THE DETERRENT POTENTIAL OF
VOLUNTARY R&D DISCLOSURE
Floris Dekter
Thesis Supervisor: Msc. E. Dirksen University of Amsterdam
Master thesis International Management 19th
Abstract
Publicly traded firms that voluntarily disclose R&D information without any regulatory requirements prove to be difficult to explain by prevailing literature on disclosure, spillovers and competition. A more recent stream of literature challenges traditional believes and has identified several mechanisms through which firms can obtain advantages by engaging in selective revealing strategies in which knowledge is purposefully and strategically disclosed to outside actors, including rivals. Building on these insights, this thesis aims to explore the deterrent potential of voluntary disclosure and its ability to re-shape the strategic behavior of rivals in an innovation system. In this thesis it was hypothesized that greater voluntary R&D disclosure through scientific publications results in a reduction of inventive efforts in the form of awarded utility patents by competitors. Furthermore, it was predicted that only when firms are able to signal a technological advantage through a clear and observable message, rivals are deterred. In order to test these hypotheses a longitudinal study was conducted, including a sample of eleven multinationals in the pharmaceutical and medicine manufacturing industry between 1997 and 2009. Although a negative relation between voluntary R&D disclosure and rivals’ inventive effort was found, the results were not significant.
Key words: selective revealing, strategic disclosure, voluntary disclosure, research and development, patents
Acknowledgments
This thesis has greatly benefited from the guidance of my thesis superviser Msc Dirksen and I appreciate all the helpful comments on my work. Furthermore, Hanneke Bekkering, Tijn van Vugt, Joost-Pieter Duizendstra and Daan van den Oetelaar have been of great help, motivating me throughout the process of writing this thesis.
Table of contents
1. INTRODUCTION 1
2. BACKGROUND 4
2.1 DEFINITION AND POSITION OF SELECTIVE REVEALING 4 2.2 THE DETERRENT POTENTIAL OF SELECTIVE REVEALING 6 2.3 SELECTIVE REVEALING IN ORDER TO MISLEAD RIVALS 8 2.4 SELECTIVE REVEALING AS A PATHWAY TO COLLABORATION 9
2.5 LITERATURE OVERVIEW 10
3. THEORY AND HYPOTHESIS 14
3.1 INNOVATION TIMING 14
3.2 COMPETITIVE DYNAMICS 15
3.3 INVENTIVE EFFORT AND STRATEGIC STAKES 16
4. RESEARCH DESIGN 20
4.1 DATA COLLECTION AND SAMPLE 20
4.2 VARIABLES 23 4.2.1 DEPENDENT VARIABLE 23 4.2.2 INDEPENDENT VARIABLE 24 4.2.4 MODERATING VARIABLES 25 4.3 MODEL SPECIFICATION 25 5. FINDINGS 27
6. DISCUSSION AND CONCLUSION 30
6.1 SUMMARY OF FINDINGS 30
6.2 MANAGERIAL IMPLICATIONS 31
6.3 LIMITATIONS 32
6.4 AVENUES FOR FUTURE RESEARCH 32
6.5 CONCLUSION 33
7. REFERENCES 34
8. APPENDICES 40
APPENDIX A: USPTO DEFINITION OF CLASS 424: DRUG, BIO-‐AFFECTING AND BODY TREATING COMPOSITIONS 40
APPENDIX B: SEARCH PROCESS SCOPUS 42
1. Introduction
A resource-based approach to strategic management focuses on internal resources and capabilities as fundamental drivers of firm performance and competitive advantage. It is for this reason that outgoing interfirm knowledge spillovers are considered to be detrimental, especially to innovators for which knowledge represents the most crucial resource as it accounts for the greater part of value added. If firms want to keep appropriating above normal rents from their inventions they must therefore protect the underlying knowledge base, for instance through patents or keep it hidden from competitors (Barney 1991; Peteraf 1993; Grant 1996). In other words, firms must maximize incoming and minimize outgoing knowledge spillovers.
A small but growing body of literature challenges this view and has identified several mechanisms through which firms can obtain advantages by engaging in selective revealing strategies in which knowledge is purposefully and strategically disclosed to outside actors, including rivals. Advantages are, among others, a reduced cost of public financing (Myers and Majluf 1984), gains in firm reputation and market enlargement through collective invention processes (Allen 1983), and possible cross-spillovers and agglomeration economies (Krugman 1991). More recent advances have explored the deterrent effects of selective revealing strategies. For example, Clarkson and Toh (2010) show that firms can be deterred from a technological space by allocating less effort in that space when being shown what resources rivals already possess within that space. In similar vein, Pacheco-de-Almeida and Zemsky (2012) argue that even without any compensating mechanisms, innovators may still find it optimal to freely disclose proprietary knowledge to competitors. By disclosing, innovators may persuade, i.e. deter competitors to switch from concurrent to imitative development strategies, alleviating competitive pressure. Yet another stream of literature seeks to uncover how selective revealing is used as a signaling mechanism in order to engage in cooperative behavior with competitors (Alexy et al. 2011).
Building on these insights, this thesis aims to further explore the strategic potential of voluntary disclosure and its ability to re-shape the strategic behavior of rivals in an innovation system. As such, the impact of voluntary disclosure on competitors’ patent activities is empirically examined in this thesis. It is argued that the inventive effort of rivals, measured in awarded patents, decreases as firms disclose
2
more research and development (R&D) information. This main effect is however affected by firm-specific contingencies, such as financial strength. In order to test these predictions, a sample of eleven firms in the pharmaceutical industry (NAICS1 3254) is taken. The desire to further explore the strategic potential of selective revealing strategies displayed in this thesis, is reflected in the following research question:
What is the impact of R&D disclosure through scientific publications as a selective revealing strategy on competitors' patenting efforts in the pharmaceutical industry?
