This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

Transcription

1 This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier s archiving and manuscript policies are encouraged to visit:

2 Journal of Economic Behavior & Organization 84 (2012) Contents lists available at SciVerse ScienceDirect Journal of Economic Behavior & Organization j ourna l ho me pag e: ebo Strategic uses of patents in markets for technology: A story of fabless firms, brokers and trolls Julien Pénin BETA, CNRS-UMR 7522, University of Strasbourg, 61 avenue de la Forêt Noire, Strasbourg Cedex, France a r t i c l e i n f o Article history: Received 28 June 2011 Received in revised form 7 May 2012 Accepted 8 September 2012 Available online 23 October 2012 Keywords: Markets for technology Technological firms Patent Troll Broker Start-up a b s t r a c t This paper analyses the strategic uses of patents in markets for technology. In particular, we study the behaviors of technological firms, patent brokers and patent trolls and explore their consequences on the amount of R&D investments of the economy. We show that patent brokers, in their pure form, are R&D increasing (i.e. they increase the R&D investments of both supplier and manufacturing firms) while patent trolls, in their pure form, are R&D decreasing. We also show that, paradoxically, an optimal positive level of trolling may exist if the probability to secure market transactions and the probability to encounter trolls are positively correlated. Then we discuss the normative implications of this research. Among others, we argue that changes in patent laws are needed in order to limit the possibility of trolling; however, the difficulty is that those changes might also affect the development of technological firms and brokers. Several possible evolutions of international patent laws are hence discussed Elsevier B.V. All rights reserved. 1. Introduction In the patent literature, non-practicing entities (hereinafter NPEs) are firms that rarely or never practice their patents, and instead focus on earning licensing fees (Shrestha, 2010, p. 114). In this context, practicing a patent means manufacturing and selling the product or process described in the patent and using the patent to secure a monopoly over this product or process. NPEs do not use their patents in this way because, contrary to manufacturing firms, they do not manufacture any innovative tangible product. NPEs are an important feature of the knowledge-based economy, which is largely linked to the development of independent technological firms (following the logic of open innovation, Chesbrough, 2003). Yet, the issue of the social desirability of NPEs is vividly debated in the economic, legal and managerial literature on intellectual property rights. On the one hand, many authors are worried about the emergence of perverse patenting strategies adopted by patent trolls 1 (or patent sharks) and warn against their potential detrimental effect on R&D investments, especially those coming from manufacturing firms (Jaffe and Lerner, 2004; Lemley, 2007; Merges, 2009; Bessen et al., 2011). The damage caused by patent trolls may be specifically important in fields where technology is complex, such as electronics and software. On the other hand, other scholars emphasize the importance of NPEs in order to secure markets for technology, thus fostering a vertical division of labor and Corresponding author. address: penin@unistra.fr 1 The expression patent troll was first used by Peter Detkin in 2001, while he was working for Intel and complained about the patenting strategy of a NPE, Tech-Search, with which he was in litigation. Detkin claimed that a patent troll is somebody who tries to make a lot of money off a patent that they are not practicing and have no intention of practicing and in most cases never practiced (cited by McDonough, 2006). According to this definition, technological firms and even universities would therefore be patent trolls since they do not manufacture any tangible good but use their patents to collect licensing fees (Lemley, 2007) /$ see front matter 2012 Elsevier B.V. All rights reserved.

3 634 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) sustaining the emergence of technological firms specialized in knowledge production (McDonough, 2006; Magliocca, 2006; Golden, 2007; Shrestha, 2010). In this paper, we argue that most of the divergences between those two standpoints stem from the ambiguity surrounding the definition of NPE, a term which encompasses many different types of organizations: technological firms, intellectual property brokers, patent trolls and even universities. Most of the time, scholars on opposite sides of the debate tend to use terms differently, thereby confusing patent brokers with patent trolls. Our purpose is therefore threefold: first, we provide a clear definition of technological firms, (pure) patent brokers and (pure) patent trolls, thereby dismissing any ambiguity between these three actors of the knowledge economy. Second, we investigate the consequences of patent trolls and patent brokers on the R&D investments of technological and manufacturing firms. Third, we discuss policy changes which might contribute to improving the functioning of markets for technology. In the next section, we define patent trolls, patent brokers and technological firms (Section 2). In the third section, we propose a model to assess the implications of (pure) patent trolls and (pure) patent brokers on R&D investments (Section 3). We show that although patent trolling, in its pure form, is detrimental to R&D investments, patent brokers are not, since they encourage the emergence of firms specialized in knowledge production. This result has important policy implications, which we examine in the last section (Section 4). In particular, we argue that one important challenge for policy makers is to design intellectual property rights systems which reduce the possibility of trolling but which are not detrimental to the emergence of technological firms. 