The EU impact assessment of Standard Essential Patents reform: Nice try but please try again!

The EU process to reform its policy on Standard Essential Patents (SEPs) should have been much better. When the proposal was unveiled in April this year, many observers and experts were surprised – including experts in EU authorities and member state governments that have followed the issue. Earlier this week, the European Patent Office stated its concerns with the proposal in a letter to the European Parliament – pointing, for instance, to the lack of patent and technical competence at the EUIPO (the authority designated to manage the new SEP policy). The Commission has been analysing and discussing matters related to SEPs for a long time. It has commissioned external reviews and evaluations, and routinely communicated its desire to have a system that is balanced between SEP holders and implementers. When it then released its proposal and its accompanying impact assessment, the package was not just unbalanced and expressing a highly simplistic view of how the SEPs market work; it was also riddled with unsubstantiated and unverified claims.

There are several other issues in the report which should be highlighted. In this article, I will draw attention to the glaring problems of the proposal and, especially, the impact assessment that undergirded it.

Transparency and patent pools

First, the Commission has assumed that there is a lack of transparency regarding the number of SEPs for a specific technology, which is impairing the development of products. However, this is a strange view of how it works. Many SEPs users have long experience of working with SEPs holders and licenses. For most actors, they develop their products first, and then they engage in licensing discussions. It’s only in rare instances that these discussions lead to litigation. When they do go to litigation, it is even rarer that the underlying concern is an insufficient understanding of the need to license on the part of the implementer. True, one could make an argument that there are too many relevant patents or that too many patents declared essential to a standard is not essential. However, that is also a difficult argument to make given that patents often are declared essential because they can be essential for some implementers using a standard – even if they aren’t for the dominant part of the SEP licensees. Perhaps there is overdeclaration (there is an inconclusive academic literature on overdeclaration and its potential significance) but that is not a transparency issue that confronts most licensees.

The types of problems that have been the basis for litigation, pertaining to licensing issues, increasingly often get settled outside the court, for instance through pool licensing. This mechanism was widely endorsed by the Commission in 2014. A study prepared then for the European Commission’s DG Grow said that patent pools are one-step solution for licensing a bundle of SEPs owned by different entities thereby aiming to mitigate transaction costs, avoid royalty stacking, and create a level playing field. (P. 13 of the study). Patent pools are now the dominant way for several implementer-sectors to access SEPs (especially the automotive sector).

The Commission has now shifted its approach in the new regulation and the assessment report. Under Article 11 of the proposed EU SEP regulation, the Commission has imposed new transparency requirements. The Commission fails to capture that there is no provision for the pools mentioned in the assessment which bars patentees from entering into bilateral licensing negotiations (Daimler settled a pending infringement case for a car-level patent license with Nokia, similar to the deals with Conversant and Sharp). This points that patent pools do not bar bilateral license deals. As such the Commission’s analysis on enlisting pools as a problem, is incorrect. The pool is not the problem, because if the pool rates were to become burdensome, bilateral licensing will eventually prevail, this will marginalise the pool and lead it to adapting to the rates, and as such the issue will be solved by markets. The underlying truth to this problem is all about “transactional efficiencies with the deals struck.”

Implementers normally take a pool license because it is more efficient than taking numerous bilateral licenses. In contrast to wanting more transparency, pools actually tend to be transparent in their standard royalties. Avanci for instance has been consistent in this approach. Avanci‘s pool includes 55 licensors and covers a wide majority of cellular SEPs. However despite the transparency, some users have resisted entering to an agreement. This indicates that disputes taking place are not due to the lack of transparency on information of an aggregate rate to be imposed on a standard, but on the calculation of the rate and whether that is FRAND.

In September 2022, 80 brands were included under the Avanci 2G-4G license which the Commission pointed to in the report (P.12). However, the Commission on the following page (P.13) put a figure which points to the “estimated accumulated sales on licensed and unlicensed connected vehicles based on Avanci’s announcements of licenses signed up to August 2022.” The figure becomes misleading and misrepresentative, primarily due to the fact that the Commission has already emphasized the September 2022 announcement in the text preceding the figure. Recently, Avanci announced the launch of 5G Connected Vehicle Program. Avanci currently holds an 80% share of granted and active patents in the worldwide 2G-5G spectrum. This statistic suggests that a significant portion of the automotive industry has licensed patents from Avanci’s pool. Overall, the impact assessment appears to have some issues in comprehending the dynamics of patent pools at present. Moreover, the Commission not only discredits Avanci despite it maintaining a consistent pool rate and adding value to it, but also enlisted Sisvel as a former operating company, when in fact it is a major pool administration with specialized patent assertion entities (footnote 128).