By answering this question, this thesis aims to make a modest contribution to our understanding of selective revealing strategies in Business Studies. Approaching selective revealing as a deterrence mechanism differs from most previous studies in this field, in which selective revealing is considered either an accelerant of innovation, a pathway to collaboration, or a catalyst of direct compensation mechanisms such as gains in firm reputation. Some studies do approach selective as a deterrence mechanism, these studies are however concerned with providing either a theoretical model instead of an empirical data analysis (Pacheco-de-Almeida and Zemsky (2012), or include a rather different operationalization of selective revealing and have as such a different independent variable than the one of this thesis (Polidoroh and Toh 2010). From this, and even more so from section 2.5, the literature overview of this thesis, it becomes clear that research on selective revealing as a deterrent mechanism is very much needed as it’s a relatively untouched topic in Business Studies. In addition to making a modest contribution to the field of selective revealing, this thesis also aims to improve the academic understanding of resource-based theory (RBV). Along with other studies, this thesis challenges the classical RBV view, which emphasizes the need to protect resources and to minimize outgoing spillovers. Furthermore, as the results of this thesis can be of interest to managers, by providing managerial implications of the results this thesis tries to make a modest contribution to improving managerial understanding of selective revealing strategies.
In order to ground this thesis in theory and provide context, a definition of,
1 The North American Industry Classification System (NAICS) is used by government departments and
agencies to classify firms according to the type of economic activity in Canada, Mexico and the United States of America (United States Census Bureau 2013).
and background information on selective revealing will be provided. Furthermore a comprehensive theoretical overview will be presented in which relevant streams of literature and theories are displayed. In this overview, literature on innovation timing, competitive dynamics and inventive effort are incorporated. As such, the background and theoretical section of this thesis will produce a clear research gap that justifies the research question of this thesis. The discussion will focus on the significance and implications of this thesis’ findings, and contains a clear answer to the research question. Subsequently, managerial implications, limitations and promising avenues for further research are outlined in the conclusion.
This thesis focuses on how selective revealing can be best used to shape and influence the real behavior i.e. actions and reactions of competitors in order to capitalize on innovations and eventually secure leadership in product markets. As such this thesis is concerned with for-profit organizations engaging in selective revealing strategies, a group of firms this work is bound to. More specifically, the focus will be on multinational enterprises (MNEs) because in particular these firms engage in large-scale R&D activities and play an important role in the development and commercialization of new technologies in the pharmaceutical industry (Krogh et al. 2012).
4
2. Background
In order to explore the deterrent potential of voluntary R&D disclosure and how it can be used as a competitive tactic, the concept of selective revealing itself will be defined and its potential to shape and influence the behavior of competitors will be discussed. In doing so, the various strategic functionalities of selective revealing and the studies in which these functionalities are put forward, are presented in this section. In figure 1 the structure of this background on selective revealing is schematically shown, note that the academic attention to the various aspects of selective revealing through time is reflected in the topic consecution. A comprehensive overview of influential papers and various perspectives on selective revealing is provided in table 1 on pages 11-12 of this thesis.
Figure 1. Overview of the topic consecution of the background on selective revealing.
2.1 Definition and position of selective revealing2
Following Henkel (2006) and Alexy et al. (2013: 7), selective revealing is defined as the voluntary, purposeful, and irrevocable disclosure of specifically selected, often knowledge-based, and otherwise kept proprietary resources to competitors and the general public. Disclosure is considered voluntary because the information that is disclosed exceeds the mandatory requirements set by official regulators. Furthermore, because managers determine the content, these disclosures reflect conscious decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision- decision-
2 For some it may seem that the concepts selective revealing and voluntary R&D disclosure are used interchangeably in this thesis, this is however not the case. In order to prevent confusion, it is necessary to explain how these concepts relate. Selective revealing is the overarching strategy that firms adopt in order to enhance their competitive position. Engaging in such a strategy, firms disclose R&D information on voluntary bases, for instance through scientific publications or press releases.
De^ining selective revealing Classical view on selective revealing: Represents a loss of value for
the uniqueness of resources is undermined Bene^its of selective revealing: Bene^its include gains in ^irm reputation and the potential for
standard setting
Selective revealing as a strategy tool:
To deter rivals To mislead rivals To collaborate with rivals
making processes, and are as such of a strategic nature, even more so because these disclosures are meant to influence the actions of competitors.3
This thesis is specifically concerned with disclosure of verifiable information about R&D projects though for example, e.g. scientific journals, presentations at conferences or through press releases, that reflect a strategic stake in a particular technological class. Patent applications or patent reexaminations in order to signal such a stake are specifically excluded and fall beyond the scope of this thesis. Much like patenting however, disclosing information is a signaling mechanism for it sends information to competitors and other stakeholders like capital markets (James 2011; Coff et al. 2007; Harhoff et al. 2003). As such, disclosing or concealing information can be used to re-shape the strategic behavior of rivals in an innovation system. Due to the paradox of disclosure, engaging in selective revealing strategies is however not without costs or risk. Once knowledge is disclosed to outside actors, it becomes a public good and is as such no longer a source of sustained competitive advantage. Also, because competitors are indeed provided with valuable information regarding the resources of rivals, they can compete more effectively (Arrow 1962).
The notion of purposefully disclosing valuable resources contradicts with the classical view and established literature in competitive strategy, emphasizing the need to protect resources and to minimize outgoing spillovers. In this view, resources are considered to be the main sources of economic rent and the fundamental drivers of performance and competitive advantage. Therefore, firms who are able to gain and defend a favorable position in factor markets, i.e. the competition for resources, are likely to enjoy a favorable position in product markets as well and enjoy higher rates of survival. The underlying mechanism at work stems from the imperfect nature of these factor markets, in which competitors are often not able to buy certain needed resources and are forced to imitate or substitute them. Engaging in tactics in which knowledge is purposefully and voluntarily shared with competitor’s, decreases the value a firm can extract from that knowledge, for its uniqueness is undermined, and represents a loss in this classical view (Barney 1991; Barney 1986; Peteraf 1993; Dierickx and Cool 1989). Hence, firms should seek to reduce knowledge outflows, for example by trying to reduce key employee turnover, create geographical distance between themselves and main competitors, or protect R&D information through
3 Theory on how selective revealing may elicit responses from competitors, and how selective
6
patents (Alcacer and Chung 2007: 764-765).
A more recent body of literature challenges this view and has successfully shown that selective revealing and allowing outgoing knowledge spillovers may positively affect firm performance. As mentioned earlier, examples include a reduced cost of public financing (Myers and Majluf 1984), gains in firm reputation and market enlargement (Allen 1983), and possible cross-spillovers and agglomeration economies (Krugman 1991). Even more recent advances have explored the deterrent effects of selective revealing strategies. It is this stream of literature that will be further explored in the following section of this theoretical overview.