2. The actors in markets for technology Markets for technology have entailed the emergence of at least three new players in the knowledge economy: technological firms, intellectual property brokers and patent trolls. At first glance, those actors exhibit many similarities. They are all NPEs, they all rely strongly on intellectual property rights and they are all uninterested in cross-licensing their patents (they are looking for royalties). This explains why they are often confused by manufacturing companies and external observers (Lemley, 2007). The original definition of a patent troll, given by Detkin (see footnote 1), contributed to this confusion. Yet, beyond their apparent likeness, trolls, brokers and technological firms differ largely along several dimensions. Technological firms. Technological firms (often called fabless firms) are firms that specialize in upstream R&D activities. They produce knowledge and new technologies that they then sell to manufacturing firms (located further down the value chain) which use them to develop, produce and sell tangible goods. Technological firms are therefore clearly NPEs since they do not manufacture the good described in their patents. The link between technological firms and manufacturing firms is operated via markets for technology (Rivette and Kline, 2000; Arora et al., 2001; Arora and Merges, 2004; Gambardella et al., 2007; Arora and Gambardella, 2010), which ensure the flow of knowledge from the former to the latter. Technological firms need strong patents in order to secure their investments in R&D. Without patents, these firms would find it difficult to sell their knowledge to manufacturing firms, the latter having to some extent the possibility to free ride and to obtain the knowledge for free. In other words, the patent system contributes to solving Arrow s paradox (1962), thus fostering the emergence of markets for technology and encouraging the rise of firms specialized in knowledge production (Anand and Khanna, 2000; Arora and Merges, 2004). Patent brokers. Patent brokers are complementary to technological firms. They are companies specializing in technology transfer and, more specifically, in intellectual property transfers. In a sense, patent brokers mediate between technological firms and manufacturing companies on markets for technology (Benassi and Di Minin, 2009). Concretely, patent brokers can take two forms: They can buy patents and other intellectual assets from technological firms and then sell them to other firms (mostly manufacturing firms). Or, they can negotiate the transfer of the patent without actually buying it. However, in both cases, patent brokers do not invest in R&D themselves. They are composed of experts (in law, finance, management, etc.) who specialize in patent financial evaluation, patent marketing via specialized web platforms, patent transfer and finally, patent litigation. Therefore, they are clearly NPEs but of a different kind than technological firms. For McDonough (2006), the emergence of patent brokers (or patent dealers as he calls them) is a natural evolution of markets for technology and an indication of their maturity. 2,3 Although patent brokers can develop heterogeneous strategies, 4 their primary role is to reduce transaction costs and to facilitate the transfer of intellectual assets from technological to manufacturing firms. 5 In particular, patent brokers help reduce two important failures inherent to markets for technology: incomplete information and externalities. 2 Yanagisawa and Guellec (2009) provide a detailed inventory of the patent brokers operating in Europe. 3 The activity of intellectual property brokers relies massively on the technology of information and communication and, more precisely, on Internet marketplaces, such as patent online auctions (Dushnitsky and Klueter, 2010). Yet, Lichthenthaler and Ernst (2008) have recently questioned the efficiency of these pure online exchanges as they are completely disconnected from other transfer channels, such as consultancy, which involve face-to-face meetings. 4 Benassi and Di Minin (2009) distinguish between consultants (who provide general information and insights during transactions), shields (who keep the transaction anonymous), technology evaluators and promoters (who provide information about the value of the technology), patent enforcers (who protect against opportunistic behaviors), and patent aggregators (who assemble complementary patents in a single portfolio in order to increase its value). 5 Transaction cost economics predicts that if transaction costs are too high, firms may as well merge in order to avoid these costs. However, by reducing transaction costs, brokers reduce the interest of merging. Also, hybrid organizations such as patent pools and clearinghouses (Merges, 2001) enable firms involved in licensing activities to preserve the flexibility of the market mechanisms without the associated risk.

4 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) First, in a context where information about inventions is far from perfect, where uncertainty about the value of inventions and intellectual property rights is high and where knowledge transactions are complex, they help identify promising technologies (they connect buyers and sellers), define their economic value, and transfer them to the firms which want them the most. In other words, they contribute to improving the information on markets for technology, thus reducing the lemon problem raised by Akerlof (1970). Second, patent brokers also help protect technological firms from attempts of free riding by manufacturing firms. Indeed, by specializing in technology transfer, they remove the burden of intellectual property litigation from R&D companies, which are hardly specialists in that area. Without patent brokers, technological firms, inexperienced when it comes to intellectual property, may hence fear free riding behaviors from manufacturing firms, thus lowering their incentives to invest in R&D. In this situation, the role of patent brokers is to provide a credible threat of litigation in case manufacturing firms try to free ride, thus reducing the externality problems and Arrow s paradox (1962) in markets for technology (McDonough, 2006). Lamoreaux and Sokoloff (1999) extensively documented the role of patent intermediaries in helping to develop a market for inventions in the nineteenth century in the United States, thus allowing individual inventors to earn returns from their inventions. It is interesting to note that, when the technology base is complex i.e. multi-component, as in the field of electronics (Grindley and Teece, 1997) organizations such as clearinghouses, pools and collectives can also be considered to act as patent brokers (Merges, 2001; Aoki and Schiff, 2008). These specific institutional arrangements are indeed developed in order to promote downstream use of innovations by economizing on search and transaction costs in licensing, as well as potentially mitigating the conditions that lead to the tragedy of the anti-commons and other coordination problems in multilateral licensing (Aoki and Schiff, 2008, p. 189). In a sense, they make navigating the patent thicket possible (Shapiro, 2001). In practice, one or a group of patentees can manage the patents of their partners, but often an outside organization, i.e. a broker, is contracted to manage the portfolio and the licensing agreements. In conclusion, the main role of patent brokers is to help secure mutually advantageous technology deals which, without them, would not be realized. One important point may nevertheless qualify their importance: patent brokers are not knowledge brokers. They do not transfer knowledge, but merely a property right, i.e. a piece of paper. They transfer the