By Garrard R Beeney and Renata B Hesse
- From growing numbers of SEPs to questions over screening methods, despite their impressive track record there is still mistrust of patent pools.
- However, the four most frequent criticisms simply do not stand up to the fact that this is a purely consensual licensing model on both sides.
- Patent pools can still be improved and critiques should therefore be analysed for future licensing innovations.
Dozens of patent pools operating for dozens of years have built up an impressive track record, leading to widespread recognition of the model’s benefits. Nevertheless, several widespread and high-profile critiques will not go away.
Whether a pool meets market needs or creates harmful disincentives depends on choice. If the pool terms are unacceptable, licensees will not accept the licence. If an implementer genuinely feels that the pool terms do not meet its IP needs, it has three alternatives:
- negotiate bilateral licences outside the pool (ie, the option that would exist without the pool alternative);
- test its theory of the value of patented inventions to its products through mediation or arbitration (assuming agreement on the process) or pursue litigation; or
- proceed without using the IP rights of others.
The fourth alternative – using intellectual property without offering a fair return to the owner – is not an option that benefits innovation, consumers or even the holding-out implementer, whose ability to sell new products in order to meet consumer demand often depends on incentivising innovation by paying a fair return on valuable inventions.
At the end of the day, any criticisms of pools must account for the fact that they are entirely optional.
With this in mind, analysis of some of the most common issues raised demonstrates why the reasoning behind them is unfounded.
“There has been a proliferation of SEPS”
While the growing number of standards and the increasing complexity of solving technological problems have led to more patented inventions being declared essential, it is difficult to see how more SEPs is a valid criticism of anything – much less of patent pools.
In an age of accelerated technological development led by a growing number of innovators, innovation often builds on existing inventions. Thus, for example, a new generation of video compression may use many of the tools of the predecessor technology, but add multiple innovations to increase functionality.
This is happening more frequently as innovators and standards-development organisations (SDOs) seek to avoid technological obsolescence by not only developing new technologies, but also making these backwards compatible.
The rapid advancement from MPEG-2 to AVC and then HEVC is an excellent example of this, as is the development of 3G to 4G and then 5G. The natural by-product of this is a rise in the number of patented inventions in standards.
Properly viewed, the increasing number of SEPs is therefore beneficial – although there is a legitimate question as to whether patentees may over-declare to SDOs the number of patents that are essential. Even there, the issue is far from straightforward. While such over-declarations may create licensing burdens, the process itself has benefits as declared patents are likely subject to obligations to license on FRAND terms. Indeed, many SDOs recognise that under-declaration is more of a problem than over-declaration and impose penalties accordingly.
In any event, typical pools correct for over-declaration by requiring an independent determination beyond any SDO essentiality declaration before including it in the licensed portfolio. Thus, to the extent that a proliferation of SEPs is perceived to be a problem, pools provide a solution.
“Pool screening methods are ineffective”
Critics also claim that the methods of determining essentiality are ineffective. These typically involve an independent expert deciding whether one or more claims of a patent read on a portion of the licensed standard. If the evaluator finds that a portion of the standard cannot be implemented without infringing the examined patent claim, then this is added to the pool; if it can, then the patent is rejected and not included. Pools may also allow patentees to build on prior essentiality determinations by independent experts through self-certification.
Patent pools invest substantial resources in this evaluation process. They retain worldwide networks of qualified patent experts and require licensors to fund and present robust evidence of essentiality. The European Commission’s 2020 pilot study on essentiality assessments found that “[t]o date, patent pools have the most sophisticated systems in place for essentiality assessment”. The cost of evaluation – often in the region of $10,000 per patent – creates incentives for patentees to carefully screen their portfolios and only submit for evaluation patents that are likely to be deemed essential.
While the process cannot account for good-faith human error, there is no empirical evidence that pools include non-essential patents to the extent that licensing terms are affected. Thus, there appears to be no support for the claims that:
- portfolio evaluation methods are ineffective;
- licences include, in any material number, patents that affect the royalty; or
- licenced portfolios contain, in any material number, patents that the licensee does not need or want to practise the licensed technology.