2.2 The deterrent potential of selective revealing
The study of how firms use selective revealing as a strategic deterrent tool has received little attention in Business Studies literature, although recently some scholars have explored this academic field and contributed greatly to our understanding of the subject. In particular, patent race literature offers some insights on how firms might benefit from disclosing proprietary information and in what way rivals are deterred by doing so. The theoretical models in patent race literature suggest that firms benefit, and therefore have strategic incentives to signal a technological lead via disclosure for it encourages rivals to drop out of an R&D race. The implication of these models, relevant for the study of how selective revealing is used as a strategy tool, is that credible R&D disclosure seems to reduce a rivals propensity to invest in similar technologies and reduce the number of competing patents that a rival files (Choi 1991; De Fraja 1993; Lippman and McCardle 1987).
Drawing on this literature, Pecheco-de-Almeida and Zemsky (2012) argue that even without any compensation mechanisms, innovators may still find it optimal to freely disclose proprietary knowledge to outside actors by engaging in selective revealing strategies. By disclosing, innovators may persuade competitors to switch from concurrent to imitative development strategies and as such deter these competitors to compete head to head by changing the competitive dynamics. This may be particular useful when what is at stake is not if, but when rivals develop similar technologies, which is the case in almost all industries. By engaging in a selective revealing strategy the focal firm can as such achieve delayed diffusion, rather than reduce diffusion. A central notion to take from this study is that even when
spillovers bring no direct benefits to innovators and simply result in stiffer competition, it may still be worthwhile to disclose information voluntarily, as such the authors contribute to a formal theory of the timing of technology development. Relatedly and drawing on theories regarding the allocation of inventive effort and theories on strategic stakes, Clarkson and Toh (2010) question whether the deterrence principle based on demonstrating a high stake and commitment in a particular activity, commonly found in product markets, is also applicable in factor markets i.e. in the competition for resources, with a high stake and commitment resulting in a credible threat to act in a manner unprofitable for competitors should they enter the product market (Dixit 1980; Clarkson and Toh 2010). This seems to be the case, for the authors rather convincingly show that firms can be deterred from a technological space by allocating less effort in that space when being shown what resources rivals already possess within that space. The findings of their paper suggest an approach to understanding resource heterogeneity based on how rivals’ resources are presented rather than hidden from firms, an understanding that differs from the prevailing literature. Furthermore, it suggests that a firm’s path of resource accumulation evolves through avoidance of rivals’ paths, and deterrence may therefore constitute a viable alternative theory of resource heterogeneity. De Fraja (1993) adds to the studies mentioned above and states that disadvantaged firms face a greater risk that rivals will expropriate the disclosed R&D information and use it in order to further enhance their position relative to the disadvantaged firm. The author explicitly focuses on firm-specific contingencies, showing that these exist and influence the relationship between disclosing proprietary information and the deterrence of rivals. Also consistent with the argument that firms can have strategic motivations to voluntarily and selectively reveal R&D information, are the findings of James and Shaver (2009). These authors try to explain firm heterogeneity in levels of R&D disclosure that cannot be attributed to financing needs or regulatory requirements, and argue that firms that are able to credibly signal a technological need through disclosure can deter rivals from investing in similar technologies (James and Shaver 2009).
A central and recurring concept, although differently framed and named in the various papers constituting the literature on deterrence, is that of strategic commitment or stake. Consider the following situation. At the most basic level, there
8
exist an attacker and a defender. The defender seeks to prevent an action by the attacker, who moves first, attacking or not. The defender may decide in the event of an attack to either fight or yield. The attacker is however uncertain about the defender’s ability and willingness to fight. Two conditions must be met for deterrence to occur. First, the attacker must believe that the defender is willing and able to fight – in relation to strategic disclosure, the defender must actively signal a strategic stake or commitment. Second, the cost to the attacker due to attacking rather than yielding must exceed the potential value of the prize the attacker is fighting for. If these two conditions are met, it is likely that the attacker will conclude that the prize is not worth fighting for and the defender has successfully deterred the attack4
(Clark and Montgomery 1998). The announcement of Starbucks last year to open 3000 new stores in the Americas by 2017 is an example of actively showing a strategic commitment, potentially deterring competitors. Turning to R&D disclosure, press releases issued by companies, capturing information regarding the strategic significance of R&D projects are another way of expressing a strategic commitment in a particular market. The same goes for scientific publications. Prior research by Murray and Stern (2007) shows that firms use scientific publications as a primary mechanism to disclose information about their inventions. As such, these publications are able to signal a strategic stake or technological advantage.
Deterring rivals from investing in similar technologies or markets is however not the only strategic outcome one can achieve by voluntarily disclosing information. In order to fully comprehend the strategic potential of selective revealing, sections 2.3 and 2.4 explain the ability to mislead rivals or engage in collaborative behavior by engaging in selective revealing strategies.
2.3 Selective revealing in order to mislead rivals
It becomes clear from the discussion above, that much like patenting, selective revealing should be regarded as a signaling mechanism. Also, much like patenting strategies, one can image that firms may use selective revealing as a way to mislead rivals or as a bluff. Langinier (2000; 2005) considers a dynamic model of innovation races with asymmetric information between leaders and followers regarding the improvability of an innovation. The author states that in this setting, an innovator that
4 A more comprehensive and theoretical explanation of the concept of strategic stake will be provided in section 3.3 of this thesis.
has already produced an innovation may decide to strategically use patenting decisions to mislead rivals. Eliciting responses, a patent filing usually signals that an innovation actually cannot be improved, sometimes however it will be interpreted as a decoy, and consequently will sent a competitor to an unprofitable field of research. In similar vein, regarding selective revealing, Alexy et al. (2013: 36) write: “Specifically, a firm may disclose knowledge it considers a dead end, hoping that externals commit substantial resources to find that out for themselves and giving the focal firm the opportunity to achieve lead time in an area it considers crucial.” Jansen (2010) provides more insights and a more fundamental understanding of how firms are affected by false signals. The author argues that firms apply similar R&D technologies to obtain innovations, i.e. the costs of investment are positively related amongst firms. Hence, the release of good news (selective revealing) by one firm will make its rivals more optimistic about their own opportunities in the R&D race and will therefore increase their incentive to invest themselves (Jansen 2010). The central notion here is of course that the release of good news is based on fiction rather than actual good news. Although the idea of selective revealing as a way to mislead rivals has hardly received any attention in Business Studies, and science has yet to produce systematic evidence in order to enhance academic knowledge of false signals within selective revealing strategies, it is not very difficult to presume that selective revealing is able to convey a decoy message to rivals for as mentioned above, like patenting, selective revealing is a signaling mechanism.
In the following section selective revealing is discussed as a competitive tactic to engage in collaborative behavior wilt competitors.
2.4 Selective revealing as a pathway to collaboration
Besides being a deterrent mechanism and decoy mechanism, selective revealing can also be used to engage in collaborative behavior with competitors, as Alexy et al. (2011) convincingly argue in their paper. These authors conceive selective revealing as a strategic mechanism to re-shape the collaborative behavior of other actors in a firms’ innovation ecosystem and propose that selective revealing provides an alternative to known collaboration mechanisms such as alliances, joint ventures or
10
acquisitions.5
The authors propose a model of selective revealing as a deliberate, strategic action to improve the conditions for innovation and furthermore suggest that selective revealing is a novel mechanism to shape the collaborative behavior of external actors. First, selective revealing may initiate active collaboration even under conditions of high partner uncertainty, high search costs, and when known partners are unwilling to collaborate. Second, it may cause passive and possibly unknowing collaboration by externals even when these are merely free riding on the selectively revealed knowledge by making future involuntary knowledge spillover more valuable to the focal firm and induce the external to become isomorphic. The authors further outline internal and external factors that should positively impact the firm’s propensity to engage in selective revealing, and point out the role of modularity, existing capabilities, and substitutive threats in this context. Finally, Alexy et al. specify four forms of selective revealing depending on whether the firm aims to improve its access to technologies (through problem-revealing) or markets (through solution-revealing) and whether it aims to extend existing paths or create new ones: issuespreading, agenda-shaping, product-enhancing, and niche-creating. In similar vein Henkel et al. (2013) find evidence to support the claim that in the case of open source operating system Linux, firms increasingly waived their intellectual property rights on software drivers for these firms realized that this may be beneficial for their business. It becomes clear from these, and other papers, that when firms face considerable pressure to engage in collaborative behavior with competitors, selective revealing is a viable option, one that firms already seem to use.
2.5 Literature overview
In order to gain a better understanding of the academic field surrounding selective revealing and the various perspectives on the matter, key contributions regarding selective revealing are bundled in a comprehensive literature table, table 1. For each article the research question, main findings, perspective on selective revealing and the context are outlined.
5 These arrangements usually entail a contract in order to minimize unwanted spillovers or moral
Table 1. Review of the literature on selective revealing.
Citation Research
Question
Main Findings Perspective on
selective revealing
Context
Allen (1983) What are the characteristics of collective invention, what are its consequences for the rate of invention, and what are the causes of its existence? The release of technical information to competitors is an essential feature of collective invention which allowed cumulative advance Selective revealing as a pathway to collaboration and accelerant of innovation
Case study focusing on the British iron and steel industry in the 19th century
Myers and Majluf (1984)
How are corporate financing and investment decisions affected by information asymmetries? Asymmetric information between managers and the public market leads to underinvestment
Voluntary disclosure can reduce the cost of public financing
Theoretical model
De Fraja (1993) Do firms want to disclose (part of) their discoveries ?
Firm may choose to disclose scientific knowledge to competitors, even when direct benefits are absent, for the probability of success prevails above the size of this success.
Selective revealing in order to increase a firm’s own profit as a consequence of a rivals’ success Theoretical model Harhoff et al. (2003) What incentives do users have to disclose proprietary
innovations ?
Users are motivated to freely reveal by expectations of benefits from manufacturers and other users Selective revealing as a signaling mechanism for obtaining external support and standard setting
Theoretical model; motivating examples and cases
Henkel (2006) Which firm characteristics determine revealing behavior? Under what conditions does openness leads to informal development collaboration?
The more important obtaining external support for development is, the more likely it is that firm selectively reveal information
Selective revealing in order to minimize competitive losses and to obtain external support. Selective revealing as a pathway to collaboration
Case study focusing on open source software Linux
Coff et al. (2007) Why and when do firms disclose information about technological breakthroughs that may direct rivals’ attention to their valuable capabilities?
Firm disclosure is driven by the need for complementary resources, for attention in order fully exploit advantage, by managerial opportunism
Selective revealing as a signaling mechanism for obtaining external support
Empirical sample of over 2400
breakthrough patents
James and Shaver (2009)
Do firms disclose R&D information for strategic motivations, and if so, when?
Firms are more likely to disclose when they have stronger cash positions, operate in strong IP protection regimes, have greater technological interdependencies
Selective revealing as a signaling mechanism in order to re-shape the behavior of rivals
Empirical sample of 302 communications equipment and pharmaceutical firms
12
Clarkson and Toh (2010)
Is the deterrent principle, based on demonstrating commitment and high stakes, applicable in the competition for resources?
The inventive efforts of rivals are negatively correlated with firms’ strategic stake (measured as awarded patents) in a technological space Selective revealing as a deterrent mechanism Empirical sample of 253 firms in 75 technology classes
Jansen (2010) What role does the appropriability of an innovator’s revenue play in explaining which information revelation strategy a firm uses? What are the implications for firms’ profits and the probability of innovation?
Disclosure strategies are affected by spillovers size, which has implications for firm profit and the probability of innovation
Selective revealing as a mechanism to re-shape the R&D investment behavior of rivals
Theoretical model
Alexy et al. (2011) How can selective revealing be used to re-shape the collaborative behavior of rivals in an innovation system? Selective revealing is an effective collaboration mechanism, specifically under conditions of high partner uncertainty, high coordination costs and unwilling potential collaborators
Selective revealing as a collaboration mechanism
Theoretical model
Polidoro and Toh (2011)
How does the threat of potential substitutes affect a firm’s inclination to deter imitation in the first place?
With regard to patents, firms are not uniformly inclined to deter imitation. Rather, this inclination decreases in response to
substitution threat
Selective revealing as a mechanism to re-shape the behavior of rivals and to manage the tension between inimitability and nonsubstitutability Empirical sample of 1480 patents protecting 510 pharmaceutical drugs Pacheco-de-Almeida and Zemsky (2012)
Why do firms, even without compensation mechanisms, freely disclose proprietary information?
By sharing proprietary information, firms may persuade (deter) competitors to switch from concurrent to imitative development strategies, slowing down competition Selective revealing as a mechanism to re-shape the behavior of rivals and to manage the tension between inimitability and nonsubstitutability
Theoretical model
Even though the literature overview is not fully comprehensive in the sense that all the literature regarding selective revealing in captured is one table (this is rather impossible to do due to practical problems and also beyond the scope of this thesis), the key contributions are included in table 1. Looking at this table, it becomes clear that in fact little research has been done in the field of selective revealing, especially regarding selective revealing as deterrent mechanism, while it seems that selective revealing strategies as a pathway to collaboration has received most attention in the literature. In the introduction of this thesis it was already stated that previous studies either consider selective revealing as an accelerant of innovation, as a catalyst of direct compensation mechanisms such as gains in firm reputation, or in fact as a
pathway to collaboration. Furthermore, the selective revealing literature (in general as well as the part that focuses on the deterrence) consists for the most part of theoretical papers and case studies. While only Clarkson and Toh (2010) have studied selective revealing as a deterrent mechanism, the authors however use patent reexamination certificates within a technological space as a proxy for voluntary disclosure and technological stake. As such their (operationalization of the) independent variable differs slightly from the one in this thesis. Rather than looking at patent reexamination certificates, voluntary R&D disclosures through scientific publications are used as a proxy for strategic stake in this thesis. Furthermore, in this thesis it is argued that scientific publications can signal a technological lead, rather than merely a technological stake, that can potentially deter rivals. The fact that little research has been done means that it is not possible yet to fully comprehend the strategic potential of selective revealing in general and as a deterrent mechanism, as well as the managerial implications. It is not merely of academic interest that it is necessary to address this blind spot in the Business Studies literature. The very nature of our economy and capitalist system forces firms to strive for optimal competitive strategies. A full understanding of the deterrent – and other – potential(s) of selective revealing strategies, and more particularly, the voluntary disclosure of R&D information is very useful in this regard. Empirically exploring the relationship between firms’ scientific publications and patent applications of competitors, this thesis should be considered a modest attempt at addressing the above described research gap.
In sum, although previous studies on the deterrent potential of selective revealing and its other strategic functionalities have greatly enhanced our understanding of this concept, in order to further our understanding of this topic, it is necessary to go beyond theoretical models and explore empirical operationalization’s of selective revealing as a concept. In order to do so, the literature on innovation timing needs to be taken into account. Furthermore, competitive dynamics theory and theory on inventive effort and strategic stakes are indispensable in this regard and will be drawn on in this thesis.
14
3. Theory and hypothesis
While in the previous chapter the strategic potential of selective revealing emerged, the aim of this chapter is to present a theoretical foundation for the various hypothesis of this study, as well as the hypothesis itself. A priori theoretical reasons to expect relationships between variables are necessary for the claims made otherwise are generally considered rather weak (Burnham et al. 2008: 165). The innovation timing literature is addressed in this theoretical framework as differences in innovation timing are considered both a precondition and outcome of interfirm spillovers, and are as such relevant for this thesis (Pacheco-de-Almeida and Zemsky 2012: p. 774-776). The competitive dynamics literature is incorporated for it offers insights on how engaging in selective revealing may elicit responses from competitors, and how selective revealing may even reshape behavior. Furthermore, theory on inventive effort and strategic stakes are incorporated for it will explain how such a stake or commitment reduces the inventive efforts of rivals. This chapter concludes with a brief overview of the pharmaceutical and medicine manufacturing industry, the context in which the hypotheses of this thesis are empirically examined.
3.1 Innovation timing
Innovation timing has been extensively studied in both empirical and theoretical literature. The timing of innovation is of importance to interfirm spillovers and hence to this thesis, for spillovers imply that innovation activities among firms take place sequential. It is for this reason one can make a distinction between technology leaders and followers, and that imitation and substitution occurs.6
Furthermore, interfirm spillovers have the potential to reduce the distance between followers and leaders by accelerating a followers’ innovation. As such, firm differences in innovation timing are, as Almeida and Zemsky write: “both a precondition for and an outcome of interfirm spillovers.” (Ameida and Zemsky 2012: 774).
A situation in which firms develop innovations sequentially can arise for instance when a first-mover was uniquely aware of the market opportunity of the innovation or through sustainable technology leadership (Almeida and Zemsky 2012; Lieberman and Montgomery 1988: 41-47). Sequential development can however also arise due to the conscious choice of firms to act as followers for these firms can
potentially benefit from interfirm spillovers while spending relatively little resources themselves. In particular with large spillovers, followers can acquire the technology relatively fast and at a fraction of the leaders’ costs (Almeida and Zemsky 2012: 784). This free-rider effect occurs in industries in which imitation costs are usually lower than innovation costs. Furthermore, followers may benefit from a delayed resolution of technological and market uncertainty (Lieberman and Montgomery 1988).
3.2 Competitive dynamics
Besides an understanding of the consequences of differences in innovation timing, an understanding of competitive dynamics is of importance to this thesis. The competitive dynamics literature offers insights on how engaging in selective revealing may elicit responses from competitors, and how selective revealing may even reshape behavior. Drawing on the seminal work of Joseph Schumpeter (1934; 1942), the competitive dynamics literature attempts to uncover why some firm interactions create destructive patterns while others have beneficial consequences. As such, it is concerned with real behavior of firms in the form of actions (moves) and reactions (countermoves) over a specific time frame that affects competitors, competitive advantage and performance. Firms act in order to enhance their profitability, increase market share or their overall competitive advantage or industry position, this in turn elicits competitive reactions from competitors trying to block or imitate the initial action, creating competitive dynamics7
(Ketchen et al. 2004; Smith et al. 2001; Chen et al. 1992).
