Patent trolls are just a small piece of the patent aggregation puzzle. A myriad of electronics companies, with and without production capabilities, pursues patent aggregation by building patent portfolios and monetising them beyond manufacturing. So far, it is unclear whether such patent aggregation activities have adverse effects on innovation. This article replaces the myopic discussion about patent trolls with a wider picture—that of patent aggregation. It analyses the rationales and evolution of this phenomenon, orders the domain of patent aggregators, and presents the effects patent aggregation activities can have on technological development. Only the interplay of patent and competition laws is considered apt to reconcile patent aggregation with innovation.

Low manufacturing costs of today's global electronics product markets drive price competition to almost perfection.¹ Competition shifts on quality and innovation bringing Intellectual Property, particularly patents, onto centre stage. However, quality and innovation depend on costly trial-and-error research and development investments, whose returns can hardly be achieved through direct sales of patent implementing products alone.² Here patent aggregation comes into play as an alternative path to raise revenue and finance further R&D. Many electronics firms, such as Qualcomm or Huawei, couple end-product sales with patent aggregation practices such as patent licensing and selling (so-called practising entities PEs).³ Other firms, such as France Brevets or Sisvel, completely forego manufacturing and specialise in patent aggregation activities (so-called non-practising entities NPEs). As a natural consequence of the property nature of patents, both PEs and NPEs can enforce their patents against infringers. Whether patent aggregation equates with anti-innovative patent trolls cannot be a summary judgment dependent on the vertical integration of the patentee under scrutiny.⁴ Rather, it is a factual case-by-case assessment of the effects on innovation of the specific patent aggregation activity considered. EU competition law warrants such an approach, which ensures a positive relationship between innovation and competition law. In light of technological development, patent aggregation activities can either be proscribed as anti-competitive or justified even if otherwise anti-competitive. The remainder of this article impartially analyses the rationales, evolution, and multifaceted impact on innovation of patent aggregation. It concludes that only the interplay of patent and competition laws reconciles patent aggregation activities with technological development concerns.

Patents are one of the public mechanisms to incentivise innovation. Together with direct public research, public procurement, subsidies, and prizes, patents remedy the public goods pitfalls of inventions favouring their disclosure. Once an inventor reveals an unpatented invention, he can hardly prevent others from using it (i.e., non-excludability), also because a single use does not preclude (i.e., non-rivalry) or diminish subsequent ones (i.e., joint-consumability). Without patents, society would experience fewer inventions or, at best, inventions only in limited fields. On the one hand, rational inventors would not undertake R&D investments, lacking the prospect of being able to appropriate their results.⁵ On the other hand, they would focus on those technological areas where appropriation can be achieved without patents yet otherwise, such as through lead-time, sophisticated product design or secrecy.⁶ In contrast, the patent right to exclude anyone from practising one's invention, enforceable in courts through property and liability remedies, enables patentees to disclose their inventions while inhibiting free riding.⁷ Hence, during the years of patent protection, patentees can profitably commercialise their inventions, recover R&D costs, and devote rents to new inventions.⁸ Accordingly, patents incentivise both inventive activity and the disclosure of its results, while maintaining up-front technological neutrality for R&D investments. Moreover, the public grant of patent exclusivity to inventors is not cheap, as there is no such thing as a free lunch. Patentability, disclosure, and maintenance fee requirements raise the bar for inventors and limit patent rights for the essential benefit of society. In addition to the aforementioned right to exclude, patents also confer the often neglected right to include. In fact, patent rights are as transferable as any other property right.⁹ This transferability permits inventors to share their knowledge while obtaining remuneration for it. In so doing, patentees can specialise in inventing and outsource the risks of making the invention into a final product.¹⁰ Because of patents, each industry player can do what he does best: inventors invent, intermediaries intermediate, and implementers implement.¹¹ Insofar as patent aggregation activities involve knowledge transfer, collaboration, and the division of innovative labour, it is wrong to associate them with patent trolls.

Patent aggregation is commonly conceived as the monetisation of patents beyond the sale of patent-implementing products.¹² Despite being not very informative, such conception has the merit of tracing back some long-lasting forms of patent aggregation. In this sense, the pooling of technologically-related patents from different owners into a single licensing program dates back to the mid-XIX century. The Sewing Machine Combination of 1856, the Motion Picture Patents Company of 1908, and the Manufacturers Airplane Association of 1917 are all examples of patent pools created to license technologies owned by multiple parties as a single package obviating anti-commons and double-marginalisation issues.¹³ Next to patent pools, cross-licenses, whereby two parties grant each other access to the respective patent portfolios, have been a feature of all cumulative technologies industries for decades.¹⁴ However, modern patent aggregation practices started with the rise of open innovation business models. Facing the globalisation of production chains and distribution channels, companies like IBM, Xerox, and Texas Instruments began by the late 1980s to heavily license their technologies across entire industries including to the competition. Nor did they hesitate to recur to patent infringement litigation against reluctant licensees.¹⁵ Intermingling collaboration and rent-seeking motivations, these patent aggregation pioneers achieved memorable royalty earnings, which showed the immense value lying behind dormant patent portfolios. Since then, very few companies let their patents sit without providing direct monetary value.¹⁶ The patent licensing potential attracted the entrepreneurship of a new genre of firms, so-called NPEs, exclusively operating upstream in the technology supply chain as providers of patented technology or freedom-to-operate services.

Considering the contemporary patent monetisation practices of both PEs and NPEs, patent aggregation more precisely comprises any activity where patents are aggregated through direct prosecution or transfer and are then exploited for non-manufacturing purposes. Apart from patents, patent applications, patent commercialisation rights or even utility models can be similarly the object of aggregation activities. Ownership or control of the patents is equivalent insofar as the patent aggregating company decides the access terms to the patents.¹⁷ Following such a definition, patent aggregation conducts identify two subsequent stages. The first stage involves building patent portfolios, namely sets of related patents with a common function. While patent prosecution unequivocally identifies the filing of patent applications, a transfer is manifold. Indeed, the ownership or control of patents can be transferred either directly through patent purchases and exclusive long-lasting licenses with sub-licensing rights, or indirectly by merging with or acquiring patent owners.¹⁸ Electrical engineering firms venture all patent portfolio-building paths. For example, electrical engineering is, according to data from the EPO Annual Reports, the technological field with most European Patent applications—at least since 2001—and with most grants since 2017. Likewise, patent transfers have increased in Europe since the 1990s, with electrical engineering patent transactions involving more patents on average per transaction than any other sectors.¹⁹ Last, patent portfolio transfers often drive mergers and acquisitions of electronics firms, such as between Nokia and Alcatel-Lucent or Google and Motorola Mobility.²⁰

No a priori difference exists between a homegrown portfolio or an externally-developed one, since both confer the same strategic benefits and can be exploited beyond manufacturing. After all, what difference does it make for the law if a PE—take IBM for example—directly makes money out of its homegrown patents or an NPE—say Sisvel—does the same with somebody else's patents? From an EU competition law perspective, probably none since both PEs and NPEs can raise collusion, exclusion or exploitation issues. Indeed, any patent portfolio multiplies rather than sums the strategic effects conferred by its individual patents for the better or worse of innovation, thus exacerbating all competition law concerns over individual patent exercises. While a patent gives the rights to exclude and include anyone over the protected claims, a patent portfolio gives such rights over a much broader scope of subject matter. In doing so, patent portfolios confer market foreclosure power, and their magnitude attracts external innovation, shields from litigation, diversifies risks and hedges against uncertainties. It also improves bargaining positions, enhances freedom-to-operate by counter-infringement threats, increases the patentee's weight over patent politics, and allows to attract and retain investments.²¹

The second stage of patent aggregation starts after a patent portfolio is built. Patentees have four non-manufacturing exploitation options, namely out-licensing, selling, enforcing or defensive-holding. First, patentees such as the above-mentioned Huawei or Sisvel, out-license their patents to third parties, either out-sourcing the implementation of their inventions or expanding it to new markets and applications. Second, patentees sell their patents. However, while the financial parameters of licenses allocate innovation risks between licensor and licensee, the one-off nature of patent sales determines an all-or-nothing approach to the allocation of innovation risks between licensor and licensee. Because of the possibility of over- or underpayment, patent sales usually involve technologies out of the core business of a patentee, see Google's 2018 battery patents sale to Amperex Technology or Intel's 2019 cellular wireless connectivity patents sale to Apple.²² Alternatively, patent sales contribute value to restructuring proceedings, as it happened for Nortel, Kodak or Qimonda.²³ The third non-manufacturing exploitation option is to enforce patents, which is especially lucrative for electrical engineering technologies. In fact, the incremental, convergent, interoperability-driven, and fast-pace characteristics of electrical engineering technologies determine a high likelihood of patent infringement in today's global electronics product markets.²⁴ Because patentees are entitled against infringing products to injunctions stopping entire production lines, enforcement quickly leads alleged infringers to settle and accept licenses. Indeed, patent enforcement rarely ends in litigation, which has uncertain outcomes but certain length and costs. For example, Lemley, Richardson and Oliver recently estimate that less than one-third of patent infringement assertions result in litigation in the U.S.²⁵ To counter the spread of nuisance or litigation-cost-based patent enforcement, defensive-holding emerged as the last non-manufacturing patent exploitation option. While PEs have always held their portfolios to shield-off infringement litigation through the threat of counter-suit in mutually-assured-destruction (MAD) settings, a few NPEs started by the mid-2000s to build patent portfolios upon which they offer freedom-to-operate as a service to their clients. Nowadays, diversified business models pursue patent aggregation practices.

A patentee's position in the supply chain materially affects his options as to how to build and exploit a patent portfolio. Accordingly, the primary distinction of patent aggregators is between PEs and NPEs. PEs are vertically integrated patentees, which operate on both the upstream technology market and the downstream product market. Having both proprietary technologies and manufacturing operations, PEs at times out-license, in-license or even cross-license patents to avoid infringement litigation. Instead, NPEs only operate upstream in the market for patented technology. Without production capabilities, NPEs are not vulnerable to counter-suit nor do they have reputation concerns vis-à-vis consumers. Business model considerations distinguish six main types of NPEs, namely patent pools, technology-transfer offices (TTOs), R&D firms, patent funds, patent assertion entities (PAEs), and defensive patent funds. Patent pools, TTOs, and R&D firms are the more straightforward types of NPEs for innovation. They build their patent portfolios to reach profitable licensing contracts with manufacturers in support of the latter's commercialisation activities. Instead, the remaining NPE types are more controversial since their activities usually omit technology transfer and are so remote from invention implementation efforts.

1. Patent pools, starting with the above-mentioned Sewing Machine Combination, license patents bundled from different owners. Nowadays, more than 30 patent pools worldwide manage licensing programs mostly of complementary standard-essential Information Communication Technologies. Patent pools' licensing expertise, independence and neutrality from industry participants make them attractive business partners for PEs.²⁶
2. TTOs provide prosecution and commercialisation services for the inventions of universities and other public research organisations.²⁷ For example, Max- Planck Innovation is the TTO responsible for bringing the inventions of Max Planck Institutes to industry, inasmuch as KT-CERN does the same for the European Organisation for Nuclear Research.
3. R&D firms, such as InterDigital, Experi (previously known as Tessera) or Rambus, internally develop technologies, prosecute patents over them, and then monetise in the market the patented technologies.²⁸ Differently from PEs that outsource the fabrication of their patented products to third parties, yet directly market them, R&D firms supply their patented technologies to original equipment manufacturers.
4. Patent funds are recent NPEs that raise money from PEs and/or from capital markets, acquire patents that fit coherent patent portfolios, and then exploit them benefitting from arbitrage to achieve a return on investment.²⁹ Intellectual Ventures is perhaps the most emblematic patent fund backed by PEs (e.g. Microsoft, Intel, Sony etc.), and by institutional investors (e.g. Charles River Ventures, McKinsey and Co., Sequoia Holdings etc.). Acacia Research, Quarterhill or Marathon Patent Group are instead examples of patent funds publicly traded on stock exchanges. Finally, Intellectual Discovery, IP Bridge, France Brevets and Ruichuan IPR Funds (Zhigu) are sovereign patent funds respectively created by the governments of South Korea, Japan, France, and China.
5. PAEs, previously known as patent licensing and enforcement companies and pejoratively called patent trolls or sharks, build patent portfolios in light of their infringement value rather than their technological merit, to profit through patent licensing, settlements, and damage awards in courts.³⁰ Some, such as Papst Licensing, Conversant (Mosaid), or Unwired Planet are formerly PEs that started to monetise their homegrown patents and then abandoned manufacturing operations to become completely PAEs. Others, such as Innovatio IP Ventures, IPNav or Dominion Harbor directly started as PAEs. Last, certain PAEs develop a specific contractual relationship with PEs known as patent privateering. In a patent privateering agreement, a PE assigns some of its patents to a PAE, for little or no consideration up-front, yet retains a financial interest in the subsequent enforcement of the divested patents.³¹
6. Defensive patent funds, which emerged as a market response to PAEs, buy patents, either with their own capital or upon members' solicitation and finance, to provide freedom to operate as a service to their own members or subscribers.³² Defensive patent funds, such as AST, RPX or SynPat commit to never assert their patents. However, their defensive holding may be followed by other patent aggregation activities both to recoup patent acquisition costs and to prevent free-riding of non-members (which otherwise would benefit from the commitment not to sue). Thus, defensive patent funds also out-license the acquired patents to third parties and offer subscriptions to new firms (a so-called catch and hold strategy) or resell the bought patents after having licensed their members (a so-called catch and release strategy).³³ See Table 1.
   The classification of patent aggregators informs the analysis of the effects of patent aggregation on innovation. It has the benefits of improving the understanding of the markets where patent aggregators operate, identifying who their clients, suppliers, partners, and competitors are. However, it cannot be rigidly applied since patent aggregators may engage in many patent aggregation conducts over or at the same time. The discussion must objectively focus on the specific activity pursued by the patent aggregator, which impacts innovation, and not on (to some extent) arbitrary entity labels.

The relationship between patent aggregation and innovation depends above all, on what we mean by innovation. Defining innovation as the exploitation of new markets or as new ways of doing business, patent aggregation is in itself an innovation. On the one hand, patent aggregation involves the exploitation of patent portfolios into both totally new markets, such as the market for freedom-to-operate services, and previously under-exploited ones, such as cross-industry technology markets. On the other hand, patent aggregation represents a new way of managing patents that enables companies unlock their idle value. In this regard, the stigma against the first companies that started patent aggregation practices can be explained by their role as mavericks disrupting traditional industry practices.

More profound is the effect of patent aggregation on innovation when it is understood as the dynamic process of bringing new products or processes to markets. Each patent aggregation activity presents pros for innovation that can be offset by corresponding cons. See Table 2. First, patent aggregation activities allow for economic specialisation and division of R&D labour.³⁴ Yet, they might entail an inefficient allocation of innovation risks, whereby patent aggregators profit from patented inventions at the expense of those who put such inventions into practice. While prosecuting patents to license or sell them renders economic specialisation and division of innovative labour benefits, prosecuting in order to enforce or hold pre-empts the implementation of patents into products against innovation. Second, patent aggregation practices ameliorate the market for patents.³⁵ However, rents from patent aggregation activities might as well be insufficiently redistributed to original inventors or reinvested into R&D. Any non-manufacturing exploitation of externally developed patent portfolios poses such opposing effects. Third, certain patent aggregation activities accelerate technology diffusion and enhance follow-on innovation, whereas others lead to either unilateral or coordinated innovation suppression. Therefore, the enforcement or holding of patents without implementing them into products undermines innovation more than their licensing or sale, all else being equal. Fourth, patent aggregation that fights technology free-riding (so-called patent hold-out) is often borderline with opportunistic, exclusionary or exploitative patent hold-up, namely outright foreclosure or excessive rent extraction from alleged infringers corresponding to nuisance or litigation costs rather than technological value. Patent hold-up is more likely if the patent licensing or enforcement presents opportunistic features, such as occurring in parallel with industry fairs, consumers shows, and initial public offers, involving soon-to-be-expired patents or targeting the retail and user levels rather than the manufacturing one. Fifth, all patent portfolio-specific advantages for innovation have downsides. For example, portfolio licensing, whereby patents are licensed as a bundle, shortens negotiation times, minimises transaction costs, avoids litigation, provides freedom to operate, certainty and predictability to parties. However, it might also shield weak patents from invalidation, anti-competitively tie unnecessary patents, and overburden the licensee. Last, one-stop-shop efficient patent aggregation activities, solving double marginalisation and royalty-stacking issues contrast with fragmentary and anti-commons ones. Of course, out-licensing only complementary patents saves transaction costs.³⁶ More contradictory is the case for sales or enforcement of externally developed patents representing the disintegration of previously united portfolios as it occurs in privateering.