This is underscored by the fact that while there are powerful implementers with substantial resources that could re-examine any pool licensed patent portfolio for essentiality, no such project seems to have taken place. Rather, critics typically focus on a single (and dated) evaluation of 85 patents asserted in court by three highly litigious SEP owners, using this as a basis to argue that a high percentage of litigated patents are invalid or non-essential.
In addition to a lack of evidence, the theory that pools are affected by licensing non-essential patents suffers from another critical fallacy. As explained, pools cannot force their terms on implementers; any execution of a pool licence is by definition consensual. If the implementer does not believe that the pool terms reflect the value of the licensed portfolio, it can simply reject the pool offer and negotiate bilateral licences.
Again, we do not claim that every pool is free of patents that may be invalid or non-essential. Rather, pools generally employ structures to guard against such inclusion and there is no evidence (notwithstanding the incentive to investigate the issue) that pool licensing terms are affected by the inclusion of patents that the licensee does not need or want.
A slight variation of this criticism is that pools may license SEPs that are essential to optional, rather than mandatory, portions of standards. While this is no doubt true, there is no evidence that doing so affects licence terms. What is more, critics again misinterpret what actually happens in the real world.
The alternative to inclusion is exclusion. It appears that most, if not all, implementers would prefer that a pool provide a licence to practise optional portions of a standard rather than require multiple licence negotiations.
This is backed up by US Philips Corp v International Trade Comm’n (424 F 3d 1179, 1191 (Fed Cir 2005)), where the Federal Circuit noted that including patents that not all licences need in a pool generally causes no harm to competition.
After decades of successfully meeting market needs, there simply is no persuasive evidence that pool licensing terms are adversely affected by the inclusion of non-essential patents. Thousands of commercial licensing transactions strongly suggest that the opposite is true.
“Multiple pools, royalty stacking and comparable licences”
Another focus of criticism is the fact that multiple pools may form around a single standard (eg, there are three pools around the HEVC video compression standard). This is less efficient than a single pool. However, the fact that not all licensors could agree on terms for a single pool is hardly an indictment of the typical pool structure. Without these three pools, implementers would have to enter into 56 separate licence transactions with 56 distinct participating licensors – compared to a mere three.
We know of no justification (or indeed any principled contention) that the licensors in the three pools were required to license their intellectual property under a single set of terms. Pools are voluntary organisations in which licensors opt to provide an alternative to the marketplace. While three HEVC pools are less efficient than one, they are still fare more efficient than a world without HEVC pools.
Critics also claim that pools are responsible for stacked royalties, but evidence suggests that the opposite is true. ‘Royalty stacking’ refers to the fact that an implementer must pay royalties to multiple patentees in order to implement a single standard. However, when multiple patentees develop a standard – collaboration often being the cause of rapid innovation – royalties for that technology will necessarily be stacked. That is simply the function of collaborative standards development, not a function of pools.
Pools reduce the amount of stacked royalties – both in terms of the number of licensing transactions that an implementer must enter, and as several studies have found, likely with respect to the aggregate payable royalty as well.
Judicial authorities have also addressed methods of dealing with the issue of stacked royalties. One way to determine a FRAND royalty rate is the so-called ‘top-down’ method, by which the total reasonable rate for use of a standard is calculated and then divided by the total number of patents essential to that standard. Although subject to shortcomings (eg, imprecision in calculating the total number of SEPs and treating all SEPs as entitled to the same proportion of the royalty stack regardless of the invention’s quality), this provides an alternative method for calculating a total royalty.
“Patent pools wrongly license only at the end product”
One branch of dissent focuses on the tendency of pools to license end products, rather than products such as integrated circuits or components, which may also practise the licensed patents. Objectors claim that end-product licensing violates a licensor’s FRAND commitment to license to all implementers and/or sets a royalty unrelated to the value of the licensed technology because the patent is not practised at the level of the end product.
