Resumen de Three essays on outbound open innovation and intellectual property protection
This thesis consists of three chapters studying topics related to the corporate innovation management. More specifically, the first two chapters discuss the outbound open innovation phenomenon and its relationship with the firm’s behavior in technology and labor markets, respectively. The last chapter links the firm’s litigiousness with its collaborative activities.
Chapter 1 (titled “What’s there to Gain? Outbound Openness and Markets for Technology,” co-authored with Araksya Ayvazyan), focuses on the outbound type of open innovation, in a setting, where the firm makes its knowledge (or part of it) for free to the outside parties, and investigate its implications for the focal firm. Specifically, it addresses the question of “Why do corporate firms adopt an outbound open approach in their intellectual property (IP) strategy, especially without direct financial gains in exchange?” and suggests two main mechanisms that the firm may use to capitalize on such a practice. These channels primarily concern strategic openness’ facilitating inward and outward knowledge flows, which in turn, induces the focal firm’s engagement in markets for technology, in terms of transactions for IP rights. This may give rise to potential externalities for the opening up firm.
An important element in our theoretical arguments is that outbound openness reduces transaction and negotiation costs, as well as litigation threats (Wen, Ceccagnoli, & Forman, 2016). Due to the decrease in access costs and litigation risks, other firms get encouraged to become more involved (Boudreau, 2010) and subsequently build more knowledge. Further, the focal firm can respond to the advances in innovation in different ways: by selectively buying innovations, that other firms have developed via relying on the liberated knowledge, or by making further internal developments through combining its expertise with the new knowledge created by others in a specific technology. At the same time, because strategic openness indirectly enforces the outsiders’ commitment to the liberated knowledge and technologies, as they create more complementary assets, thereby increasing their demand for the subsequent knowledge, the focal firm gets more opportunities for selectively selling its other internally developed knowledge.
We test our predictions on patent data, making use of IBM’s patent pledge of 2005 as a shock to the level of openness in the IP strategy of IBM (Wen et. al, 2016). Having the sample period from 1999 to 2010 allows us to implement a difference-in-differences approach to explore the consequences of the openness decision on the firm’s engagement in markets for technology and the degree to and channels through which IBM utilizes the follow-on spillovers. In particular, we conduct analyses at two different levels. The first set of analyses includes the level of knowledge domain, represented by technological classes, according to The United States Patent Classification (USPC). We use these class-level analyses (with class-year type of observations) to explain the temporal variation outcomes of patent trading (i.e. selling and buying) activities, using a score of openness for each technological class. We identify the treated group of technological classes, if the class includes any of the patents that IBM pledged in 2005. The classes without any pledged patent, to which IBM had significantly contributed up until 2005, serve as the control group of technological classes.
In the second set of analyses, we use data at the patent-level (with patent-year type of observations). The goal of these analyses is to track the effect of the patent pledge on patent trading in a more detailed and direct manner and to test the impact on building upon the subsequently created external knowledge, by examining the pledged patents and their spillovers, e.g. citing patents, in comparison to similar non-liberated patents and the latter’s spillovers. In these difference-in-differences analyses, we compare the pledged patents, i.e. initial treated group, to a selected group of similar patents, i.e. initial control group, which we construct by matching each pledged patent with other patents using a text matching algorithm to measure technological similarity, following Arts, Cassiman, & Gomez (2017). We extend these initial treated and control groups of patents by their corresponding direct or indirect forward citations.
Altogether, our results show that after 2005, IBM buys and sells more patents, proportionally to the level of openness in its IP strategy. However, we find stronger evidence for increased selling, rather than buying activities by IBM. We also find that IBM continues to create further knowledge developments, building upon external sources of subsequent knowledge more than on its own subsequent knowledge. The results are robust to integrating the launch of Open Invention Network (OIN), – an organization, established by major players in the software market to support the Open Source Software (OSS) community in advancing OSS programs, by offering patents for royalty-free licenses, – in 2005, in our analyses. Additional checks on the effect of openness over time reveal that the impact takes some time to show up in the firm’s patenting and IP trading activities and fades away after a few years. Overall, our results provide support for the proposed mechanisms.
I study in the second chapter, “Opening Up Intellectual Property Strategy: Implications for Inventor Mobility,” whether the practice of outbound openness makes the inventors of the firm more or less mobile. Specifically, I focus on the effect of the firm’s openness level in a field of knowledge on the mobility of its inventors that have expertise in this field. In other words, I examin how a firm’s move towards outbound open innovation strategy via opening up the firm’s intellectual property (IP) affects inventor mobility from the firm. Interestingly, the answer is not straightforward because there are arguments to expect both an increase as well as a decrease in inventor mobility as a consequence of the move towards openness. The key argument relies on the degree of complementarity or substitutability between the codified knowledge and the inventors’ tacit knowledge, and how it is altered by outbound openness.