right to use a given technology to firms which already know how to use it. In other words, they operate on markets for patents and not on markets for knowledge (Fischer and Henkel, 2009). This reduces their scope of action and their social desirability. As stated by Lemley (2007), a market for rights of exclusion has a lower value for society than a market for knowledge. Patent trolls. Although similar to patent brokers at first glance, the strategy of patent trolls is very different. Patent trolls are patent holders who try to provoke hold-up situations, i.e. who do not want to grant licenses (at least not too early), but provoke infringement. Hence, while patent brokers in their pure form try hard to grant licenses (they look for customers, advertise their technologies on the Internet, etc.), patent trolls keep their patent portfolios hidden and want to be infringed 6 (Henkel and Reitzig, 2008; Merges, 2009; Reitzig et al., 2010). In other words, trolls are not engaged in licensing activities but in litigation. The patenting strategy of trolls, therefore, constitutes a radical hijacking of the primary role of patents. Originally, patents were designed to prevent infringement, whereas trolls use them precisely in order to be infringed. Delayed infringement has a value for patent trolls because, most of the time, their patents have low value in an ex ante negotiation (before a manufacturing firm has made sunk investments to manufacture the product), however they become highly valuable in ex post negotiations, after important investments have been sunk by manufacturing firms. This hold-up strategy enables trolls to earn a disproportionately large fraction of the value of a technology (a fraction which is far higher than the intrinsic value of the component brought by the troll) (Shapiro, 2001; Farell et al., 2007; Lemley and Shapiro, 2007). The Blackberry case which opposed NTP, a US NPE firm, and Research in Motion (RIM), the firm which developed and commercialized the Blackberry provides a useful illustration of the way patent trolls operate (Magliocca, 2006). A patent troll business model contains 4 steps (Pénin, 2010). First, the troll buys cheap patents from financially distressed firms (or universities) in sectors where the probability of infringement is high. 7 For instance, sectors in which technology is complex (Kingston, 2001), such as electronics, are highly favorable to patent trolling. Second, trolls engage resources in order to identify potential infringers. Third, as soon as an infringer has been identified, trolls wait for sunk investments to be made before launching their attack. Fourth and last step, they threaten (to enforce production stoppage, for instance) and compromise in order to earn substantial compensation from the attacked manufacturing company. The amount of compensation may vary from one case to another. 8 It is interesting to note that, according to our definition, trolls are characterized by their behaviors and not by their adhoc features or status. As Merges puts it (2009, p. 1611): One can act as a troll, but it will not be true that one simply is a 6 The word patent troll is precisely expected to reflect the fact that trolls remain hidden while waiting for their victims. It is well-known that in Scandinavian mythology, trolls are ugly monsters who stay hidden in the woods, waiting for innocent victims to cross the wood, and then attack and rob them. The analogy with patent trolls is obvious. 7 Cheap patents may not mean low quality patents. Recent studies stress the good quality of trolls patent portfolios measured by the number of received citations (Fischer and Henkel, 2009). 8 It is worthwhile to note that, since patent trolls are NPEs and do not practice their patents, they cannot infringe patents held by other firms. Hence, they still diminish the defense of the manufacturing firms they attack (Grindley and Teece, 1997).

5 636 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) troll. Any patent holder who is hiding and waiting for infringement behaves like a troll. 9 Hence, technological firms and even universities can be considered trolls if they behave accordingly. Yet, in many instances, unlike trolls, universities and technological firms are actively looking for licensees and have no interest in hiding or rejecting reasonable licensing deals. To summarize, the development of markets for technology, sustained by strong intellectual property rights, has contributed to the emergence of new economic actors. In this section, we have defined three such actors and identified their differences. The next section endeavors to develop a simple model in order to (1) formalize the behaviors of pure patent brokers and pure patent trolls and (2) explore the consequences of trolls and brokers on R&D investments realized by both technological and manufacturing firms. 3. The consequences of patent trolls and patent brokers on R&D investments Let us assume that two firms, a supplier of technology (called S) and a manufacturer (called M), are playing a two-stage game. At the first step, S chooses to invest an amount x in R&D. At the second step, M observes x and chooses its own level of R&D, y. M s R&D investments consist of customization and development, which therefore complement S s R&D. The total value created by the two firms is V(x,y) = x y (where and [0,1] and + + <1. 10 Furthermore, let us assume that R&D costs are linear and similar for the two firms: C(x) = cx and C(y) = cy. M can access the knowledge produced by S via two different mechanisms: First, via a market arm s length transaction (technology licensing, Katz and Shapiro, 1985). In this case S receives a share ˇ (0 ˇ 1) of V, so that S and M s revenues are respectively ˇV(x,y) and (1 ˇ)V(x,y). Second, via knowledge spillovers. In this case, M benefits from gratis externalities equal to sx (0 s 1) and keeps for itself the total amount of the created value V(sx,y). The probability that S manages to find a buyer M and to handle a complex intellectual property transaction is p. 11 When S chooses its level of R&D investments it does not know whether it will be able to find a buyer, but only the probability p that it will. Yet, at the second stage of the game, M knows whether or not the transaction was a market one. The subgame perfect Nash equilibrium (SPNE) of this game is given by (x ; (y 1, y 2 )) where: x = ( 1 ) 1/(1 ) ( ) 1 /(1 ) (pˇ) 1 /(1 ) ((1 ˇ)) /(1 ) c 1 y 1 = [(1 ˇ)] (1 + )/(1 )(1 ) ( ) /(1 ) (pˇ) /(1 ) C 1/1 1 y (1 + )/(1 )(1 ) 2 = (1 ˇ) /(1 )(1 ) ( ) /(1 ) (pˇ) /(1 ) S /(1 ), c 1/(1 ) 1 and where x * stands for S s R&D investments in the first stage of the game and y 1 and y 2 for M s R&D investments, respectively in case of market transaction and spillovers. S and M R&D investments at the equilibrium are always increasing with the probability of securing a market transaction p. This might therefore legitimate the intervention of market intermediaries (brokers), who aim to lower transaction costs in order to improve market based technology transfer (decrease spillovers). Consequently, let us assume now that S can decide whether or not to rely on a broker in order to transfer its technology to M. When S relies on a patent broker, the probability to secure a market based transaction becomes q > p. Yet, the use of patent brokers is not free of charge. If they successfully manage to secure a market transaction, brokers receive a fraction ı of V. We assume further that the earnings of patent brokers are debited exclusively from the share that goes to technological firms, so that S earns now (ˇ ı)v. The introduction of patent brokers changes the structure of the game. Now, S, having chosen x, must decide whether to rely on a patent broker. It will choose to do so only if this increases its expected profit, which requires that q/p > ˇ/(ˇ ı). If this inequality does not hold, R&D investments of both firms are identical to those found earlier in the game without a broker (x ; (y 1, y 2 )). However, when it holds, S chooses to rely on a broker, which implies that the R&D investments of both firms at the SPNE become (x b ; (y 1b, y 2b )) where: x b = ( 1 c ) 1/(1 ) ( ) (1 )/(1 ) (q(ˇ ı)) (1 )/(1 ) ((1 ˇ)) /(1 ) 1 9 An important consequence of our definition of patent trolls is that it may be very difficult, if not impossible, to study patent trolling empirically. Collecting a database of firms assimilated to patent trolls is indeed strongly complicated by the fact that an organization may only occasionally behave like a troll. Furthermore, observing during litigation that a firm is in a situation of hold-up is not sufficient to conclude that the other firm is a troll. One must show that the firm acted deliberately in order to provoke the hold-up (Fischer and Henkel, 2009). In other words, existing databases of patent trolls are likely to also encompass firms that practice trolling only occasionally or involuntarily. 10 Firms have perfect information about the value of the technology. Many authors have explored the issue of technology licensing when the value of the technology is unknown for the seller, for the buyer or for both (Gallini and Wright, 1990; Martimort et al., 2010; Jullien et al., 2011). 11 In a sense, p can be interpreted as the strength of the patent system for technological firms such as S and s as the intensity of knowledge flows in case of spillovers. Different combinations of p and s account therefore for different technological regimes and degrees of appropriation (Winter, 1984).

6 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) y 1b = [(1 ˇ)] (1 + )/(1 )(1 ) ( ) /(1 ) (q(ˇ ı)) /(1 ) c 1/(1 ) 1 y (1 + )/(1 )(1 ) 2b = (1 ˇ) /(1 )(1 ) ( ) /(1 ) (q(ˇ ı)) /(1 ) S /(1 ), c 1/(1 ) 1 and where x b stands for S s R&D investments in the first stage, knowing that it will rely on a broker in the next step of the game, and y 1b and y 2b for M s R&D investments in case of market transaction and spillovers, respectively, and knowing that S is investing x b. It is straightforward to observe that when q/p > ˇ/(ˇ ı), x b > x, y 1b > y 1 and y 2b > y 2. In other words, as stated by our first proposition, the presence of intellectual property brokers is always R&D increasing. If q/p < ˇ/(ˇ ı) brokers increase S and M s levels of investments and, if q/p < ˇ/(ˇ ı), S will not hire a broker, which will therefore not change the total level of R&D investment. This result sheds light on McDonough s statement claiming that: The activity of patent dealers in their pure form benefits society (2006, p. 204). Proposition 1: the presence of patent brokers is either neutral (does not decrease) or increases the R&D investments of both S and M Let us now envisage the possible intervention of patent trolls. As seen in the previous section trolls operate late on markets for technology. They hide and as soon as a manufacturing firm has engaged R&D investments they claim patent infringement and ask for indemnification. The main consequence of patent trolls is therefore to introduce uncertainty with respect to M s freedom to operate. Let us assume that the probability to meet a patent troll is r. In case M encounters a patent troll, the latter claims a share ε of V and M receives therefore only a payoff equal to (1 ˇ ε)v. The new SPNE with patent trolls is given by (x T ; (y 1T, y 2T )) where: x T = ( 1 c ) 1/(1 ) ( ) (1 )/(1 ) (pˇ) (1 )/(1 ) ((1 ˇ rε)) /(1 ) 1 y 1T = [(1 ˇ rε)] (1 + )/(1 )(1 ) ( ) /(1 ) (pˇ) /(1 ) c 1/1 1 y (1 + )/(1 )(1 ) 2T = (1 ˇ rε) /(1 )(1 ) ( ) /(1 ) (pˇ) /(1 ) S /(1 ), c 1/(1 ) 1 and where x T stands for S s R&D investments in the first stage, and y 1T and y 2T for M s R&D investments in the second stage in case of market transaction and spillovers, respectively. Again, one can immediately observe that x T, y 1T and y 2T are always smaller than x *, y 1 and y 2 respectively. In other words, as stated by our second proposition, patent trolling is always R&D decreasing in the economy. This result is in line with Jaffe and Lerner (2004) and Henkel and Reitzig (2008) who explain that patent trolls strategy is based exclusively on value destruction and that patent trolls create no surplus for society (see also Merges, 2009). It perfectly illustrates what Reitzig et al. (2010) call collateral damage for R&D manufacturer. The likelihood of encountering a troll decreases the incentives of R&D manufacturers to invest in R&D which, in turn, decreases the R&D investments of technological firms. This is also in line with the research carried out by Bessen et al. (2011), whose empirical study stresses the considerable cost of patent trolls for society. Proposition 2: the presence of patent trolls decreases the R&D investments of both S and M The result emphasized in proposition 2 holds under the hypothesis that trolls do not transfer revenues to upstream technological firms and that they do not increase p. Indeed, pure patent trolls are unlikely to do so, since they are, by definition, not interested in technology transfer (Bessen et al., 2011 show that most of the welfare loss induced by trolls is not transferred toward upstream inventors). Yet, since p and r both depend to some extent on the characteristics of the patent system, they are likely to be positively correlated. For instance, if the patent system is strongly in favor of patent holders and makes it difficult for manufacturing firms to free-ride on non-practicing entities, it increases both p and r. Conversely, if the patent system makes it difficult for non-practicing firms to enforce their patents, it decreases both p and r. Hence, let us now assume that both p and r are positive functions of, which might reflect the strength of the patent system for non practicing firms. In this case it is easy to see that at the SPNE with patent trolls, the R&D investments of both S and M are not monotonic functions of (it follows immediately from the expression of x T, y 1T and y 2T ). This implies that, as stated by proposition 3, paradoxically, there might exist a non-zero optimal level of trolling that must be tolerated in order to benefit from markets for technology and from the rise of specialized technological firms.

7 638 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) Proposition 3: the optimal level of trolling might not be zero when r and p are both positive functions of. The intuition underlying this proposition is the following: When varies, it triggers two opposing effects. On the one hand, an increase in favors market transactions and decreases spillovers, which is R&D augmenting; On the other hand, it also favors trolling, which is R&D decreasing. The optimal level of must thus balance these two effects. 4. Policy implications Pure trolling activities, as defined in the previous section, penalize manufacturing companies and are welfare decreasing. Henkel and Reitzig (2008) thus suggest that, in order to adapt to trolls, manufacturing firms can and should react, including by modifying their patenting strategy. For instance, manufacturing firms can largely reduce the possibility of trolling by developing collaborations and information exchange early in the innovation process in order to identify trolls. They can also voluntarily limit the number of patents they apply for in order to facilitate the work of examiners and to improve patent information. Finally, they can develop modular technologies that protect them from trolling, by making it easy to change the piece of technology being attacked, without having to rethink their product entirely. However, Henkel and Reitzig also conclude that these private solutions to trolling are likely to have only a limited effect in shutting down trolling activities, without an accompanying change in patent law. This result calls therefore for important evolutions in the patent system in order to limit the possibility of trolling (especially with regard to the potentially enormous costs for society put forward by Bessen et al., 2011). However, one major difficulty here is that, although trolling in its pure form is likely to entail important social costs, the activities of technological firms and pure knowledge brokers have a positive impact on R&D investments (and by extension on social welfare). One central challenge for policy makers is thus to design patent systems which limit the activity of trolling but which are not detrimental to technological firms and brokers, the latter being a critical part of a knowledge-based economy (Merges, 2009). The rise of patent trolling is linked to three elements (Reitzig et al., 2007): First, the number and the quality of delivered patents; second, the structure of patent litigations; and third, the facility for trolls to operate in order to place other firms in a situation of hold-up. First, patent trolls raise the issue of the number of patents delivered and, most of all, their quality. In an ideal world, only inventions that are sufficiently new and inventive should be granted a patent. Yet, in the real economy of the past 20 years (mostly in the US), thousands of inventions that were not particularly new or inventive have been granted patents (Jaffe and Lerner, 2004). This proliferation of low quality patents and the complex way in which they are written feed patent trolling, among others because it contributes to the blurring of information for manufacturing companies. In particular, it makes the freedom to operate analysis more complex because, although patents are public, it is increasingly difficult to find relevant and exhaustive information in the patent literature (Merges, 2009). This is especially true of complex technologies, like electronics, in which hundreds of overlapping patents may exist for a given piece of technology. Also, the proliferation of low quality patents gives trolls access to cheap raw materials (existing patents). Indeed, although recent studies (Fischer and Henkel, 2009) show that trolls tend to rely on rather good quality patents, trolling strategies are compatible with low quality patents. In our model, the effect of the quality of patents held by trolls can be illustrated by adding an uncertainty with respect to the outcome of the litigation process, an uncertainty which varies according to the quality of the patent. When a manufacturing firm encounters a troll, it would therefore have a lower probability of winning the trial if the patent is of good quality. In this case, it is easy to see that it is usually more profitable for trolls to rely on high quality patents, however it may also pay to use low quality patents provided that they are sufficiently cheap to obtain. Hence, an initial political measure to limit patent trolling might be to increase the standard required to obtain patents, or at least to respect existing ones. This might strongly increase the quality of patents delivered, improve the visibility of manufacturing firms and decrease uncertainty. Consequently, this change might make the job of trolls more complex as, if the number of delivered patents decreases, it will be more and more difficult for trolls to hide the patents with which they operate. This is reflected in our model where aggregated R&D investments are negatively linked to r. Thus, under the assumption that r depends on the number of delivered patents (the higher the number of patents delivered, the greater the likelihood of encountering a troll), reducing this number directly increases the R&D investments of the economy. 12 Another measure that may contribute to increasing patent quality is to increase the cost of obtaining a patent (Harhoff, 2009). Standard economic theory predicts that an increase in fees should increase the quality of patents applied for by firms, due to a self-selection effect (in particular, it may deter the strategic use of patent offices by firms). Many countries already apply increasing renewal fees in order to encourage firms to maintain only highly valuable patents. It might hence be desirable to continue to increase renewal fees in order to reinforce the self-selection effect. Indeed, De Rassenfosse and von Pottelsbergue de la Poterie (forthcoming) provided empirical evidence on the link between patent fees and firms patenting behaviors by surveying the fees at most patent offices in developed countries. They found that fees (pre-grant and post-grant) do affect the patenting behaviors of firms. However, the demand for patents is also found to be relatively inelastic, which 12 However, making the delivery of patents more costly and/or more difficult might also penalize pure technological firms (it might decrease p). As a consequence, as stated by proposition 3, an optimal patenting policy must be the outcome of a delicate balance between limiting trolling and encouraging technological firms.

8 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) means that an increase in fees must be significant in order to decrease the number of applications. Yet, it is also possible that the demand elasticity is higher for low value patents than for high value patents. To summarize, although it is still very exploratory, this discussion surrounding the role of patent fees in order to increase the quality of delivered patents has the merit of providing an argument against most patent practitioners and lawyers who fiercely advocate decreasing patenting costs. Second, the structure of patent litigations in most countries makes the activity of patent trolls very attractive. Policies that reduce the interest of firms in trolling can play on three parameters of patent litigation: the probability of winning for patent holders, the rules of indemnification and the possibility for patent holders to stop the activity of infringers. Nowadays (and this is especially true in the USA with the implementation of the Court of Appeal of the Federal Circuit (CAFC)), the firm which is accused of infringement has a very high probability of losing the trial (Jaffe and Lerner, 2004). This obviously encourages patent trolling by inducing manufacturing firms to compromise, independently of the quality of the patents held by trolls. Indemnification rules also often make the practice of trolling very attractive. These rules vary from one country to another (Reitzig et al., 2007), but they often imply that it is more profitable for a firm to be infringed than to bargain a licensing agreement ex ante. 13 During a patent litigation, the firm whose patent is infringed can ask the judge to shut down the activity of the infringing firm (this can be done after the verdict is announced, or even before by asking for a preliminary injunction, which may be granted if the judge considers that the survival of the infringed firm is threatened). This is a highly problematic issue for manufacturing firms that have engaged in costly sunk investments in order to manufacture their products. The possibility of requiring them to stop their activity obviously increases their willingness to compromise. A second policy measure to limit trolling might hence play on those three dimensions of litigations. First, in the US, a rebalancing of the ruling of the courts more in favor of defenders might help to make trolling less profitable. Yet, this is obviously a very delicate issue for policy makers as the outcome of a trial rests exclusively in the hands of the judge. Second, it might be possible for judges to improve indemnification rules in order to make trolling less interesting. In our model, for instance, aggregated R&D investments are negatively linked to ε, which depends directly on the indemnification structure of patent litigations. Thus, changes to the indemnification rules resulting in a decrease in ε would in turn increase the total R&D investments of the economy. Again, this is also likely to affect ˇ, the share that goes to technological firms. Third, a correction of the injunction rule, making it more difficult for fabless firms (including trolls and brokers) to shut down manufacturing activities, might also seriously reduce the attractiveness of trolling. Such a correction can also make sense because, in the case of litigation between a troll (or another fabless firm) and a manufacturing company, the injunction rule might not be justified. Indeed, since the troll does not commercialize any tangible products, the activity of manufacturing companies does not threaten them and there is therefore no reason to shut down the activity of the infringer. The third characteristic that favors the trolls business model is based on the facility for firms to provoke hold-ups. A manufacturing firm can be trapped in a hold-up situation for three reasons (Reitzig et al., 2007): it can be the victim of a submarine patent, which cannot be detected before its attribution (this is specific to the US system). It can fail to identify a relevant patent during its freedom to operate analysis, the multiplication of patents making such analyses more and more complex. Finally, it may have identified the relevant patents but may have considered their holders to be not dangerous (a partner for instance). Yet, the owner of a patent can change (buyout, bankruptcy) and the patent ownership may pass on to a troll. A third set of policies might thus include measures to reduce the ease of hold-up. First, it is admittedly possible (and desirable) to improve information on patents and to lower the number of low-value patents (by increasing the requirement level or the costs, as discussed above). It is also possible to implement a system of compulsory licensing which may prevent fabless firms from refusing to grant a license to a manufacturing company. Yet, those two measures, although they might improve the situation slightly, might not really prevent firms from strategically maneuvering in order to provoke hold-ups and to earn a bigger share of the total value. Consequently, another measure that aims at radically decreasing the possibility of provoking hold-ups might be to make it possible for manufacturing firms to publicly signal their manufacturing projects during a given period of time, at the end of which, if no patent holder has presented itself, they achieve complete freedom to operate. A website (largely advertised) could be specifically dedicated to this purpose. On this website, manufacturing firms could publicly announce the launch of a new innovation project before they engage sunk costs. During a certain period of time (for instance 3 months), patent holders would therefore have the opportunity to approach ex ante the manufacturing company. Genuine patent brokers would certainly not hesitate to do so. At the end of the announcement period, it would no longer be possible for holders 13 For instance, many courts assess the amount of royalties that infringing firms must pay to patentees on the basis of the average rate of royalties of the sector. This rule introduces obvious self-selection problems, since firms who expect more than the average (because they have a valuable technology) have an incentive to bargain ex ante while firms who expect less have an incentive to wait for litigation (Reitzig et al., 2007). Another feature that favors trolling is that in the USA, the loser of a trial does not have to reimburse the money advanced by the winner, as is the case in France for instance. This rule, by decreasing the cost associated with trolling also favors its development.

9 640 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) of patents applied for before this period to claim royalties from the manufacturing firms, therefore ensuring that they have freedom to operate. 14,15 In theory, this solution would make it feasible to discriminate between pure trolls and pure brokers (since only the latter would have an incentive to approach manufacturing firms ex ante) and would therefore significantly decrease the attractiveness of trolling behaviors. However, in practice, it is likely that many manufacturing firms would be reluctant to signal their projects and to reveal their details to competitors, especially given the strong evidence that, for many firms, lead time advance is more important than patent protection (Cohen et al., 2000). 5. Conclusion This paper has explored the behaviors of pure patent trolls and pure patent brokers and their consequences on global R&D investments. We have showed that while patent brokers in their pure form always increase total R&D investments, pure patent trolls do not. This work therefore clearly stresses that one must distinguish between different kinds of NPEs. Not all are trolls. Furthermore, it teaches us that the development of markets for technology may not always be welfare increasing, given that it may also fosters the proliferation of trolling behaviors (Reitzig et al., 2010). We have also discussed the policy measures that can be implemented in order to limit patent trolling. Some of them have already been adopted or will soon be (Magliocca, 2006; Golden, 2007; Lemley and Shapiro, 2007). For instance, in the USA, the patent reform act passed in 2007 defines the payment of royalties on a more reasonable basis. Still in the USA, it is currently almost impossible for fabless firms to require a preliminary injunction against manufacturing firms (this follows the recent ruling of the Supreme Court in the case MercExchange vs. Ebay Inc, McDonough, 2006). The quality of attributed patents is also significantly improving. Similarly, in Europe, directive 2004/48/EC attempts to harmonize litigation practices across European countries. However, it is important to keep in mind that, although these changes may contribute to limiting patent trolling, the objective of policy makers may not be to eliminate patent trolls completely as such an action might also endanger technological firms. Indeed, patent trolls, patent brokers and technological firms all represent different facets of the same phenomenon: The rise of markets for technology. In this sense, they all rely on strong patent systems. Hence, changes that limit trolling might also be detrimental to brokers and technological firms. As emphasized by McDonough (2006), it seems inevitable that the fight against patent trolls which aims to weaken patents held by NPEs in one form or another would also affect technological firms, thus harming the knowledge-based economy at its core. In the end, it is likely that a balance will have to be found. As usual, with respect to patent systems, the optimum is a matter of compromise and a certain level of trolling may in fact be socially desirable. To conclude on an optimistic note, it is important to recall that our analysis of patent trolls is entirely static. However, in an evolutionary perspective, the episode of intense trolling experienced in the last decade in the USA may, paradoxically, have been welfare increasing for at least two reasons: First, because it forced changes to the patent law, adopting a more balanced policy (especially in the US). Second, because it forced manufacturing firms to improve their technology watch practices and their freedom to operate analysis. It is therefore likely that, nowadays, manufacturing firms are better prepared to react to the aggressive behaviors of some NPEs. References Akerlof, G., The market for lemons: quality uncertainty and the market mechanism. Quarterly Journal of Economics 84, Anand, B.N., Khanna, T., The structure of licensing contracts. Journal of Industrial Economics XL (VIII), Aoki, R., Schiff, A., Promoting access to intellectual property: patent pools, copyright collectives and clearinghouses. R&D Management 38 (2), Arora, A., Gambardella, A., Ideas for rent: an overview of markets for technology. Industrial and Corporate Change 19, Arora, A., Merges, R., Specialized supply firms, property rights and firm boundaries. Industrial and Corporate Change 13, Arora, A., Fosfuri, A., Gambardella, A., Markets for Technology: The Economics of Innovation and Corporate Strategy. MIT Press, Cambridge, MA. Arrow, K.J., Economic Welfare and the Allocation of Resources for Invention. In: The Rate and Direction of Inventive Activity: Economic and Social Factors. Princeton University Press, pp Benassi, M., Di Minin, A., Playing in between: patent brokers in markets for, technology. R&D Management 39 (1), Bessen, J., Ford, J., Meurer, M., The private and social costs of patent trolls. Boston University School of Law Working Paper, pp Chesbrough, H., Open Innovation: The New Imperative for Creating and Profiting from Technology. Harvard Business School Press, Boston. Cohen, W.M., Nelson, R.R., Walsh, J., Protecting their intellectual assets: appropriability conditions and why US manufacturing firms patent (or not). NBER. Working Paper Interestingly, this solution already exists with respect to new real-estate projects. In this case also, property developers often face hold-up situations provoked by neighbors who can block the project for a while by arguing diverse nuisances. To counter those trolling behaviors, in many countries real-estate property developers must publicly announce and describe (at the city hall, for instance, or by displaying a sign at the place where the project will occur) their project before starting it. The purpose of this is to allow neighbors to react to the project. Once a given period has elapsed, it is no longer possible for neighbors to block the project. 15 This system exhibits similarities with the system of pre and post opposition provided by the European patent office. In both cases, other firms are asked to provide information about existing patents that the examiner may have missed. However, the system proposed here differs from a simple opposition in the sense that if no more information has been provided after the end of the period, the manufacturing firm is warranted entire freedom to operate for its project, while the opposition system concerns only the deliverance of a patent. In other words, this system gives firms a right to use while the opposition system provides only a right to exclude.

10 J. Pénin / Journal of Economic Behavior & Organization 84 (2012) De Rassenfosse, G., von Pottelsbergue de la Poterie, B. The role of fees in patent systems: theory and evidence. Journal of Economic Survey, forthcoming. Dushnitsky, G., Klueter, T., Is there an ebay for idea? Insights from online knowledge marketplaces. European Management Review 8 (1), Farell, J., Hayes, J., Shapiro, C., Sullivan, T., Standard setting, patents and hold-up. Antitrust Law Journal 74, Fischer, T., Henkel, Y., Patent Trolls on markets for technology an empirical. Analysis of Trolls patent acquisitions. In: 4th EPIP conference, Bologne, September Gallini, N., Wright, B., Technology transfer under asymmetric information. The RAND Journal of Economics 21, Gambardella, A., Giuri, P., Luzzi, A., The Market for patents in Europe. Research Policy 36 (8), Golden, J.M., Patent trolls and patent remedies. Texas Law Review 85, Grindley, P., Teece, D., Managing intellectual capital: licensing and cross-licensing in semiconductors and electronics. California Management Review 39, Harhoff, D., Patent value and patent rating. In: 4th EPIP Conference, Bologne, September Henkel, J., Reitzig, M., Patent sharks. Harvard Business Review. Jaffe, A., Lerner, J., Innovation and its discontents: how our broken patent system is endangering innovation and progress and what to do about it. Princeton University Press. Jullien, B., Pouyet, J., Sand-Zantman, W., Make or sell? Innovation Licensing and market structure. In: AFSE Thematic Meeting on Law and Economics, Besanç on, France, June 9 10, Katz, M., Shapiro, C., On the licensing of innovations. The RAND Journal of Economics 16 (4), Kingston, W., Innovation needs patent reform. Research Policy 30, Lamoreaux, N., Sokoloff, K., Inventors, firms and market for technology in the late nineteenth and early twentieth century. In: Learning by doing in Markets, Firms and Countries. NBER Book, University of Chicago Press, pp Lemley, M., Are universities patent trolls? Fordham Intellectual Property. Media & Entertainment Law Journal 18, Lemley, M.A., Shapiro, C., Patent hold-up and royalty stacking. Texas Law Review 85, Lichthenthaler, U., Ernst, H., Innovation intermediaries: why Internet marketplaces for technology have not yet met the expectations? R&D Management 17 (1), Magliocca, G., Blackberries and Barnyards: Patent Trolls and the Perils of Innovation. Notre Dame Law Review 82, Martimort, D., Poudou, J.C., Sand-Zantman, W., Contracting for an innovation under bilateral asymmetric information. Journal of Industrial Economics 48, McDonough, J.F., The myth of the patent troll: an alternative view of the function of patent dealers in an idea economy. Emory Law Journal 56, Merges, R.P., The trouble with trolls: innovation, rent seeking, and patent law reforms. Berkeley Technology Law Journal 24 (4), Merges, R.P., Institutions for intellectual property transactions: the case of patent pools. In: Dreyfuss, R., Zimmerman, D.L., First, D. (Eds.), Expanding the Boundaries of Intellectual Property. Oxford University Press, Oxford, pp Pénin, J., Le problème des «patent trolls»: comment limiter la spéculation sur la propriété intellectuelle dans une économie fondée sur les connaissances? Innovation: cahiers d économie de l innovation 32 (2), Reitzig, M., Henkel, J., Heath, C.H., On sharks, trolls, and their pa$tent prey unrealistic damage awards and firms strategies of being infringed. Research Policy 36, Reitzig, M., Henkel, J., Schneider, F., Collateral damage for R&D manufacturers: how patent sharks operate in markets for technology. Industrial and Corporate Change 19, Rivette, K.G., Kline, D., Rembrandts in the attic: unlocking the hidden value of patents. Harvard Business School Press. Shapiro, C., Navigating the patent thicket: cross licenses, patent pools, and standard Setting. In: Jaffe, A., Lerner, J., Stern, N. (Eds.), Innovation Policy and the Economy, vol. 1. MIT Press, pp Shrestha, S.K., Trolls or market-makers? An empirical analysis of non practicing entities. Columbia Law Review 110, Winter, S.G., Schumpeterian competition in alternative technological regimes. Journal of economic behavior and organization 5 (3 4), Yanagisawa, T., Guellec, D., The emerging patent marketplace. OECD Science, Technology and Industry Working Papers, 2009/09.