At least in the pool context, an obligation to license every implementation of a licensed patent would lead to uncertainty and additional costs, and would vastly decrease pool efficiency. By way of example, assume that a product in which the chip, the component including the chip and the finished customer product all infringe a portion (or all) of a pool patent portfolio. If the pool were required to license all those contributing a component to the finished product, it would be costly and inefficient – if not impossible – to track which products were licensed under which patents.
If some but not all chips on the market were licensed, the pool would have to track which fabricator’s chips were included in which components (and which component manufacturers held a separate licence) to determine which finished products still required a licence because they did not include a licensed chip or component. Similarly, a component manufacturer would have to determine which of the chips that it sources were licensed so that – assuming that patent exhaustion applied – it did not pay a royalty on a component that was fully licensed just because it incorporated a licensed chip. The same analysis would then have to be carried out by each finished product vendor.
Moreover, it is likely that in a large portfolio not all claims would be infringed by the chip (eg, system claims). Licensing at all levels of distribution could lead to endless disputes about which finished products are licensed under which claims as a result of upstream component licensing. This would lead to additional costs, long-running disagreements and inefficiencies. For pools, allowing end-product licensing only furthers the well-recognised benefits of the model.
These principles were recently recognised by the US Ninth Circuit Court of Appeals in Federal Trade Commission v Qualcomm Inc (969 F 3d 974, 984 n 5 (9th Cir 2020)). The court noted that end-product licensing is widespread (suggesting that it is the “industry norm”) for good and sufficient reasons, including that licensing at all steps in product distribution would “lead to inefficiencies”.
Similar conclusions have been reached by European courts, which have found that end-product licences are FRAND. (See also the UK decision in Unwired Planet v Huawei Technologies approving end-level licensing for 3G, 4G and 5G mobile services royalties as FRAND and the German decision in Nokia v Daimler, holding that Daimler’s offer to Nokia was not FRAND because it was based on the price of the component.)
To the extent that a component manufacturer believes that FRAND requires component licensing, it can – as several have – make such a claim in connection with bilateral licensing. But pools should not be required to license all steps in the product distribution process.
Further, a licence at the end product only avoids questions about whether different components use different parts of the portfolio and should be charged different royalties. Instead, it increases competition by charging competitors at one levelt he same royalty rate so that no one licensee is advantaged over another.
The courts have also recognised that royalty rates may logically and lawfully be based on the value of the technology to the licensed end product, rather than on the value of the smallest saleable patent practicing unit (SSPPU). As the Qualcomm decision held: “No court has held that the SSPPU concept is a per se rule for reasonable royalty calculations.”
Going further, the court rejected the theory that “patent royalties cannot be based on total [end product] price”. End-product licensing should apportion to the patented technology only that part of the end product value attributable to the patent(s) at issue.
In sum, imposing an obligation on pools to license at all levels of product distribution could defeat some of the very goals that pools seek to achieve: low costs, consensual licensing and the reduction of disputes.
Popular suggestions that would decrease the benefits of patent pools to implementers
Based on these criticisms, commentators occasionally offer suggestions for change. However, a closer look at these alternatives reveals that they would frequently prove counterproductive, making pools far less efficient and productive, and thus far less appealing to innovators.
“Licensors should require pool administrators to commit to FRAND licences”
First, while pool administrators that are not members of SDOs and have not participated in the standards-setting process have no contractual FRAND obligations, in practice, this is something that the administrators of the world’s largest and most successful pools regularly commit to.
Second, the FRAND commitment is made by individual licensors and the development of what terms are FRAND in real-world licensing (and the concomitant obligations that FRAND places on implementers to negotiate in good faith) has taken place in the context of bilateral (ie, not pool) licensing. Thus, for example, a FRAND obligation by an implementer to make a good-faith royalty counter-offer may not apply in the context of a non-negotiable pool royalty rate.
Moreover, the intent of FRAND – to make standardised technology widely available and free from the economic costs of patentee hold-up and to provide a fair return to patentees for their investment and innovation free from licensee hold-out, thus encouraging participation in standard-development activities – is innate to the pool structure. Pools make licences available to all that wish to implement the technology, and there is a standard rate and terms for similarly situated licensees. Any implementer is free to reject a pool offer at its discretion.