Arguments can be made both for a reduction and for an increase in inventor mobility, following the facilitation of knowledge out-flow via opening up firm’s IP strategy. On the one hand, scholars argue that aggressiveness in IP rights’ use and enforcement can reduce employee turnover (Kim & Marschke, 2005; Ganco et al., 2015). Then, by considering the practice of granting access to the firm's knowledge to outsiders as a “soft” strategy, the firm’s inventors will become more, rather than less mobile. The reason is that with more openness, the perceived risks and costs of recruiting the firm’s knowledge labor will be reduced, resulting in increased likelihood that more inventors will leave the firm. In addition, it has been shown that openness in the innovation process increases the involvement of the outsiders, due to a reduction in costs and risks in using or implementing the opening up firm's knowledge (Murray et al., 2016; Wen, Ceccagnoli, & Forman, 2016). Therefore, with more engagement in the knowledge fields, the demand for the experienced labor in the liberated technologies will increase. On the other hand, allowing for inside-out flows of knowledge may make the firm's inventors less mobile. The underlying logic is the following. By voluntarily granting access to proprietary knowledge, the focal firm reduces the incentives for outsiders to recruit the firm’s inventors. Thereby, the voluntary leakage of information can serve firms as a tool to reduce the drain of skilled labor embodying such information (Lewis & Dennis, 2001; Guarino & Tedeschi, 2006).
As the theoretical arguments from the prior literature provide conflicting conjectures, I decide to look closely at the factors driving the effect. In this paper, I draw on arguments linking the degree of complementarity or substitutability of the inventors' tacit knowledge with the codified opened-up knowledge. Specifically, I suggest that in the presence of complementarity (substitutability) between the inventors’ knowledge and the opened-up knowledge, the mobility of inventors outside the firm is likely to increase (decrease). The main rationale I propose is that outsiders’ demand for the inventor will increase (decrease), when the inventors’ knowledge is perceived as necessary (replaceable) for exploiting the opened-up knowledge. I elaborate on the potential situations and scenarios where outsiders perceive the two types of knowledge, i.e. the inventor’s tacit knowledge and the opened-up knowledge, as substitutable or complementary. As such, I discuss instances when outsiders a) practiced inventing around before the decision of the firm to open up (part of) its knowledge, b) seek to commercially implement the opened-up knowledge, or c) aim to internally develop follow-up innovations. Section III provides reasons for why the substitutability mechanism is more likely to dominate. First, outbound openness per se reduces the outsiders’ need to invent around the liberated knowledge, which provides a substitute to hiring inventors from the opening up firm. And second, unlike the knowledge and the experience of the inventors that created the opened-up knowledge, those of the inventors with general expertise in the field (the focus of the current study), are not deemed complementary to the codified liberated knowledge in the contexts of commercial implementation or building upon practices. The reason is that these inventors do not necessarily possess specific information on how to commercially exploit or advance the opened-up knowledge. Thus, I expect that inventors with general knowledge in the field will experience lower turnover rates after their employers’ decision to open up their IP strategy. To gain further insights on the suggested direction of the effect, I investigate inventors’ knowledge characteristics that may moderate the effect of outbound openness on inventors’ mobility. Specifically, I hypothesize that a) when the inventors’ knowledge is more basic, b) when the inventors work with lower number of co-inventors, and c) when the inventors’ knowledge is in areas characterized with higher technological cumulativeness, the negative effect of outbound openness will be strengthened.
In order to address my hypotheses, I utilize the decision by International Business Machines Corporation (IBM) to (partially) open up its IP at a specific point in time to analyze whether and how this decision changed labor mobility patterns. IBM is as a very good case, since it used to be known for a strict adherence to the closed model of innovation (Mayle, 2006). However, IBM made a radical shift to the open innovation model (Chesbrough, 2003), by supporting open-source software (OSS) through many actions, as such pledging 500 patents in 2005. I use this patent pledge as a proxy for IBM’s adoption of openness in its IP strategy in multiple areas of knowledge. By extracting information about IBM’s inventors and their patents from the USPTO database, I detect inventor mobility and investigate the implications of IBM shift to open governance of their IP’s on the leaving of their inventors. I use a special version of difference-in-differences, with a continuous treatment variable, in order to capture the variations in the levels of the inventors’ knowledge openness. The results suggest that opening up the corporate innovation strategy, on average, lessens the chances of outbound labor mobility.
