evolution of technological capabilities. Strategic Management Journal, 17: 21-38.

220.

Szulanski

1996. Exploring internal stickiness: Impediments to the transfer of best practice within the firm. Strategic Management Journal, 17: 27-43.

221.

Thomke

Kuemmerle

2002. Asset accumulation, interdependence and technological change: Evidence from pharmaceutical drug discovery. Strategic Management Journal, 23: 619-635.

222.

Tong

X. S.

Frame

J. D.

1994. Measuring national technological performance with patent claims data. Research Policy, 23: 133-141.

223.

Tortoriello

Krackhardt

2010. Activating cross-boundary knowledge: The role of Simmelian ties in the generation of innovations. Academy of Management Journal, 53: 167-181.

224.

Tortoriello

Reagans

McEvily

2012. Bridging the knowledge gap: The influence of strong ties, network cohesion, and network range on the transfer of knowledge between organizational units. Organization Science, 23: 1024-1039.

225.

*Tsoukas

2005. Do we really understand tacit knowledge. Managing Knowledge: An Essential Reader, 107: 1-18.

226.

Tushman

M. L.

1978. Technical communication in R & D laboratories: Impact of project work characteristics. Academy of Management Journal, 21: 624-645.

227.

Tushman

M. L.

1979. Impacts of perceived environmental variability on patterns of work related communication. Academy of Management Journal, 22: 482-500.

228.

Tushman

M. L.

Katz

1980. External communication and project performance: An investigation into the role of gatekeepers. Management Science, 26: 1071-1085.

229.

Tushman

M. L.

Romanelli

1983. Uncertainty, social location and influence in decision making: A sociometric analysis. Management Science, 29: 12-23.

230.

Tushman

M. L.

Scanlan

T. J.

1981. Boundary spanning individuals: Their role in information-transfer and their antecedents. Academy of Management Journal, 24: 289-305.

231.

Tzabbar

2009. When does scientist recruitment affect technological repositioning? Academy of Management Journal, 52: 873-896.

232.

Tzabbar

Kehoe

R. R.

2014. Can opportunity emerge from disarray? An examination of exploration and exploitation following star scientist turnover. Journal of Management, 40: 449-482.

233.

Uzzi

1996. The sources and consequences of embeddedness for the economic performance of organizations: The network effect. American Sociological Review, 61: 674-698.

234.

Uzzi

1997. Social structure and competition in interfirm networks. Administrative Science Quarterly, 42: 417-418.

235.

Van de Ven

W. P.

Van Praag

B. M.

1981. The demand for deductibles in private health insurance: A probit model with sample selection. Journal of Econometrics, 17: 229-252.

236.

Watts

D. J.

1999. Networks, dynamics, and the small-world phenomenon. American Journal of sociology, 105(2), pp.493-527.

237.

Wezel

F. C.

Cattani

Pennings

J. M.

2006. Competitive implications of interfirm mobility. Organization Science, 17: 691-709.

238.

*Wooldridge

J. M.

1999. Distribution-free estimation of some nonlinear panel data models. Journal of Econometrics, 90: 77-97.

239.

Wooldridge

J. M.

2015. Control function methods in applied econometrics. Journal of Human Resources, 50: 420-445.

240.

Yatchew

Griliches

1985. Specification error in probit models. Review of Economics and Statistics, 67: 134-139.

241.

*Yayavaram

Ahuja

2008. Decomposability in knowledge structures and its impact on the usefulness of inventions and knowledge-base malleability. Administrative Science Quarterly, 53: 333-362.

242.

Zaheer

McEvily

Perrone

1998. Does trust matter? Exploring the effects of interorganizational and interpersonal trust on performance. Organization Science, 9: 141-159.

243.

Zucker

L. G.

Darby

M. R.

2001. Capturing technological opportunity via Japan’s star scientists: Evidence from Japanese firms’ biotech patents and products. Journal of Technology Transfer, 26: 37-58.

244.

Zucker

L. G.

Darby

M. R.

Brewer

M. B.

Peng

1995. Collaboration structure and information dilemmas in biotechnology: Organizational boundaries as trust production. Working paper no. 5199, National Bureau of Economic Research, Cambridge, MA.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB