3.2.1. A class of heterogeneous actors

From an analytic perspective, one of the main difficulties related to evaluating this phenomenon is that the term “patent troll” is not an established legal term. Over time, the term “patent troll” was substituted for or was used interchangeably with the terms “non-practicing entities” (NPEs) and “patent assertion entities” (PAEs) as well as “patent monetization entities” and “patent aggregators”. This multiplication of designations makes it difficult to understand the phenomenon. The common point of these entities, physical or moral, is that they do not have the intention or the capacity to produce, commercialize or use the patented invention at issue, but rather than renounce their patrimonial rights, they seek to leverage them strategically for active licensing and/or make claims for their rights before the courts. In reality, these terms conceal much more complex strategies. As indicated in an annual report by the Federal Trade Commission [FTC 16] dedicated to the topic, a significant blind spot in all previous studies concerning trolls is related to the fact that most of these studies “have focused on publicly observable litigation behavior and relied on publicly available litigation data”⁸. Relying exclusively on public data does not allow for accessing key information, such as the confidential terms of licensing agreements and their amount, though these are essential for understanding the business model of PAEs. In what follows (similarly to [FTC 16]), we will use the generic term of patent assertion entities (PAEs). This term has the advantage of a certain neutrality and does not immediately condemn all of the actors in this category as trolls. It simply refers to the fact that these actors concentrate on asserting patents acquired from a third party rather than filed by themselves. Two features characterize PAEs.

First, a PAE does not manufacture, distribute, or sell products. Patents represent the main assets of this type of entity and its revenue comes from receiving royalties in case of license or obtaining damages in case of litigation. The fact that they do not take part in manufacturing activities means that they are not concerned by counterclaims for patent nullity. This is all the more true because these entities are not very affected by acts of retaliation on the part of “target” companies, because unlike many operational companies, they do not find themselves in a cooperation/confrontation logic, also known as coopetition, to produce goods⁹. Inversely, they can earn a reputation as aggressive in terms of litigation when they deploy a “scorched earth” strategy.

Secondly, PAEs do not have any inventive activity and do not invest, or invest very little, in R&D¹⁰. Their primary activity consists mainly of an intensive patent-purchasing policy, often at a discount from bankrupt companies (Lemley et al. [LEM 16]) and monetizing them. The choice of the “cease and desist” timing is not insignificant. When acquiring and then asserting patents, PAEs target individuals and companies when they are most vulnerable, that is, when they have created a potentially infringing invention (even if they obtained a different patent than the one held by the PAE) in their product, and when this product has already been commercialized. The activity of PAEs gives rise to what are often called ex post patent transactions because all licensing or out-of-court settlements take place after the product in question has been developed or commercialized. As indicated by the FTC, this contrasts with ex ante patent transactions in which the technology and the related rights pass from an inventor to a manufacturer before the product is developed or commercialized [FTC 16].

Confronted with an increase in complaints against PAEs for alleged anti-competitive behavior, the Federal Trade Commission (FTC) published a report in which it analyzed the behavior of PAEs in 2016¹¹. For clarification purposes, the FTC established a distinction between two types of PAEs [FTC 16]. The first category is represented by “portfolio PAEs” whose main activity is to acquire industrial property titles in order to build patent portfolios and generate revenue in the form of licenses or by attesting, if applicable, their rights against alleged infringers. Portfolio PAEs are unique in that they finance their patent acquisitions through capital collected from investors, including patenting firms themselves. Generally, they hold portfolios containing hundreds or even thousands of patents. Their activity consists of negotiating licenses whose value is generally greater than several million dollars¹² (65% of the licenses issued by litigation PAEs have generated revenues of more than one million dollar per licensee and 10% revenues of more than 50 million dollars). The second category is represented by entities that the FTC calls “litigation PAEs”. They acquire patent portfolios that are often composed of less than 10 patents each, and hold them in newly created affiliated entities that have little or no working capital¹³. Their strategy is more focused on legal threats than on technology transfers, in the sense that they regularly initiate claims against potential licensees.

3.2.2. The business model of litigation PAEs

Traditionally, by negotiating and granting licenses on invention patents, PAEs facilitated the creation of technology markets, as described in Chapter 1 (Hosie [HOS 08], McDonough [MCD 06], Morgan [MOR 08], Myhrvold [MYH 10]). Arora et al. [ARO 04] discussed the pivotal role of PAEs in redefining the petrochemical industry. Recent criticism more specifically concerns PAEs who attempt to exploit their negotiation power to attest their rights aggressively, even to extort royalties (Merges [MER 09]). Boldrin and Levine [BOL 13] as well as Magliocca [MAG 07] denounces the behavior of some of these actors whose business model, according to them, is tantamount to blackmail¹⁴.

In reality, the business model of litigation PAEs, which the FTC seems to equate to the category of patent trolls (without ever naming them as such), seems to follow a well-defined sequence, namely purchasing and securing of patents, often large, in domains affected by a patent thicket and therefore conducive to overlapping patents¹⁵, potentially concealing these patents until the market for a certain technology develops and the patents held can play a role in them, and finally, obtaining “compensation” for infringement from firms developing the technology (Henkel and Reitzig [HEN 08]). The hold-up is due to the fact that, in this sequence, the PAE waits to assert its patent or patents until the supposed infringers have sunken recoverable costs for the industrialization and commercialization of the invention and are therefore in an unfavorable position to defend their interests. Consequently, the economically weak patent holders may prefer to accept a disadvantageous agreement instead of running the risk of not surviving a trial. This can be explained by the substantial legal fees in some courts, especially the principle of treble damages¹⁶ in the United States and China, as well as the slowness of their legal procedures. The combination of these two elements, namely high costs and long delays, are effective tools to exert pressure on supposedly infringing companies and get them to settle out of court. The FTC study [FTC 16] shows that the average royalties received by litigation PAEs amounts to less than $300,000. According to some estimates, this amount represents the lower limit of legal fees for the defense in a patent infringement action, which could be seen as proof that litigation PAEs favor settling out of court. Given the relatively low amounts, legal actions initiated in this way must, according to the FTC be interpreted as intending to cause nuisance¹⁷ [FTC 16, p. 10].

In practice, litigation PAEs often use pressuring and leverage methods. Very often, they target several companies for a single allegedly infringing patent, and capitalize on the fact that the companies that are first to straighten out their situation will obtain considerable savings compared to other competitors targeted by the legal action.

This incentive strategy is inspired by the prisoner’s dilemma. Concretely, it means that the best strategy for all of the alleged infringers would be to cooperate and agree not to pay the amounts requested. This is especially true since the literature shows that patent holders only win one-quarter of cases (Allison et al. [ALL 14], Janicke and Ren [JAN 06])¹⁸ and only 8% of cases concerning PAEs (Allison et al. [ALL 11]). However, the substantial legal fees suffered by each allegedly infringing company, compared to the lower royalties required by settling out of court, pushes each company to prefer to settle matters out of court. In other words, as indicated by Ewing and Feldman [EWI 12], it is economically and individually logical for manufacturing companies to surrender to a PAE’s demands rather than fight them, even if the patents concerned have a low chance of being recognized as being infringed by a judge. This non-cooperative behavior between the accused firms is reinforced by risk aversion. Feldman [FEL 12] explains that, given the uncertainty surrounding the scope of patents, the absence of quick, reliable and inexpensive means to resolve this uncertainty, and the possibility of having to provide considerable damages or even suffer an injunction against an entire range of products, a company that considers itself within its rights may nevertheless capitulate to the patent holder’s demands and pay the requested amount¹⁹.

This is true even if the patent is low-quality, or not connected with the product, for the sole purpose of avoiding the costs and risks of litigation. In other words, the saying that “a bad settlement is always better than a good trial”, is especially accurate in this case. All the more so because, in certain courts, like the United States, legal fees are often greater than the “compensatory” indemnities requested and received in case of victory for the defending firm²⁰. The potential pressure from clients and the negative impact on the brand image of the alleged infringer are other factors that can explain the capitulation of manufacturing firms to litigation PAEs. These elements come on top of the actual legal fees and are strong incentives to be the first to obtain favorable licensing conditions. What follows can be described as a “bandwagon effect”, which means that the other alleged infringers will also make a licensing and/or financial agreement. They will be even more incentivized because the first contractor is influential and lends strong credibility to the validity of the patent put forward by the PAE. Litigation PAEs rely on the fact that there is an uncertainty about the patent scope and its quality. Some authors (Lemley and Shapiro [LEM 05]) go so far as to say that patents are “probabilistic property rights” analogous to lottery tickets²¹.

In summary, the business model of a “troll” consists, as noted by Reitzig et al. [REI 07], of being infringed. “Trolling” strategies are therefore misappropriations of the original role of the invention patent. However, from a legal perspective, the injunction strategies of patent trolls are not only profitable (Reitzig et al. [REI 07]) but also legal²² because the patent law offers all title holders the possibility of enforcing the monopoly of exploitation that is conferred to them when they obtain a patent.

3.2.3. What is the scale of this phenomenon?

From an empirical point of view, some studies have focused on the PAE phenomenon. It is difficult to draw general conclusions given that these works neither adopt the same methodology, nor use the same data or the same period for analysis. In addition, the phenomenon is difficult to grasp given that most of the cases do not go to trial²³. However, all of the studies seem to confirm the idea that although this is a growing phenomenon in the United States, it is rather rare in Europe [HEL 12].

A first series of works conducted by Jeruss et al. [JER 12] examined all of the patent disputes that took place in front of American courts between 2007 and 2012, namely 13,000 cases and nearly 30,000 claimed patents. They showed that in 2012, legal action initiated by monetization entities represented the majority of patent disputes in the United States. More specifically, these entities were responsible for 58.7% of legal proceedings in 2012. This is a substantial increase compared to 2007, when they only amounted to 24.6%. Out of the 10 entities who filed the greatest number of litigations about patents over the years in question, all were patent monetization entities.

A second series of works by Chien [CHI 12a] showed that PAEs represented 61% of legal proceedings concerning patents in 2012 compared to 45% in 2011 and 29% in 2010. Although infringement actions only represented 20% of cases in 2006, this number rose to 57% in 2012. Love [LOV 13] also documented the legal behaviors of PAEs. His 2012 study analyzed a sample of 472 patents selected from recently expired patents that were the subject of legal action. Love’s data showed that PAEs were responsible for twice as many disputes as practicing entities. They asserted their patents at a later age than other rights holders, which reinforces the idea that they wait until a “prey” presents itself. According to Love [LOV 13], PAEs pursue four times more alleged infringers per patent than other rights holders. This can be explained by the fact that before the adoption of the America Invents Act in the United States, some courts allowed patent holders to include different defendants in a single request, on the grounds that the decision about the scope of the patent would provide a sufficient basis for the consolidation, even if the acts of infringement were independent²⁴. Regardless of the year referenced, the literature also shows the existence of forum shopping, that is, parties choosing to bring their case before a court of first instance (district court) where they know that the jury and the rules of civil procedure, especially in terms of timeline, will be more favorable for them. On this issue, the Eastern District of Texas court is the preferred choice of patent holders to initiate infringement actions. This is true for both PAEs and practicing entities. The Eastern District of Texas court has a reputation for having juries and rules of procedure that are friendly to patent holders. The same judges have responded to new junction rules by allowing proceedings initiated by a single plaintiff against several defendants to be consolidated, although the trials remain distinct²⁵.

For their part, Jeruss et al. [JER 12] retraced the history of the transfers of patents included in their databases and showed that there is a strong market for the transfer of patents prior to litigation, in so far as about 52% of patents were transferred. This is consistent with the development of a technologies market analyzed in Chapter 1. One of the interesting points emphasized by these authors is the existence of a market for post-expiration transfers. The parties transfer the patents, after the expiration of these patents, to other parties, who then initiate legal proceedings for infringement on the same patents. This is possible because American legislation allows for the retrospective collection of damages for a maximum of 6 years. More fundamentally, this suggests the presence of what could be described as a distinct market offering a residual value for expired patents.

Finally, Chien [CHI 12b] surveyed 223 tech start-ups and noticed that 79 of them were threatened with legal action by PAEs in order to get to them reach a licensing agreement. This research also shows that most of the defendants were small-scale: 55% of the start-ups surveyed earned less than 10 million dollars in revenue and 66% earned less than 100 million dollars. However, Wagner [WAG 12] suggests a recent trend of targeting larger companies. In a complementary study, Chien [CHI 13] surveyed nearly 300 venture capitalists and start-ups who benefited from venture capital, and showed that patent assertions by PAEs not only increased frictions related to technological transactions, but also reduced the value of the start-ups they pursued and gave rise to substantial indemnities. This research shows that 75% of the venture capitalists surveyed and 20% of the start-ups with industrial property experience were threatened with litigation by a PAE. In total, according to Chien [CHI 13], nearly 90% of all technological venture capital operations were affected.

Due to the current provisions for indemnification, including the high compensations awarded by American courts and the existence of triple damages in the United States, patent assertion strategies appear “profitable” and encounter considerable success²⁶.

Chien [CHI 09] showed that the damages received by litigation PAEs has been, on average, more than double the damages received by practicing entities during the period studied and that PAEs won 29% of their cases, compared to 41% for practicing entities. The data from Lex Machina [LEX 16] shows that the median damages received by PAEs have greatly increased, rising from about $9.9 million over the 1996–2000 periods to $13.3 million between 2011 and 2015. It is nevertheless best to be cautious in interpreting the results given that they are strongly influenced by a relatively low number of very substantial amounts. In addition, the data from Lex Machina [LEX 15, p. 8], representing 42,805 patent cases judged before the Federal Circuit from 2000 to 2014, show that 13.4% of these cases terminated on merits, but only 1.8% of them involved compensatory damages for the plaintiff. The Lex Machina [LEX 16] still seems to show a turning point in the dynamic of patents issued. On the other hand, damages reached their highest level in 10 years, with a total amount of 10.2 billion dollars in 2015. If we break it down, the damages awarded to NPEs are almost three times higher than damages awarded to practicing entities over the past 5 years. What’s more, the litigations conducted by NPEs were concentrated in five district courts (out of a total of 94). These were responsible for 45% of all decisions involving NPEs. Finally, 80% of the district court decisions were appealed and 53% of the decisions that were appealed were modified.

Data from PWC [PWC 14, p. 6] completes this description and shows a significant difference between the median damages awarded to practicing entities compared to non-practicing entities, like PAEs, of about 2.5 million dollars for the former versus 8.5 million dollars for the latter between 2010 and 2013, despite the fact that practicing entities, unlike PAEs, can collect lost profits in addition to reasonable royalties. The Lex Machina study [LEX 16] also demonstrates a significant difference between the median damages awarded by judges compared to those awarded by juries²⁷. Between 1996 and 2000, the median damages awarded by the judges in bench were $7.2 million and $6.9 million for juries. Between 2001 and 2005, there is an inversion with $1.4 million in damages awarded by judges and $13.2 million by juries. Between 2011 and 2015, the difference grew even more and fell to $0.6 million for judges versus $10.1 million for juries.

An interesting aspect to highlight is that PAEs are increasingly seeking to control the fees that they must incur to defend their interests during a legal battle. To do this, PAEs attempt to exploit economies of scale by contesting the same patents in different courts. They also often rely on contingent fees with lawyers in order to mitigate the costs of litigation. In practice, this means that PAEs are establishing a link between the result of the action and lawyer’s fees, which generally correspond to a percentage of the amount recovered by the client.

The lawyer only bills the client if he wins the case or if it is favorably settled out of court, or in other words, “No win, no fee!”. It is undeniable that in patent infringement actions, the high cost of court fees for both parties can influence the result of the litigation just as much as the appropriateness of the claim. According to an economic study conducted in 2009 by the AIPLA (American Intellectual Property Law Association) [AME 13], in patent infringement cases valued between 1 and 25 million dollars, the total court costs were about 3 million dollars (roughly 60% of which was related to the discoveries process). When the amount was greater than 25 million dollars, the court costs were about double. For smaller cases, when the amount at issue was less than 1 million euros, the court costs were sometimes greater than the amount at issue and the costs incurred by the end of the discovery process corresponded to approximately 60% of the total.

3.2.4. The consequences for innovation

According to Bessen and Meurer [BES 14], intellectual property has become a source of tension. Litigations involving PAEs resulted in direct costs of 29 billion dollars for potential infringers in 2011. In a context of patent thickets, due to a system in which patent applications are secret for 18 months²⁸ and the patent issuing process is slow²⁹, the threat of hold-up is very real (Shapiro [SHA 01a]). Consequently, a company can be a victim of a “submarine patent”, undetectable before its attribution. In practice, this means that the aforementioned company would be unable to identify the relevant patent or patents, as the number of patents filed increases the complexity of studies about the freedom to operate³⁰. For instance, Bessen et al. [BES 11b] showed that the activity of litigation PAEs contributes to reducing incentives to innovate and a significant loss in terms of social welfare. When choosing whether or not to invest, companies must consider the cost related to the risk of inadvertent infringement³¹. Some studies suggest that the practices of litigation PAEs have negative effects on sales and innovation in many industries. Tucker [TUC 14] showed that defendants suffer a considerable decrease in sales during the trial and that litigation can negatively affect producing entities, independent of the result of the trial.

Bessen et al. [BES 11b] showed that litigation tends to drive down stock market prices for the alleged infringers and that this is greater than the revenue recovered by litigation PAEs, which implies a clear social loss.

According to [BES 11b], the increased risk of litigation translates into a static effect on the total social welfare. Litigation activities divert the production of resources but without creating any value. They are also accompanied by a loss in trading gains related to an increased market power resulting from the application of a profit margin in the royalty due to the PAE; even its stacking (following the logic of double marginalization detailed in Chapter 1 for the analysis of patent thickets) with those of the licenses already mobilized by the producing firm. It is this dead loss that causes several authors to say that litigation PAEs operate a “tax on innovation” (Bessen et al. [BES 11b], Cooper [COO 14], Posner [POS 13]³², Tucker [TUC 14]³³). In addition to this static effect, there is a dynamic effect, because the lower return on investment for innovative firms targeted by PAEs reduces their incentive to innovate. We could object that these effects are neutralized as soon as the action by the PAEs is anticipated and that consequently, producing firms are proactive by better examining which patents are likely to be infringing. In other words, the threat exerted by litigation PAEs would have a preventative effect and would generally lead to a greater respect for patents. However, Bessen et al. [BES 11b] showed that the evidence contradicts this optimistic perspective. The literature about litigation shows that the effects in terms of welfare are characterized, in reality, largely by a deadweight loss (Bhagat and Romano [BHA 98, BHA 02]).

The risk of a static deadweight loss, and its dynamic echo in the form of fewer incentives to innovate, is multiplied by the fundamentally asymmetrical relation between the litigation PAE and the company that it targets. Asymmetry in the negotiation is due to the fact that the PAE tries to litigate as late as possible, when the infringing company has already incorporated the patented technology into its product. Having committed irretrievable investments, the “target” company is therefore ready to make concessions. This hold-up strategy allows the PAE to collect a disproportionate amount of the final product (Lemley and Shapiro [LEM 05], Shapiro [SHA 01a]) in the form of royalties that are a multiple of what the target company would be ready to pay ex ante (Reitzig et al. [REI 07]). In particular, this type of strategy can be found in standard-setting procedures that involve essential patents. Thus, Rambus³⁴ was accused by European and American competition authorities of having committed a “patent ambush” set by intentionally concealing the patents and patent applications it held concerning technology used by the Joint Electron Device Engineering Council (JEDEC) standard, and then claiming royalties for these patents³⁵. This type of behavior, which could constitute a violation of the rules of competition, strengthened the debate about regulations concerning standards related to patents (Besen and Levinson [BES 09a], Devlin [DEV 09], Hovenkamp [HOV 08]). Similarly, companies can turn to PAEs to temporarily “borrow” or buy (such as “leasing-back”) certain titles that are used by the PAEs against the competitors of their clients. This type of behavior can also be likened to raising a rival’s costs (Salop and Scheffman [SAL 83, SAL 87]). The problem is that it is difficult to detect and even more difficult to discourage. The asymmetry in the bargaining power related to the hold-up is often paired with an asymmetry of information about the quality of patents held by the litigation PAE. Some studies show that litigation PAEs seem able to negotiate royalties that are clearly disproportionate in light of the damage caused or the quality of the patents in question (Sag and Rohde [SAG 06]). PAEs succeed at doing this through threats (especially of injunctions) that they could impose to stop the allegedly infringing production.

In the context of the coming implementation of the European unitary patent and its corollary, the unified court, there is a concern that this court could become an attractive forum for the activity of litigation PAEs in Europe, given that it will be possible for them to obtain a single prohibitive measure for all participating countries, creating a leverage effect. However, the European Patent Office is known to be more vigilant than the USPTO in its issuing process and less inclined to apply the principle of rational ignorance³⁶. Although the practices of patent trolling are mainly concentrated on the other side of the Atlantic for the time being, they are still sparsely observed in Europe³⁷ and Asia. Although most studies and authors cited above see this as a problem, others, on the contrary, insist on the positive effects.

3.2.5. A longstanding and potentially beneficial role

As noted earlier, complaints against PAEs, and more specifically litigation PAEs, are numerous. A critical analysis of the most common complaints, however, prompts us to take a more nuanced look at the role played by patent assertion entities in the patent system.

One of the most common complaints against PAEs is the fact that they do not take part in manufacturing activities and only buy titles to extract rents, as opposed to the actual implementation of the invention by manufacturing entities (presumed to be more deserving). Yet, as Khan [KHA 13] emphasizes, the founding principle on which patent rights were built, especially in the United States, has never been that an invention must be actively exploited by the rights holder³⁸. The main concern of legislators was to provide incentives to innovate and facilitate the disclosure of information about patents, not the production of goods. It was also to ensure that other inventors benefit from the discovery, either through licenses, circumventing the invention, or the expiration of the patent. From a legal point of view, patents are property rights. Like all forms of property, the decision whether or not to exploit a patent is entirely at the discretion of the title holder, in the same way that owners of physical goods are allowed to determine their use or not as they see fit. Imposing restrictions on patent holders concerning the use they make of them is a violation of patent law. To do this would be to forget that some inventors are not able to exploit their inventions themselves. Finally, this would also affect the signal function that was discussed in Chapter 1 and which is characteristic of the recent development in the use of patents by economic actors; a development which tends to disassociate the patent from the industrialization and commercialization of the patented invention. In reality, the troll phenomenon is not new and is not an anomaly. Khan [KHA 14] showed that PAEs were the norm in the 19th Century and that patent holders who granted licenses were also the most productive inventors. According to Chien [CHI 14], PAEs are also a response to the hold-up and reverse hold-up problems, that is, the propensity of some companies to deliberately ignore others’ patents, especially patents held by SMEs. Most of the time, firms argue that they do not have to sign a licensing agreement because the patent is invalid or not infringed, or even give as pretext that the licenses proposed are not fair or reasonable³⁹. In that case, patent holders do not have any other solution than to make recourse ex post to assertions if the manufacturers ignore the requests ex ante. Chien [CHI 14] showed that because small companies do not have the means to defend their patents, due to the high cost of litigation as well as the risk that the patent may be invalidated in the end, they sometimes turn to PAEs to assert (indirectly) their rights. From this perspective, we can consider that litigation PAEs effectively support SMEs, and more particularly start-ups, in the development of a technologies market.

The second complaint is based on the fact that the patents used by litigation PAEs are supposedly of questionable validity. In other words, many patents concerned by these legal proceedings could potentially be invalid, which would make it difficult for litigation PAEs to prove a violation. Contrary to this perspective, Fisher and Henkel [FIS 09] demonstrated the quality of the patents held by PAEs. Based on an analysis of 565 patents acquired by portfolio and litigation PAEs between 1997 and 2007 versus 1,130 patents acquired by manufacturing companies used as a benchmark, Fisher and Henkel [FIS 09] showed that patents held by PAEs were distinctly higher quality than those in the benchmark group. These results were confirmed by Shrestha [SHR 10] who showed that, contrary to popular belief, patents held by PAEs receive more citations than other patents and therefore a priori have a higher value. Allison et al. [ALL 09] showed that the fact that the most contested patents are disproportionately held by PAEs⁴⁰, coupled with the fact that these patents also have a higher private value, must lead us to reconsider the strong connection often made between PAEs and patents of questionable quality. Finally, it should be noted that the “bad” patents often used as anecdotes to justify and illustrate the low quality of patent portfolios held by PAEs were granted many years ago, at a time when it was difficult for examiners to refer to the relevant prior art⁴¹.

A third complaint is the supposed “explosion” of litigation initiated by patent trolls. If we define an explosion as an unusual increase compared to a previous trend, it is possible to determine to what extent recent litigation is abnormal. Khan [KHA 14] showed that the supposed “explosion” of patent litigation simply represents a parallel increase in patent requests. Khan [KHA 14] showed that the historical trend (over the period from 1790 to 2000) of litigation rates does not support allegations that disputes have “exploded” beyond the long-term norm over the last decade. In reality, it shows that litigation rates were much higher before the Civil War and during the market expansion that started in the 1870s. The increase of litigation noted at the end of the 20th Century is therefore nothing more than a return toward a long-term norm. Based on historical examples, Khan [KHA 13] showed that “patent wars” also took place in the past and that they surrounded the appearance of the sewing machine⁴² (see Chapter 4), the turbine and aeronautics (see Chapters 1 and 4). In other words, “patent wars” are historically associated with disruptive technologies. In addition, it is important to note that although according to Chien [CHI 09], 61% of legal proceedings initiated in matters of intellectual property during the year 2012 were by PAEs, their scope was limited to sectors where the probability of seeing patents infringed is higher, notably industries were the technology is complex⁴³. Allison et al. [ALL 09] showed that patent litigation affects industries related to the Internet, ICTs and young start-ups more strongly. More than 90% of litigations involving PAEs had to do with software patents or the finance industry⁴⁴. For instance, Allison et al. [ALL 12] showed that patents related to the Internet are 7.5 to 9.5 times more frequently disputed than patents that are not related to it. If there is a problem, then it is circumscribed.

Finally, it is important to note that PAEs have an intermediary role and perform the socially useful function of facilitating the development of technologies markets. They contribute to the logic of specialization inherent in the development of these markets. The threat that they pose to their potential victims prompts vigilance regarding respecting others’ patents and in doing so, generally reinforces the effectiveness of these rights. Criticizing the fact that these PAEs do not have manufacturing activities ignores the fact that while individuals or companies may excel in matters of invention, they might not be the best at transforming the inventions into innovations themselves because they may not have sufficient resources or expertise to successfully create a license policy for their technology and enforce their rights. Economic theory shows that, in this case, the most efficient result is to allow certain third parties with the required expertise or financial means to buy and manage the rights to a patent. By performing this intermediary function, PAEs allow inventors to generate revenue that they could not otherwise obtain⁴⁵. In return, this is an additional incentive to innovate and it encourages the creation of an efficient market for the exchange of technologies (Geradin et al. [GER 11], Shrestha [SHR 10]). Based on the purchase of patents and the sale of licenses, this market makes it possible to realign the incentives of market participants by allowing for higher market liquidity, and is a source of private gains. In doing so, PAEs contribute to the spread of new technologies to companies who are better equipped to commercialize them. Indeed, intermediaries like PAEs make it possible to reduce the costs of research and exchange and to increase liquidity, especially in markets where there are asymmetries of information. This point was outlined by the FTC in its report on PAEs that recognized the function of liquidity and the opportunities for risk mitigation that portfolio PAEs can offer [FTC 16]:

“The Portfolio PAE model may serve as a mechanism for shifting the financial risk of assertion activity to individuals or entities more able and willing to bear such risk, which may be more attractive to some patent owners than asserting the patents themselves. By raising capital from investors and purchasing patents with a large up-front payment, the Portfolio PAE provides the patent seller with guaranteed revenue and zero risk. The investors, who may have greater risk tolerance, then stand to enjoy the financial upside of successful assertion activities. In addition, manufacturing firms may transfer patents to Portfolio PAEs for assertion because Portfolio PAEs may enjoy lower costs, lack of reputational concerns, or licensing experience owing to their specialization in patent assertion”⁴⁶.

We could, however, object to the argument of accumulated liquidity because it may lead to more patent transactions. When the adjustment between patents and technologies is imperfect, a liquid market encourages the allocation of patents based on their litigation value to the people who are most likely to choose to enforce their rights rather than commercialize the invention. This means that patents negotiated on the market can be held disproportionately by entities that choose to capture expected cash flows through legal action. The result, in turn, can translate into more litigation over patents. More troubling still, a liquid market for patents would probably have an impact on the behavior of innovators. Inventors could find it more profitable to rely on patent protection to earn gains on their R&D investments rather than using other mechanisms that are less costly socially. The community could therefore obtain the same amount of innovation but at a higher social cost. On the other hand, the possibility of obtaining a financial return on patented inventions by selling the underlying patent in a liquid market could incentivize inventors to modify their R&D activities in order to be able to patent, even if the social value of the invention is lower. Just as stronger rights on patents do not provide more innovation, but simply more patents, a greater liquidity could, in certain circumstances, generate more patents rather than more innovation.

As indicated by Smith [SMI 76], the specialization and division of labor are endemic to efficient markets. Through their activities, some PAEs are better able to exploit patents due to their skill at selecting property titles and defending them, but also in light of their negotiation abilities, reputation as defendants and relative immunity against retaliation. This explains in part why more and more practicing entities are using the services of PAEs in order to assert their patent portfolios aggressively and obtain cross-licenses in portfolios held by PAEs.

3.2.6. Proposals for reforms

As indicated in 1966 by the US Supreme Court, “a patent is not a hunting license”⁴⁷. Although litigation PAEs can have negative effects on innovation, this should not overshadow efficiency gains that result from their activity, especially the improvement of market efficiency. Generally, if everyone has an interest in getting rid of questionable patents that are the cornerstone of the empire built by some patent trolls, then first of all, the already-existing ways to counter questionable patents should be used. In practice, this means that patentability standards must be improved, but also that litigation costs should be reduced, especially the rules for compensation in case of “inadvertent infringement”. In this regard, in recent years, we have noted the implementation of some reforms or bills filed before the American House of Representatives and the Senate moving in this direction as well as clarifications to the jurisprudence of the Supreme Court in order to remedy the supposed abuses of patent trolls⁴⁸.

In addition, some companies are seeking to organize against the phenomenon of PAEs. Aside from the industry practices regarding cross-licenses, new models of risk reduction against PAEs are emerging. For example, Google is promoting a “license on transfer” (LOT) system that guarantees to the beneficiary of a license that in case of patent transfer to another entity, the new entity is contractually obligated to renew the license.

Finally, as is stressed by Lemley in [LEM 07a], it would seem that the final frontier in the war against PAEs is to attack innovation in universities. The Electronic Frontier Foundation (EFF) filed a bill, the Reclaim Invention Act⁴⁹ whose goal is to encourage universities (which are, from this point of view, likened to PAEs) to manage their patent portfolios to maximize public benefit. More specifically, the goal is to encourage universities to sign a public patent pledge not to sell or grant any exclusive licenses to patent assertion entities. Faced with laws that “punish” them for not being sufficiently entrepreneurial and the multiplication of legal proceedings that intend to put an end to the commercialization of their inventions, universities must make a choice à la Hobson: either license the small fraction of research that could generate royalties and potentially see their legal responsibility engaged, or simply abandon the commercialization of their inventions (and the revenue necessary to finance fundamental research).

3.3.1. FRAND licenses as safeguards for essential patents

As soon as a standard involves one or more essential patents for its implementation, the problem of a potential hold-up arises. In the literature, these patents are known by the acronym “SEP”⁵⁴. Although there is no single commonly accepted definition, a patent is considered to be essential to a standard (SEP) when it is not possible, for technical reasons, to manufacture products complying with a standard established by a standard-setting organization without infringing that patent. According to Tapia [TAP 10] and an RPX report [RPX 14], there is a tendency to over-declare SEPs which leads to granting them too much importance, at least in numerical terms. A similar concept would be “Market Essential Patents” (MEPs)⁵⁵. Such patents cover a functionality that the majority of final users expect to find, such as a camera in a cell phone⁵⁶. MEPs are not subject to FRAND conditions presented here but can nevertheless provide their holder with considerable bargaining power.

In order to protect against potential hold-up behaviors by SEP holders and preserve a broad and non-conflicting diffusion of standardized techniques, most standard-setting bodies have implemented a certain number of internal rules. For instance, several of them ask their members to disclose, as soon as possible and continuously, potential patents that relate to a standard in the process of being developed⁵⁷. In the event that there is a patent (or a patent request) for the technique in question, they also require that their members negotiate with companies desiring it to reach a license contract based on fair, reasonable and non-discriminatory conditions (FRAND license) or a revenue-free license (RF license). In theory, this commitment to FRAND conditions is a tool used by standard-setting organizations to fight potential abuses, such as excess revenue requests⁵⁸, and to strike a balance between the rights of patent holders and the rights of potential licensees. In other words, the goal of FRAND licenses is to ensure effective access to a standard while preventing a single owner of essential patents from having a stranglehold over the market in question.

However, standard-setting bodies are generally not very specific about the meaning of the terms “fair” and “reasonable” in the FRAND commitment. The vague nature of FRAND commitments is obviously related to uncertainty about future demand and costs for the new technology, but also the fact that some members of standard-setting bodies benefit from this ambiguity. The fuzzy or incomplete character of certain clauses can therefore favor opportunism and in fine facilitate obtaining an excessive or discriminatory royalties rate or signing a biased cross-license agreement. However, competition authorities are hesitant to intervene in business related to private disagreements over FRAND conditions⁵⁹, preferring to defer these litigations to the courts. The basis for this reasoning is the following: FRAND commitments create contractual rights for third parties⁶⁰. Cases of not respecting contractual obligations relate, consequently, to contract law that allows for implementing the responsibility of a co-contractor. In principle, it is not up to competition authorities to set the price of FRAND royalties⁶¹. The courts and arbitration are supposed to be more appropriate to examine the excessive or discriminatory nature of royalties applied to SEPs and to determine “reasonable” royalties. Despite their reservations, competition authorities are nevertheless “forced” to intervene due to contradictory decisions made by the courts and the risk of forum shopping, or in other words, the existence of an opportunism in the choice of jurisdiction before which the litigation will be brought⁶². It is therefore necessary to first evaluate the reality of the negative consequences attributed to the SEP in terms of competition and on the pace of innovation and then, second, to examine the position adopted by the courts and different competition authorities in Europe and the United States.

3.3.2. The hold-up theory faced with the facts

Although there are many works that explore the theoretical conditions in which hold-up can be produced, this literature simply demonstrates the possibility that excessive royalties or an injunction (or the threat of an injunction) toward a potential licensee can potentially be anti-competitive. According to Abbott [ABB 15], the opinion that hold-up is a widespread or systemic problem that is harmful to consumers has not actually been verified and is mostly anecdotal. To date, there has been no strong empirical evidence to support the theory that hold-up practices through patents prevail, nor any evidence demonstrating that royalty-stacking or injunction threats would have negative consequences on competition and innovation (Barnett [BAR 14], Geradin et al. [GER 08], Geradin and Rato [GER 10], Layne-Farrar [LAY 14], Sidak [SID 15])⁶³. At least four series of works support this statement.

The first series of works, conducted by Galetovic et al. [GAL 15], sought to evaluate whether the harm identified by the hold-up theory, namely higher costs for consumers and a lower quality and variety of products, were true⁶⁴. More specifically, the authors relied on the concept of hedonic prices to determine if SEP holders reduce, all other things being equal, the lowering of prices and/or the increase in the quality of goods to the detriment of the consumers’ welfare by exercising a greater market power⁶⁵. Their rationale is that if SEP holders are likely to adopt a hold-up behavior that is harmful for innovation, then products that rely on SEPs should experience a greater stagnation of quality-adjusted prices than similar products that are not based on SEPs. In this approach, the decrease in prices with constant quality is used as a proxy-variable to measure the rate of innovation. The results show that in the United States from 1997 to 2013, the rate of innovation as reflected by the relative price adjusted as a function of quality has rarely, even never, been as high as it is today for products that are theoretically subject to hold-up. Galetovic et al. [GAL 15] showed that the price of products that are based on SEPs has dropped if we compare them to other patented products that are not based on SEPs. Similarly, Galetovic et al. [GAL 15] tested whether the tightening of conditions to grant an injunction for SEP holders, following the decision in eBay⁶⁶ made by the US Supreme Court in 2006, had accelerated the decrease in prices adjusted as a function of quality. The authors demonstrated that this was not the case because the innovation rate in industries that rely on SEPs has not increased, compared to other industries. This offset between the reality and the theory is confirmed by analyses related to the smartphone sector, often highlighted to illustrate hold-up problems. A study by the Boston Consulting Group [BEZ 15]⁶⁷ showed that despite an increase in the number of SEPs incorporated into smartphones, the price of wireless telephone services in the United States has decreased compared to the global Consumer Price Index (CPI) since 2005. The cost per megabyte of data actually decreased 99% between 2005 and 2013 (demonstrating both the innovation effort to make data transmission more efficient and the effectiveness of competition). The price in dollars per megabyte has dropped by 95% since the transition from 2G to 3G and by 67% since the transition from 3G to 4G. The average global sale price of smartphones decreased by 23% between 2007 and 2014, while the average price of low-end telephone devices dropped by 63% over the same period. In the end, contrary to the premise of the hold-up theory, innovation has continued at a rapid rate and competition between mobile phone manufacturers is very robust⁶⁸. Consequently, there is a gap between the theory and the reality of the facts. More generally, the study by the Boston Consulting Group showed that industries dependent on SEPs record the fastest decreases in prices in the United States. Clearly, at this stage, issues of hold-up seem to be only theoretical, not proven.

As appealing as the hold-up hypothesis may be, it seems, according to this first series of academic works, that it only represents a simple theoretical possibility. This can be explained by the fact that patent holders often benefit from a “first-mover” advantage if their technology is included in a standard. Consequently, an SEP holder is confronted with an inter-temporal trade-off. The first option is to practice higher royalties, meaning proceeding to a hold-up. They will generate profits in the short term, but at the risk of discouraging the adoption of the standard and leading to incremental innovations in the same vein. In the long-term, this is likely to encourage the emergence of a competing standard at its expense. The other option is to grant licenses with favorable conditions that promote the adoption of the standard and incremental innovations in the short term, so as to compensate for the relatively low inflow of gains from the licenses over the duration of the patent. The SEP holder may find the second strategy more advantageous if they plan to maintain a first-mover position in later evolutions of the standard. This strategy also confers them a “good” reputation vis-à-vis licensees and “immunizes” them against the fear of a “hold-up”⁶⁹. This is an example of the well-known principle of reputation as a solution to a problem of adverse selection in the context of repeated transactions⁷⁰. A complementary explanation is that the possibility of hold-up is not one-sided. Firms that hold SEPs can also be locked in if their technologies have a market limited to the technology covered by the standard⁷¹. Consequently, the incentives to engage in hold-up goes both ways: there is also a possibility of “reverse hold-up”, or in other words, situations where licensees use the fact that they are the main users of a standard to exercise a market power to get rates and terms below the FRAND royalties. This configuration is similar to the hold-out described in the previous section dedicated to patent trolls to the extent that it is all the more likely to occur if the SEP holder is a SME with few means to assert their rights and if the licensee is a large-sized firm. The theoretical analysis suggests that reverse hold-up can occur because, ex post, the parties face an asymmetrical risk. Patent holders are tied by their FRAND commitments, but licensees are clearly not. The threat of litigation related to FRAND commitments can therefore be used as leverage, in particular when the opposing party cannot support the costs of legal proceedings. Lerner [LER 95] examined the behaviors of 419 biotechnology companies in patent matters and showed that, all things being otherwise equal, the companies with the higher legal costs had a tendency to patent fewer innovations than their competitors with lower legal fees. This would suggest that small companies (of biotechnology, in this case) design their research strategies to avoid legal conflicts with large companies and that to protect their initial investments, they have a tendency to rely more on secrecy than on patents compared to large companies.

In a second series of works, Langus et al. [LAN 13] proposed a formal invalidation of the hold-up theory. Their works showed that despite the availability of injunctions, holders of a sufficiently weak patent will end up accepting rates that are lower than the FRAND level, notably when the cost of litigation is higher. Langus et al. [LAN 13] go so far as to demonstrate that there is a possibility of holdout or runaway, especially when licensees refuse to take a FRAND license or do so after a delay. In practice, the potential licensees can impose delays on taking licenses in order to “exhaust” the patent holder or propose insufficient royalties, even going to court in several countries, patent by patent, until all options are depleted.

It should be noted that these hold-up theories assume that SEPs necessarily confer the existence of a market power, but without saying it outright. The theoretical literature (Farrell et al. [FAR 07], Lemley [LEM 06], Lemley and Shapiro [LEM 07b]) is focused on the consequences of the market power obtained through an SEP⁷², but few works examine whether the inclusion of a patent in a standard itself creates a market power ex post. In this case, it is nevertheless important, as indicated by [FTC 07, p. 2], to distinguish two potential sources of market power: one that comes from the technology itself and one that comes from the standard.

In a third series of works, Layne-Farrar and Padilla [LAY 11a] attempted to fill in this gap. To that end, they examined whether the inclusion in a standard affected the value of a patent, as measured by forward citations and whether this had an impact on the market power or the bargaining power of patent holders. Their study, based on an analysis of patents declared to be essential for standards in telecommunications and imaging technologies, showed that if being included in a standard has a positive effect on the value of a patent, then the effect, as measured by forward citations, is very marginal or even negligible. There are two explanations for this. The first explanation is that SEPs are self-declared by their holders to standard-setting bodies but no one assesses their “essential nature”⁷³, which, itself, can change over time as the standard develops⁷⁴.

It is therefore erroneous to assume that an SEP systematically confers a market power. It is more likely that the standard-setting committees tend to “crown a winner” rather than create them because the most influential technologies are, according to Layne-Farrar and Padilla [LAY 11a], naturally the ones that are destined to be included in a standard. A contrario, their data showed that cases where a standard made a patent into a “winner” on the market are relatively limited. In other words, the existence of a greater market power precedes the inclusion into a standard and cannot be considered to be caused by it. A second explanation for the absence of a causal link between the inclusion into a standard and the value of an SEP conferred by an increased market power resides in the fact that most standards are constructed around not one but several strongly complementary SEPs. This means that any user of the standard must obtain several licenses simultaneously from the SEP holders who will not necessarily cooperate with each other to set their royalties. Like in the problem of “patent thickets” described in Chapter 1, the sum of the royalties that result is higher than the royalties that a single holder of all of the SEPs in a standard would apply. The absence of cooperation between the holders of complementary SEPs is detrimental to the good development of the market and hampers in part their market power and hold-up possibilities.

A fourth series of works conducted by Rysman and Simcoe [RYS 08], as well as Lerner et al. [LER 07], focused on the reverse causality of the previous series by studying the probability of inclusion in a pool of patents based on the value of the patent. To do this, the authors tested whether higher quality patents were more likely to be included in a pool of patents, the majority of which develop in a formal cooperative standard. Their conclusions showed that standard-setting committees tend to attract patents with the greatest value, in the sense that SEPs are the subject of a much greater number of forward citations (before inclusion) than the average patent. This means that a significant part of patents incorporated into a standard respond to the “crowning the winners” principle. These authors showed nevertheless that inclusion in a standard can have a positive effect on the subsequent importance of a patent, as perceived by the number of citations after inclusion. Lerner et al. [LER 07] showed, for example, that patents included in patent pools receive more citations. However, Rysman and Simcoe [RYS 08] showed that the effect of a pool on the importance of a patent is concentrated in a relatively short time period. They noted that the difference is limited to the years immediately before the formation of the patent pools and the years immediately following their establishment. There is rather a synchronicity between the number of forward citations and the inclusion in a standard, which only confirms the essential nature of SEPs to the standard in question.

To date, only a small number of studies have attempted to validate the hold-up theory empirically, and notably to verify if SEPs necessarily confer a market power to their holders⁷⁵.

The empirical studies mentioned here, although limited in number, demonstrate that standardization does not automatically confer market power, but rather “crowns the winners,” or in other words, the most important technologies⁷⁶. According to Layne-Farrar and Padilla [LAY 11a], these are “natural candidates for inclusion in standards”. Similarly, these studies underline the possibility of a hold-out in the context of which rights holders would be poorly remunerated for their investments in R&D. This is obviously not without consequences because this reduces the expected return of inventions, which can also decrease investments in R&D. At this stage, we can only note a lack of indirect or direct empirical evidence concerning the probable effects of this alleged behavior. It might also highlight the difficulty in rigorously defining an appropriate counterfactual to evaluate hold-up behaviors. In the absence of solid evidence of the negative effects attributed to SEPs, the jurisprudence attempted to identify the circumstances where courts and antitrust authorities can consider that a hold-up is likely to have anticompetitive effects.

3.3.3. The availability of injunctions

After a first generation of cases concerning refused licenses by vertically integrated firms wishing to extend downstream, the activity of courts and competition authorities has tended, in recent years, to concentrate more and more on supposedly abusive strategies in patent matters⁷⁷. These strategies, sometimes described as “evergreening”, are based on a sometimes faulty use of patent law⁷⁸. More prosaically, competition authorities on both sides of the Atlantic are now confronted with the difficulty of defining circumstances in which the holding and use of an SEP can constitute an abuse of a dominant position in the sense of Article 102 of the Treaty on the Functioning of the European Union (TFEU) in Europe, or represent an unfair method of competition according to the terms of Article 5 of the (FTC) Act in the United States⁷⁹. In the context of SEPs, there is a risk of anticompetitive harm if the title holder prevents other companies from using the patented technology (for example, by refusing to grant a license, or by refusing to grant it with “reasonable” conditions, or by soliciting an injunction) and consequently from implementing a given standard.

Courts and competition authorities on either side of the Atlantic have attempted to provide legal clarity as to what constitutes an appropriate framework to resolve conflicts over FRAND licenses. In particular, courts and competition authorities have been solicited, in recent years, about the circumstances in which the effective use, or threat, of an injunction action burdens the relation between an SEP holder and a potential licensee (hoping to negotiate a license) and in what circumstances this could be considered anti-competitive or, on the contrary, fall under “competition on merits”. This reflection is part of a context where the possibility for SEP holder to obtain an injunction in cases of patent violation is increasingly contested. Traditionally, SEP holders can, in the context of an infringement action, request an injunction in order to prevent the alleged infringer from continuing to use the inventions claimed in the patent. Similarly, they can request pecuniary damages in the form of lost profits or a reasonable royalty. Recently, however, some commentators have expressed the opinion that when a patent is an SEP and, in addition, is encumbered by a fair, reasonable and non-discriminatory (FRAND) license, then the holders of this patent must be prevented from obtaining an injunction, because the risk is that they will use it to seek to obtain unfairly high royalties from firms carrying out the standard (who are often competitors) and in the end, obstruct the implementation of the standard.

3.3.4. Patent ambushes

Another practice that has also been the subject of investigation in the past few years concerns what is called a “patent ambush”. This practice occurs when a patent holder who participated in a standard-setting process conceals (or declares belatedly) its patent requests or patents that are essential for the application of a standard in order to use them against market actors after the adoption of this standard. The holder is therefore in a position to require the users of the standard, who have often made substantial and irreversible investments in its implementation, to pay higher license royalties than the holder would have obtained if the technology had not been included in the standard¹²⁹.

The courts and competition authorities on both sides of the Atlantic have had to deal with these patent ambush strategies in a small number of cases¹³⁰. To this day, the most iconic case remains the Rambus¹³¹ case. In this case, the company Rambus was a member of the Joint JEDEC, a standard-setting body based in the United States, that developed a standard for DRAMs (Dynamic Random Access Memory). Rambus had participated in the standard-setting process, but without revealing its standard-essential patents. Following its exit from the JEDEC consortium, Rambus imposed its patents on users of the standard after its adoption.

Following the complaint filed on December 18, 2002 by two companies, Infineon and Hynix, the European Commission addressed Rambus with a notification of grievances¹³² on July 30, 2007, considering that Rambus may have violated Article 102 TFEU by abusing its dominant position in the DRAMs market. In its preliminary assessment, the European Commission found that Rambus held a dominant position because it created a “lock-in effect” as the other companies could not circumvent the DRAM standards. The European Commission also considered that Rambus had violated EU rules concerning Article 102 TFEU by engaging in a patent ambush by concealing the fact that it held patents and patent requests concerning the technology used in the JEDEC standard and by then demanding excessive royalties for the use of these patents.

In doing so, the Commission presented a new typology of abuse, specially developed for ambush cases. In fact, “the Commission took the view that Rambus may have been abusing its dominant position by claiming royalties for the use of its patents from JEDEC, compliant DRAM manufacturers at a level which, absent its allegedly intentional deceptive conduct, it would not have been able to charge”¹³³. In other words, the abuse consisted of demanding a price higher than what could have been requested if the company had not created specific circumstances by concealing pertinent information, when other technologies that could have substituted were available.

Two years after the communication of grievances, Rambus presented a list of commitments intended to satisfy the concerns raised by the Commission and end the proceedings launched against it¹³⁴. In the context of this set of corrective measures, Rambus accepted not to collect any royalties for the standards concerning the SDR and DDR chips that had been adopted when Rambus was a member of the JEDEC and only to collect a maximum royalty of 1.5% for the latest generations of JEDEC standards concerning DRAMs (DDR2 and DDR3). Due to the rapid obsolescence of the technologies at issue, the commitments were limited to 5 years.

Initially condemned by the American FTC on the basis of the same grievances¹³⁵, in the end, Rambus obtained a reformed decision on appeal due to insufficient proof¹³⁶. The Court of Appeal of the Federal Circuit ruled that the FTC had not demonstrated abuse because it had not rejected the possibility that the JEDEC may have developed the same standard, even in the absence of the deceptive behavior.

3.3.5. Royalty-stacking

Competition authorities have also analyzed the consequences of royalty-stacking observed on different patents in the same standard. The number of patents essential for a standard has considerably increased in recent times, doubling on average every 5 years since the 1990s (Bekkers et al. [BEK 12, pp. 20–21])¹³⁷. However, as noted by Shapiro [SHA 01a] and Lemley and Shapiro [LEM 07], when multiple essential patents are required to create a product and when they are held by different companies, the royalties will most likely be set independent of one another. In this case, a problem arises that is similar to patent thickets, namely the existence of multiple margins¹³⁸, which means that the final price can be greater than what a single patent holder, holding all of the patents, would charge. In other words, the cumulative royalties to pay for the different SEPs in a single standard can be higher than reasonable levels or even become prohibitive when it comes to commercializing the product. The idea that prevails is that royalties for SEPs should correspond to a share of the single royalty that would be collected at one time if all of the SEPs in a standard were held by a single agent¹³⁹. Despite the small number of cases related to royalty disputes, there seem to be two methods for determining this share [SID 13]. In the first method, called the “proportional contribution” method, the FRAND royalty is determined by considering the portion of the price of the final product that corresponds to the compensation of the standard as a whole, then within this compensation, the portion that amounts to each SEP considering its contribution to the standard. In the second method, called the “top-down” method, the FRAND royalty is determined by taking the average profit margin realizable on the smallest saleable component then, within this margin, the portion that amounts to each SEP considering what it contributes to the standard. In both cases, the last step is particularly delicate because it requires isolating the value contribution of an SEP to a standard, for instance by means of hedonic price estimates.

The theoretical possibility of both royalty-stacking and a hold-up was noted by the American courts in cases of determining FRAND conditions. In the cases of Microsoft v. Motorola and Re Innovation¹⁴⁰, the judges considered that the potential for royalty-stacking had to be taken into account to determine FRAND terms without requiring real proof of their existence¹⁴¹. Since then, other courts have ruled that potentiality is not sufficient and that the behavior must be proven. In Ericsson v. D-Link Systems, the Federal Circuit ruled that to be considered in an analysis of FRAND damages, the problems of hold-up and royalty-stacking must be demonstrated and not only assumed¹⁴² (Layne-Farrar and Wong [LAY 15]). However, this is difficult to establish because license agreements are often confidential and there are few factual elements available. We are therefore reduced to basing our analysis on indirect elements.

Nevertheless, it is important to keep in mind that patents that are truly essential are perfect complements, which requires a de facto connection between the patent holders, such that the SEPs cannot be the subject of a license in an isolated way. Contrary to oligopolists who can set their prices without considering the effects produced on other firms present in the market, an SEP holder cannot act unilaterally and must coordinate with other holders of patents that are essential to the same standard when setting royalties. Generally, it must be noted that the market actors that participate in the work of standard-setting bodies coordinate between themselves and often do not have another option on the market to sell their technology. They therefore have an interest in ensuring the commercial success of the standard and certainly not in condemning it, which would create a royalty-stacking. In addition, as noted by Geradin et al. [GER 08], the market addresses royalty-stacking with mechanisms such as cross-license agreements and patent pools. Similarly, the coercive measures that make it possible to enforce a patent are costly: employing them is only justified if the financial stakes are high. This means that as a whole, the problem of blocking patents and royalty-stacking are limited due to the best interest of the actors concerned. In telecommunications, experience shows that SEP holders have collectively decided to cap aggregate royalties for standards related to 3G and 4G. They have understood that royalty-stacking à la Cournot goes against their own best interests¹⁴³.

3.3.6. “Best FRAND forever” or the delicate question of royalty amounts

Litigations over supposedly excessive or discriminatory rates applied to SEPs result, in part, from the increasingly frequent transfer of SEPs following mergers and acquisitions and/or the sale of patent portfolios due to bankruptcy. Most often, the new rights holder does not consider itself to be bound by a prior commitment to FRAND licenses, especially after the bankruptcy process of the previous owner¹⁴⁴. The risk here is that the procompetitive effects of standardization will be wiped out if, following the sale of essential patents, the FRAND commitment is not also transferred¹⁴⁵. British authorities had to address this type of issue in Nokia v. IPCom¹⁴⁶ in 2009. In this case, Nokia filed a complaint against IPCom, following its purchase of patents that were essential to the GSM standard held by Bosch, after IPCom refused to offer FRAND licenses. This case was completed by a commitment decision. IPCom was committed to resuming the FRAND commitment made by Bosch for the use of the essential patents now held by IPCom.

More fundamentally, the debate about respecting FRAND conditions mainly focused on the relevance of approaches to determine the basis for calculating a “reasonable” royalty. The difficulties in the matter can be illustrated by a small number of cases in which the courts granted FRAND royalties at rates or bases well under what was requested by the licensors. Table 3.1 summarizes the different cases. In Innovatio¹⁴⁷, the royalty rate requested was reduced by the court, falling from $40 (or 6% of the final product value) to less than $0.0956 per unit.¹⁴⁸ In the case opposing Motorola to Microsoft¹⁴⁹, the latter alleged that Motorola had violated its FRAND license commitment by requesting a supposedly excessive royalty of four billion dollars. The jury estimated, in this case, that the appropriate amount for a FRAND royalty was actually $0.03471 per unit of product, for a total of 1.8 million dollars representing less than 0.1% of the initial price requested¹⁵⁰.

In Ericsson v. D-Link¹⁵¹, the jury awarded damages to Ericsson in the order of 10 million dollars for the infringement of three out of five patents, based on a royalty rate of $0.15 per infringing device, while the initial request was for $0.50 per product for the five patents at issue. On appeal, the Court of Appeal of the Federal Circuit (CAFC) ruled that the judge should have explained the 15 criteria of the Georgia Pacific test to the members of the jury to determine the nature of Ericsson’s FRAND commitment and a hypothetical royalty rate. Consequently, the court overturned the damages awarded¹⁵². Similarly, a Chinese court ruled that the royalty rate for essential patents held by InterDigital for 2G and 3G standards could not exceed 0.019% of the sale price of the Huawei devices (Alfonso and Zech [ALF 13]).

These cases are often used to “demonstrate” the existence of hold-up behaviors on the part of some companies. Yet, despite how interesting they are, a handful of legal cases does not indicate in any way that the opportunistic behaviors are widespread. In reality, they only illustrate the difficulties in calculating FRAND royalties. On the other hand, the extremely modest appraisals of the royalties set by the courts in these cases raise legitimate concerns that SEP holders could be under-compensated, not over-compensated. The difficulty in determining the “reasonable” nature of a FRAND royalty was also highlighted in Qualcomm in 2009. In this case, the companies Ericsson, Nokia, Texas Instrument, NEC and Panasonic accused Qualcomm of abusing its dominant position by refusing to grant licenses with FRAND terms for its essential patents for the UMTS standard. Due to a lack of evidence, the inquiry about its abuse of the dominant position was closed¹⁵³. In January 2017, the American FTC relaunched proceedings against Qualcomm, accusing it of having violated antitrust legislation during the sale of certain components and licenses to smartphone manufacturers, including Apple.

Faced with these difficulties and the uncertainty about the legal decisions made in determining royalties, competition authorities sought to remove the ambiguities about FRAND conditions by providing a “road map” to the courts. The basic idea is that the royalty amounts must be based on the importance of the technological contribution and not the potential for harm from a possible hold-up. The objective must be to get close to the value provided by the patented technology to the product itself, independent of the value associated with the standardization. In addition, in a report from 2011, the FTC recommended that American courts evaluate the “reasonable” character of FRAND royalties by simulating what would be the result of a hypothetical ex ante negotiation (i.e. before a standard is validated and switching costs arise) between the patent holder and a potential licensee who is “well disposed to reach an agreement”. This amount includes a term integrating the probability that the patent is not valid, the incremental cost of licensing the technology and the relative value of the patent. In doing so, there is the guarantee, according to the FTC, that the royalty reflects only the technological value of the patent, as opposed to the value resulting from a potential hold-up position that an SEP may obtain because it is included in a standard. This ex ante analytical framework is based on the incremental value of the patented technology compared to alternative available solutions before the definition of the standard. The analytical framework accounts for the rate (or range of rates) that the companies would have accepted if the alternatives competing with the SEP had not been eliminated (and the market power had not been conferred or increased) by the standard-setting process. The analysis is based on the methodology developed in the Georgia Pacific jurisprudence¹⁵⁴ in 1970 which set out fifteen criteria to determine a hypothetical royalty rate¹⁵⁵, calculated as if there had not been a patent violation but an open negotiation. In practice, in 2013, American courts adopted a modified version of the Georgia Pacific test in the cases re Innovatio IP Ventures and Microsoft Corp. v. Motorola, Inc. to recreate the conditions of a hypothetical negotiation between the parties. The new methodology (1) aims to limit what a patent holder obtains to a reasonable royalty on the economic value of the patented technology itself, outside of the value associated with the incorporation of the technology into a standard; (2) considers the importance of the SEP for the standard and the importance of the standard and the SEP for the products at issue; (3) only considers comparable patents for reference and (4) considers evidence of royalty-stacking while keeping in mind the global royalties that would apply if other SEP holders were to make royalty requests. The basic idea is that the advantages awarded to patent holders must be more directly related to the value of the technology to which they contribute. The explanation is that when compensation and contribution are aligned, economic efficiency is encouraged because investment in the development of new technologies are proportionate to the profits.

According to Elhauge [ELH 08] and Ganglmair et al. [GAN 12], the approach recommended by the [FTC 11] to reduce hold-up behaviors could translate into royalties that are too low, reducing the incentives for firms to innovate. In addition, as attractive as it may be, such a method poses practical problems. When calculating the additional value of an SEP, should only the other options assessed by the standard-setting body be considered, or all possible technologies available up to that point? In the case of complex products that include thousands of patents, is it possible to adopt a “topdown” approach and break down the product into unitary components¹⁵⁶? In the affirmative, should the relevant base for calculating royalties be “the smallest marketable unit using the patent”, like in the CAFC’s decision in Laser Dynamics Inc. v. Quanta, or the entire product¹⁵⁷? In practice, the methods based on costs are hard to mobilize because it is difficult to match costs with a patent or a part of an actor’s patent portfolio. In addition, as hold-up can only be produced once a standard has been established, we may consider it advisable to use the theoretical amount of royalties ex ante. However, this all remains very theoretical because, in reality, the license terms are most often negotiated afterwards, especially since patents are frequently not declared until very late in the process. In reality, there is no reference point to serve as a comparison. In addition, such a method is abstract due to the fact that the standard can be used in different applications and that, consequently, different evaluations of the standard can occur. This is also disregarding the fact that the payment of royalties can take different forms. For example, the payment of lump sums, permanent royalties associated with lump sums or in rate form (Caffarra et al. [CAF 14]). According to Lerner and Tirole [LER 15], the current policy in terms of licenses is formally an infinite price cap. They therefore propose implementing structured price commitments, or in other words, a combination of a FRAND license and a price cap¹⁵⁸. However, the authors admit that this may not be possible or may produce negative secondary effects.

Layne-Farrar et al. [LAY 07] maintain that FRAND rates should be calculated using the Shapley value¹⁵⁹, that is, according to their marginal contribution to the total value of a technology. Mariniello [MAR 11] discusses the limits of these approaches by emphasizing the merits of an ex ante/ex post approach, while warning that due to imperfect information ex ante, an appropriate definition of a FRAND royalty should involve “rates that are not worse than those to which the patent holder would be committed ex ante in the context of competing to define a standard, considering the information available ex post”. He then suggests a test based on four conditions for cases in which FRAND commitments cannot be violated. Mariniello [MAR 12] suggests, in particular, that the FRAND concept is not supposed to be associated with a specific price ex ante. The information available ex ante about the value of a technology is often low. The value of the technologies does not materialize until the standard is effectively implemented, when patent holders assess the relevance of their patent portfolio compared to those held by other firms and production strategies, and when the market starts to develop and end customers finally reveal their preferences. The impossibility of passing complete contracts ex ante requires that FRAND commitments must be sufficiently flexible to allow the price of the technology to adapt to the value that is revealed ex post, when the information is improved and the uncertainty is minimized. However, the inherent ambiguity of the meaning of the term “fair” in the FRAND concept leaves a wide margin for interpretation and, consequently, leaves the door open for disputes.

European competition authorities were also led to open inquiries into the application of anti-trust rules to licensing terms for so-called FRAND essential patents. These inquiries follow different interpretations of FRAND commitments by Member State courts. The Microsoft case illustrates the approach taken in Europe. In 2004, the Commission ruled, followed by the European Court in 2007, that a company in a dominant position did not have the right to require a royalty based on the strategic value (or hold-up value) of an indispensable asset, in this case confidential and patented information for operating systems on personal computers, necessary for other actors to evolve on the neighboring market for operating systems for work groups. This resulted in a decision by the Commission and a judgment by the CJEU in 2012 concerning royalty levels¹⁶⁰. This decision created the rule according to which the innovation value (or the inherent value) of an interoperable patent must be analyzed independently (that is, only in reference to the sole market of products incorporating the standard) and that the patent holder cannot profit from revenue expected by the licensee on a neighboring market (in this case that of operating systems for work group servers). The inherent value of a patent is the marginal value that the licensee can draw from the use of the patent on a relevant market compared to the second-best alternative, which must be analyzed ex ante before the lock-in. This value can, in certain cases, be verified empirically. In the particular case of Microsoft, for example, Microsoft’s interoperability protocols were available for free ex ante, that is, before compatibility with operating systems for personal Microsoft computers becomes indispensable and there is locking. The Commission concluded that ex ante it was logical to freely share information in matters of interoperability because this (and the existence of a multitude of complementary products) makes the PC operating system more attractive and increases the profits that Microsoft earns from this product. The Commission also analyzed the prices of comparable interoperability solutions. The methodology here relies on a comparison of comparable product prices, a method that was already used in the United Brands case¹⁶¹. In Microsoft, the company was required to transfer licenses with FRAND conditions, which is no longer a truly voluntary FRAND commitment. The court indicated that it was not up to the Commission to set a precise amount, and the Commission itself agreed that it should not play a regulatory role in matters of FRAND royalties¹⁶².