This freedom to reject pool terms – regardless of whether they are fair – is inconsistent with the implementer’s FRAND obligation to negotiate in good faith and accept a FRAND licence. Those that would impose FRAND terms on pools fail to address how doing so would affect a licensee’s freedom to reject.
Thus, it is at best unclear what new benefits would be provided if pool administrators, not just licensors, were required to make a legally enforceable FRAND commitment.
“Pools should provide a licensing history for pooled patents”
It is unclear what details about a licence to a small portion of the pool portfolio would tell an implementer about the pool licenec as a whole. Transparency is a worthy goal but having a pool act as the source of an exchange about licensors’ bilateral licences may do more harm than good. Properly run pools generally have bilateral licensing to individual patentees, with rules in place to prevent one licensor from influencing (or even knowing about) another’s terms. Bilateral licences should be just that, bilateral, without influence or involvement from a pool or other licensors in a pool.
Moreover, pools make transparent a far greater amount of licence information than is typically made available for bilateral licences, including terms, the licensed patents, licensors and other facts associated with the pool licence.
“Pools should impose further essentiality filters and conduct validity audits”
There is no evidence that existing essentiality determinations under current pool procedures are inadequate or that they affect pool licensing terms. Thus it is unclear what problems would be fixed by additional measures to assess essentiality. As for essentiality determinations (ie, whether one or more claims of the patent must be practised to implement the standard), patent pools generally provide (at considerable cost) an independent test of a patent holder’s essentiality declaration. There is no evidence that existing procedures are inadequate or adversely affect pool terms.
As for validity, while the system benefits from the removal of invalid patents, validity determinations are costly and complex, have been carried out at least once by the national issuing authority and are regularly the subject of good-faith disputes. It is not like an umpire simply calling balls and strikes. The cost of any such additional measures – particularly any widespread validity review – would have to be reflected in licence costs.
Without evidence that invalid patents are affecting pool terms it seems unwise to impose these.
If evidence were presented that licensing non-essential or invalid patents affected pool terms, that would be the time to discuss the benefits of new and more costly systems. However, until that occurs, expensive solutions are ill advised.
Where do we go from here?
While these criticisms are largely misplaced, there is – and should be – room for improvement in the patent pool model.
Key pool structures have evolved to optimise efficiency and enhance the results for all stakeholders. Just as these efficiency-enhancing steps have developed over time, there is no reason why continued evolution will not make pools even more valuable to our economies and innovation. Both innovators and implementers should work together to achieve any improvements that will address market needs.
For example, we may find pools offering licences to more non-essential patents or to more than one standard in the future. Perhaps implementers should be involved in the formation of terms at an earlier stage. We may also find that SDOs will provide incentives to form pools or that government initiatives (eg, tax incentives for royalties generated from pools) start to incentivise pool formation. This would be in the interest of the entire ecosystem. Other encouragements may include government subsidies to defray pool costs for additional measures in order to create increased transparency regarding licensed patents and/or other checks on essentiality or validity if necessary.
Incentives may also include additional safe harbours from competition laws as a result of pool participation or licensors providing a resource for others to consult about standardised pool terms. Many jurisdictions (eg, the European Union, Japan and the United States) already recognise such safe harbours. Competition authorities may revise existing rules to enable more effective communications about pool terms among innovators and implementers that recognise that they all benefit from reasonable pool terms.
However pools may develop in the future, we need more of them, not fewer. Increasing technological developments and the complexity of achieving true innovation, the adoption of standards in new classes of products and increasing uncertainty about fair licensing terms leading to costly litigation and substantial inefficiency – to mention just a few factors – all suggest that properly run pools should be encouraged. Honest dialogue about improving pool efficiency is a part of that.
The views expressed in this article are those of the authors and do not necessarily represent the views of Sullivan & Cromwell or any of its clients. The authors wish to acknowledge the assistance of Paul von Autenried in preparing this article.
Garrard R Beeney
Sullivan & Cromwell
Renata B Hesse
Sullivan & Cromwell