While earlier research was mainly concerned with action, response, and rival firm characteristics, more recent work is focused on the competitive dynamics between industry leaders and followers (Ketchen et al. 2004: 781). Amongst others, Ferrier et al. (1999) find that leaders are more likely to lose market share and can even be dethroned by rivals seeking to challenge the status quo, when they initiate fewer moves with a smaller latitude, slower than challengers. This, in fact mirrors Schumpeter’s notion of ‘creative destruction’ in which the gains obtained by leaders motivate rivals to elicit reactions (countermoves) in order to overtake the leaders and enjoy similar profits (Schumpeter 1934; Smith et al. 2001). Tohatosu, for example,
7 Note that the competitive dynamics literature presupposes that actions and reactions transmit
information, in the case of this thesis, through disclosing information (see on p. 9 of this thesis the section on disclosing information as a signaling mechanism).
16
the leading motorcycle producer in Japan, failed to withstand the competitive pressure from rivals and was eventually overtaken by Honda (Abegglen and Stalk 1985). Drawing on these notions, it seems rational, or even necessary for technology leaders to adopt a proactive strategy in order to prevent dethroning by followers. As such, the competitive dynamics literature stresses the importance of improving our understanding of selective revealing strategies, and in fact other proactive strategies.
3.3 Inventive effort and strategic stakes
Factors that drive the decision of how firms allocate their inventive efforts are either internal or external to the firm. Internal factors are numerous and include for instance a performance gap that needs certain R&D investments to be resolved, or an exhaustion of internal opportunities, inducing firms to explore new innovative areas. Furthermore, because a successful outcome of an R&D project is often the cumulative result of adhering to a set of consistent policies over a period of time and the knowledge accumulation of a firm consequently a very path dependent process, firms tend to allocate their inventive efforts towards familiar places (Helfat 1994: 1744-1745). In order to describe the process of how firms consider external factors when deciding how to allocate their inventive efforts, it is necessary to turn briefly to evolutionary theory. Evolutionary theory focuses on the technological changes that initiate periods of intense technical variation and selection, culminating in a single dominant design in a product class. Such an era of ferment is usually followed by a period of incremental technological progress, only to be broken by subsequent technological discontinuous advances (Anderson and Tushman 1990). Most of the time inventions occur through recombining existing components and technologies. Due to the cumulative nature of this process, firms tend to evaluate initial and all subsequent inventive opportunities in their decision where to allocate inventive efforts. If the inventive opportunities seem slim, for example due to a substantial lead time, firms may choose not to follow industry leaders (Lippman and McCardle 1987). This external allocation process reveals how signaling a strategic commitment or technological advantage may reduce the inventive efforts of rivals, and is in fact a striking example of deterrence as described in section 2.2 of this thesis. Whenever a firm is able to signal a general commitment towards R&D investments, and as such signals a willingness to preempt certain innovative opportunities, the prospects of an
R&D investment for rivals become less attractive. More specifically, it is argued in this thesis that R&D disclosures through scientific publications decrease the level of inventive effort of competitors. Prior research by Murray and Stern (2007) shows that firms use scientific publications as a primary mechanism to disclose information about their inventions. Furthermore, the authors write: “an increasing amount of industry-produced knowledge is also disclosed through scientific publication, often to serve specific strategic purposes.” (Murray and Stern 2007: 651). Firms operating in R&D intensive industries (and especially in first-to-file intellectual property regimes) have strategic motivations not only to patent novel innovations, but also to prevent competitors from doing the same. Not only will this increase a firms’ intellectual property protection and increase bargaining positions, moreover it contributes to the overall competitiveness of the firm. In conclusion, one can expect that an increase in a firms’ strategic commitment results in a reduction of inventive effort for its rivals. Hence, the first hypothesis of this thesis.
Hypothesis 1: Greater voluntary R&D disclosure through scientific publications results in a reduction of inventive efforts in the form of awarded utility patents by competitors.
The extent to which a firm is able to credibly convey its message and deter a competitor from investing in similar R&D activities depends on the quality of the signal. The quality of the signal is in turn affected by the type of disclosure and the characteristics of the sender. In section 2.2 of this thesis, a reference was made to De Fraja (1993) who argued that firm-specific contingencies matter in the relationship between selective revealing and deterrence. Only when firms are able to signal a technological advantage through a clear and observable message, rivals are deterred. Furthermore, conducting an empirical study, Clark and Montgomery (1998) find evidence that the more successful a firm is, the more the firm will be perceived as a credible defender by its rivals. Drawing on this logic, it is argued in this study that when a firm is able to increase its credibility towards an R&D commitment and the potential to capitalize on the inventive efforts, patent applications by competitors decrease even further. This statement is further supported by the fact that R&D activities (especially in the pharmaceutical industry) are usually very capital intensive.
18
Grueber and Studt (2014: 23) write in the 2014 Global R&D funding forecast, that in global life science, the ten leading R&D firms are biopharmaceutical manufacturers. Furthermore, in the Pharmaceutical Research Manufacturers of America (PhRMA) 2014 biopharmaceutical industry profile it is stated that the average cost to develop a drug in the early 2000s was 1.2 billion dollar (PhRMA 2014). Hence, one can imagine that the cash position, R&D spending figures or just sheer firm size matter in this regard. Taking this in consideration, hypothesis 2a, 2b and 2c of this thesis are the following.
Hypothesis 2a: Greater firms' potentials to capitalize on inventive efforts due to stronger financial capabilities, will subsequently reduce the inventive effort by competitors in the form of patent applications.
Hypothesis 2b: Greater firms' potentials to capitalize on inventive efforts due to higher R&D expenses, will subsequently reduce the inventive effort by competitors in the form of patent applications.
Hypothesis 2c: Greater firms' potentials to capitalize on inventive efforts due to firm size, will subsequently reduce the inventive effort by competitors in the form of patent applications.
3.4 Pharmaceutical and medicine manufacturing summary
The pharmaceutical and medicine manufacturing industry consists of hundreds of firms, but the top ten companies generally account for a significant amount of industry sales and profit (it are those industry leaders that this thesis focuses on). R&D costs in the industry have increased significantly in recent years, just as the life-cycle sales for new drugs that reach the marketplace. It seems, as DiMassi (2002: 2) writes that: “For the industry as a whole, the profitability of new drug development has kept pace with the returns the investors require to justify investing in assets that are as financially risky as new drug development”. Furthermore, industry growth has always been relatively high and the industry is expected to grow in the foreseeable future as the need for continued development of new treatments is also great due to
demographic trends and the growing socioeconomic burden of disease (Long and Works 2013).
The development of new medicines remains however a long and complex process, posing risks to individual pharmaceutical companies (PhRMA 2014: 50). It is estimated that it takes between 10 to 15 years and approximately 1.2 billion dollars to yield a single Food and Drug Administration (FDA) approved drug (Long and Works 2013: 2). Furthermore, companies are facing challenges associated with loss of patents rights, presenting opportunities for generic brands to conquer a larger share of the market. According to the PhRMA, the average patent life after FDA approval of an innovative brand-name medicine is 12.6 years (PhRMA 2014: 49-50).
These risks and competitive pressures call for an adequate response in the form of deterrent strategies. Firms in the pharmaceutical and medicine manufacturing industry that want to deter the emergence of concurrent drugs while their own drug is still under development, often turn to patent litigation as the predominant means of deterrence (Polidoro and Toh 2010: 376; Danzon 2000). Signaling a technological lead through voluntary disclosure seems however another viable option for firms in this market. Furthermore, due to the extensive time lapse between the creation and market introduction of most drugs, other forms of deterrence such as investments in excess production capacity are considered less applicable for this particular industry (Polidoro and Toh 2010: 376).
20
4. Research design
In this chapter the research design of the statistical model will be presented and as such the way in which the hypothesis are tested will become clear. Before providing a description of the statistical model however, it is shown what data is used and why, as such the way the sample of this study is formed will be explained. Furthermore, the variables are specified and the way they are operationalized will be explained. The overall data process will be schematically presented in figure 2 at page 23.
4.1 Data collection and sample
The hypotheses of this thesis are empirically examined in the context of the pharmaceutical and medicine manufacturing industry (North American Industry Classification System (NAICS) 3254) over a period of 13 years, ranging from 1997 to 2009. The year range is a reflection of the availability of data on the variables in this thesis. This thesis focuses on the pharmaceutical industry because firm R&D spending’s are amongst the highest in the world. In section 3.3 of this thesis it was mentioned that in life sciences, the ten leading R&D firms are biopharmaceutical manufacturers. Likewise, Cohen (2005: 80) writes: “The pharmaceutical industry spends more on R&D as a percentage of gross output than any industry except aerospace. Pharmaceutical industry R&D expenditures (unadjusted for gross domestic product (GDP)), as reported yearly in the Pharmaceutical Research Manufacturers of America (PhRMA) Annual Survey, increased exponentially from 1970 to 2001, with some evidence of a moderation of spending growth more recently.“ Furthermore, this industry allows for R&D allocation uncertainties and opportunities due to a wide variety of technologies used and a relatively high pace of technological change (Grueber et al. 2014: 23). In order to study the relationship between R&D disclosure and deterrence at an appropriate level, the main technology class created by the United States Patent and Trademark Office (USPTO), rather than a SIC (Standard Industrial Classification) or NAICS code is used as a proxy for technological space. As such, a significant error in measuring patenting activity is prevented. One can image that firms operate in multiple technological classes and stating a strategic stake in a particular class does not deter rivals in other classes. The sampling frameof this thesis mirrors such a USPTO class, for patent applications are measured at the same level, and scientific publications can be matched accordingly.
The unit of analysis is the firm(i)-class(j)-year(t), which means that each observation captures a firms inventive effort in a particular technological class in a certain year, as a function of scientific publications within class j in a previous year.
Class 424, compiled by the USPTO is called ‘drug, bio-affecting and body treating compositions’ in which various technologies are grouped according to similar functions or uses8
. The sample includes the eleven largest MNCs in terms of R&D expenditure that are represented in this class 424. For all included firms it is checked whether they applied for patents and disclosed R&D information through scientific publications at any point in the sample period. This sampling process yields a final sample of 143 firm year observations for eleven firms from 1997 to 2009. The preferred solution to deal with missing values in the dataset in the hot deck imputation procedure, as suggested by Myers. Hot deck imputation involves replacing missing values with values of similar donors in the dataset that match the done in researcher-determined categories (Myers 2011: 11). The nature of this thesis’ dataset violates however important boundary conditions for using this procedure as it merely includes variables that are of theoretical interest to the research question, furthermore the variables contain continuous rather than discrete values. Consequently, expectation maximization (EM) is used in order to deal with missing values. Conducting a Little’s missing completely at random (MCAR) test reveals that the missing data is indeed missing at random, an important condition for EM.
Disclosure data are drawn from scientific publications issued by firms, which are obtained from Scopus, the world’s largest abstract and citation database of peer-reviewed literature, delivering a comprehensive overview of research output in various academic fields, including science and medicine. Scopus was chosen for its database contains 4600 health science titles, including 100% MEDLINE9
coverage as well as a 100% EMBASE10
coverage (Burnham 2006: 1). As such it is very suitable for searching scientific publications related to the pharmaceutical industry. Furthermore, Scopus includes a wider vary of journals than for instance PubMed or Web of Science, while its citation analysis is faster and includes more articles (Falagas et al. 2008: 342). Other major scientific databases such as Google Scholar
8 For a full definition and description of USPTO class 424, see Apendix A on page 42.
9 The MEDLINE database contains journal citations and abstracts for biomedical literature from around
the world.
10 EMBASE is a biomedical database with over 28 million indexed records from thousands of
22
are rejected for its inconsistent accuracy (Falagas et al. 2008: 338), while Web of Science is excluded for its coverage lacks a pharmaceutical focus.
In order to obtain an accurate count on the independent variable, the Scopus database was used to search for scientific contributions from 1997 to 2009, published by the sample firms (see Appendix B for search queries and screenshots). In Scopus, first an affiliation search was conducted to account for the fact that large firms in the pharmaceutical industry often have multiple research groups, using slightly different names that publish scientific articles. Using the Scopus Affiliation Identifier, it is possible to match documents from a single organization, even if the organization is cited (slightly) differently. Second, for each firm a publication search was conduction in which the documents from different affiliations representing the same firm were combined. Additionally, a year filter ranging from 1997 to 2009 was incorporated, as well as filters limiting the search to articles published in scientific journals in subject area ‘pharmacology, toxicology and pharmaceutics’ to match technologic class 424 in the USPTO patent database. For all the firms in the sample, this resulted in 19814 articles.