This paper provides several contributions to the literature. First, by considering the complementarity and substitutability mechanisms between the codified knowledge and the inventor’s (tacit) knowledge, this study presents a new angle of looking at the dynamism between the knowledge flow and the inventor mobility. The main finding of this study suggests that strategically practicing outbound openness can bring along benefits of retaining knowledge workers for the focal firm. In addition, my findings support the proposition that firms “learn by hiring” where the knowledge is a driver for recruiting external inventors. Finally, this study contributes broadly to the literature on open innovation. Since the seminal work of Chesbrough (2006a), firms have started to acknowledge the importance of the so-called "outside-in" (i.e. inbound) and "inside-out" (i.e. outbound) knowledge flows (Tranekjer & Knudsen, 2012). The academic literature, however, has mainly focused on the inbound type of open innovation. By disentangling the effect of outbound openness from inbound open innovation, I attempt to provide a deeper and better understanding of open innovation phenomenon (Schroll & Mild, 2011).
The third chapter (titled “Patent Enforcement Strength and Collaboration: The Effect of Firm’s Litigiousness on Alliances Intensity,” co-authored with Eduardo Meleroa, David Wehrheim) argues that a weaker enforcement of property rights generates two opposite forces on firms’ research collaborations. Consequently, it may increase or reduce collaborations depending on the intensity of each of these forces. As implicit in previous research on alliances, weaker enforcement increases the expected cost of any unintended knowledge leakage that may happen during the partnership, thus discouraging it. On the other hand, to the extent that the patent system operates as a block-road to innovation by generating uncertainty and encouraging frivolous litigation (Lemley and Shapiro, 2005; Jaffe and Lerner, 2011; Mezzanotti, 2019), weaker enforcement increases the expected rewards to collaborative investments.
Within this framework, we analyze the consequences of recent moves in the U.S. patent system toward weaker patent enforcement on the prevalence of alliances, which will depend on the importance of the two opposed forces outlined above. Due to the different effects of weaker patent enforcement, its net impact is theoretically ambiguous. Nonetheless, we hypothesize that weaker patent enforcement tends to decrease the formation of alliances between partners of different sizes more than alliances between partners of similar size. These asymmetric alliances are typically characterized by involving a small partner that is particularly concerned about knowledge leakage to the counterpart (Katila et al., 2008). Alliances between partners of a similar size, on the other hand, are not expected to be discouraged, since they develop in a less unbalanced context. Secondly, we expect a relative increase in the prevalence of collaborations between relational partners, i.e., partners with a history of past collaboration together. Relational partners are likely to have developed mechanisms to avoid opportunistic appropriation of knowledge, and consequently experience less concerns of knowledge leakage (Deeds and Hill, 1999). Therefore, they are only affected by the “bright” side of a weaker patent enforcement, through a reduced legal uncertainty regarding the outcome of the alliance. Finally, we also predict that the positive aspects of a weaker enforcement for R&D collaboration agreements will tend to be relatively more important than the negative ones for firms performing more basic-science research activities. The reason is that because basic-science-related knowledge is idiosyncratically more difficult to absorb by potential partners (Lane and Lubatkin, 1998), weaker enforcement imposes fewer risks on collaborations for these companies. To test these hypotheses, we exploit the legal change in the treatment of patent infringement created by the U.S. Supreme Court decision “eBay v. MercExchange” in 2006, which ended the practice of automatically granting a permanent injunction after a patent violation. There is a consensus among legal scholars that this decision effectively weakened the enforcement of patent protection (Tang, 2006; Bessen and Meurer, 2008). We examine the impact of such a decision on firm’s decision to collaborate by exploiting heterogeneity among companies in their tendency to litigate (Mezzanotti, 2019) as the treatment variable, and combining data from Securities Data Company (SDC) Platinum for alliances, patent litigation data from Marco, Tesfayesus, and Toole (2017), patent data from Patentview dataset for co-patenting measures, data on the scientific references from Kevin A. Bryan, Yasin Ozcan, and Bhaven N. Sampat (2019), and control variables from Arora, Belenzon and Sheer (2019). Overall, our analysis suggests that the decrease in the patent protection enforcement leads companies in our sample to engage in fewer alliances. The results support our main predictions regarding the symmetric alliances in comparison to asymmetric alliances and the relational alliances in comparison to alliances with de novo partners. The moderation effect of knowledge basicness is also supported by our analyses.