Data regarding the patenting behavior of firms are obtained from the USPTO database. The Patent Technology Monitoring Team (PTMT) periodically issues general statistics and miscellaneous reports that profile patenting activity at the USPTO. Regarding technology classes, the PTMT issues extended year sets, containing patent counts in the various classes with breakouts by organization (USPTO 2014a).
Data on firm characteristics and financials are drawn from annual 10-K reports and 20-F reports provided by the U.S. Securities and Exchange commission (SEC), as well as company annuals and company financial reports. Federal securities laws require publicly traded companies to disclose information on an ongoing basis. The annual 10-K reports provide an overview of the company’s business and financial condition (U.S. Securities and Exchange commission 2014). 20-F forms are SEC filings submitted to the US SEC by certain foreign private issuers that provide company information. From over 300 reports, information regarding employee counts, R&D spending and the total current assets of firms was gathered. Financial data reported in currencies other than US dollars are converted to US dollars using
historical exchange rates. Furthermore, all financial data is conform US generally accepted accounting principles (US GAAP) to insure consistency throughout the data.
Figure 2. Overview of the data collection and analysis.
4.2 Variables
4.2.1 Dependent variable
The dependent variable – rivals’ inventive effortnonit+3 – is measured as the number of granted utility patents for firms included in the sample in the same technological class as the focal firm in year t+3. Inventive effort is measured in year t+3 in order to account for time lag between signaling a strategic stake and the subsequent reaction of rivals (for which a one year lag is deemed appropriate in line with previous research (Clarkson and Toh 2010; Polidoro and Toh 2011), as well as the time lapse between the application and the actual granting of a patent (which takes on average about two years in the case of the USPTO) (Hall et al. 2001: 9). Using a total non i count of rivals’ inventive effort, for each year every firm in the sample can be isolated and regressed against all competitors. Utility patents, also referred to as patents for invention, are issued for the invention of a new and useful process, machine, manufacture, or composition of matter, or a new and useful improvement thereof, as such, rather than reissue patents, utility patents reflects inventive efforts (USPTO
Comprising sample according to ^irm R&D
expenditures by reviewing industry
reports
Conducting a search for scienti^ic publications in
Scopus for the sample ^irms
Conducting a patent search in the USPTO database for the sample
^irms
Scanning over 300 10-‐ Ks, 20-‐Fs and annual company reports for
additional ^irm information. Importing the raw data
and constructing variables in SPSS
Statistics Handling missing data
using Expectation Maximization and constructing additional variables by combining
raw data
Conducting the statistical analysis in
SPSS
24
2014b). Although relied upon in prior research, patent-based measures pose well-known methodological issues (Ethiraj 2007; Clarkson and Toh 2010). It is argued that the propensity and reasons for patenting differ greatly amongst industries (Cohen et al. 2000). Furthermore, it is considered an imperfect measure of innovative output (Trajtenberg 1990). As an indicator of inventive effort however, it is considered a fairly reliable measure (Cohen et al. 2000). Although some might argue that a count of patent application is more appropriate in the context of this thesis, using a count of granted patents avoids having to deal with the truncation problem (this besides the fact that data on patent applications in technology classes is not available). As the time series move closer to last date in the data set, patent data as distributed by the year of patent application will increasingly suffer from missing observations consisting of patents applications that have been filed in previous years but have not yet been granted (Hall et al. 2001: 10). As such, operationalizing the dependent variable in this way is deemed appropriate for determining strategic benefits of voluntary R&D disclosure to the extent that it deters rivals from investing in a similar technological space.
4.2.2 Independent variable
In order to capture a firms’ stake and advantage in a technological space – Scientific publicationsit -, a count of scientific publications is conducted. More specifically,
R&D disclosures are measured as the number of scientific publications about the same technological class, issued by firms i in year t. Scientific publications are preferred over 10-Qs or 10-Ks in order to capture the voluntary aspect of disclosure. As mentioned on page 8 of this thesis, annual and quarterly filings are mandatory for publicly listed firms and as are as such an imperfect measure of voluntary R&D disclosure. Press releases containing R&D information are disregarded for these fail to capture the strategic aspect of R&D disclosure. Although there are databases (for instance Lexis Nexus Academic) containing numerous press releases that contain technical information regarding firm R&D, it is nearly impossible to verify whether a press release is issued by the firm or that these originate from journalists or news wires and do therefore not reflect a conscious voluntary R&D disclosure.
4.2.4 Moderating variables
In order to test hypothesis 2a, 2b and 2c of this thesis, it is necessary to incorporate firm-specific contingencies in the analysis by constructing three more variables that will act as moderators in the statistical analysis of the data. Because a moderator variable changes the strength of an effect or relationship between two variables, usually the main affect, moderators are able to indicate when or under what conditions a particular effect can be expected. As mentioned in section 3.3 of this thesis, drawing on previous research it is argued that when a firm is able to increase its credibility towards an R&D commitment and the potential to capitalize on the inventive efforts, patent applications by competitors decrease even further, exacerbating the main effect of H1.
Financial capabilities, R&D expenses and sheer firm size are used as proxies for such a commitment and potential to capitalize on inventive effort.
The variable - Financial capabilitiesit - is measured as the total current assets
of firm i in year t.
The variable - R&D expensesit – is measured as the amount of R&D spending
of firm i in year t.
The variable - Employeesit – is measured as the number of employees of firm i
in year t.
4.3 Model specification
The statistical approach that was used to test the hypothesis of this thesis, is to regress rivals’ inventive effort – measured as a count of granted patents – on the R&D disclosures – measured as a count of scientific publications – of firm i in SPSS 22. Also, possible moderating effects are explored in the regression model. Because the dependent variable of this thesis is a count of rivals’ granted patents, count models are appropriate for analyzing the hypothesized effects. Furthermore, by also taking the repeated measures on the units of analysis into account, a mixed model is constructed including fixed and random effects. An important boundary condition for this model is that the data is normally distributed, analyzing the histogram with normal curve in SPSS reveals that the dataset of this thesis does not violate this condition for the mean and median are approximately the same. Before presenting the findings of the model in the next chapter of this thesis, descriptive statistics containing frequencies, means
26
and standard deviations of the dependent and independent variable are provided, as well as a table containing pair-wise correlations.