A register, FRAND determination and essentiality checks

Furthermore, the Commission is suggesting the creation of a SEP Register, while DG Grow also envisions extending FRAND determinations to a global scale. This situation poses a challenge because patent holders in Europe may instead opt to concentrate their patent filings in a specific country, and likely not obtain patents in Europe. Consequently, for the FRAND determinations could become inaccurate due to the exclusive reliance only on the EU SEP register and its essentiality checks (P. 37).

Neither the proposal nor the impact assessment focuses on examining worldwide FRAND determinations and making adjustments to them based on the results of other jurisdictions. An EU-only SEP Register approach raises questions, since the market for technology is global and will remain global, it is very likely that companies will resolve matters elsewhere. Furthermore, if royalty determinations under the regulation remain limited to the EU, it will fail to address the fact that disputes in this regard are often of a global nature. Moreover, the assessment has not adequately considered the potential consequences if the register were to encompass a global reach. Such an expansion would involve a substantial volume of foreign issued patents. In the absence of an established FRAND rate, it is crucial to comprehend the authority foreign courts possess in establishing these rates as part of their patent portfolios. Consequently, this scenario could also result in multiple jurisdictions asserting the right to define a global jurisdiction, leading to conflicting decisions and creating more uncertainty in the determining FRAND rates.

Alternatively, upon declaring a SEP non-essential by the EUIPO, it will be delisted from the SEP register. And if the parties do decide to enforce the patent in court later without a prior FRAND determination at EUIPO, the newly established UPC for instance or some other court, it is likely that the court may interpret the patent as essential to a standard. The withdrawal of a patent from the register on the basis of essentiality checks is more complicated than the Commission thinks, and it can lead to unpredictability in the market and in courts. Before giving the EUIPO the task to perform these essentiality checks, it would have been useful if the Commission had established a facts-based understanding of how big the problem is with SEPs that may not be essential. For instance, is declared patents that are non-essential (and thus rarely used) really a driver of revenues for SEP holders?

Additionally, these checks involve the inclusion of patents, and based on the number of standards available, each SEP owner may likely propose a significant number of SEPs for assessment. The solution to solve this through an SEP register seems somewhat ambitious based on the number of SEPs (which are measured in thousands) and SEP owners in the market. Moreover, the impact assessment has acknowledged how complex the essentiality checks will be, and how long time these processes will take and the amount of technical expertise they will require. And then, as mentioned before, if the patent is removed from the process (Article 26 (1) (d)), it will exclude it from being classified for royalty determination as well. Additionally, there is no provision for appeals in this process either (Footnote 294):

“Allowing for appeals to the assessments of randomly sampled patents produces few benefits, and significant costs – appeals are likely to correct a significant share of random assessment error, which is a relatively benign and inconsequential error for purposes of assessing firms’ relative portfolio size (false positive and false negative random errors tend to cancel each other out). Appeals are however likely to exacerbate, rather than correct over-confirmation bias, as negative assessments are significantly more likely to be appealed. By reducing random error and increasing bias, appeals make the assessments of firms’ relative portfolio sizes less reliable.

Aggregate royalty rates

For the procedure for determining the maximum aggregate royalties (preferred as the most suitable option as per the assessment), the problem is that bilateral determinations of rates may prevail (highlighted earlier), and along with practical constraints as budget and time may prove the top-down royalty determinations unfeasible. A key problem with the Commission’s approach in preferring a top-down approach is that courts across the globe have rejected it, and found it risky to provide a FRAND estimate (Judge Birss’ Decision in Unwired Patent vs. Huawei). Ironically, the Commission itself has recognised that there is an absence of information about the aggregate royalties (although it says it is a major problem in SEP licensing negotiations). However the impact assessment provides no support for conclusion of the statement. The top-down approach moreover fails to account for the bilateral agreements taking place between holders and implementers. These usually tend to have several terms other than royalties that may impact the determination of royalty rates.

Additionally, the Commission has claimed that setting an aggregate royalty “may also help to overcome problems of royalty stacking.”(P. 118). Royalty stacking and standards are issues that, in theory, become a problem when the stack becomes so high that the objective is undermined (implementation and use of the standard), which may likely derail the success of a standard. This further indicates that the burden is on SEP holders to ensure that the royalties are not set high.

However, the Commission’s analysis for FRAND royalties is asserted for a more hypothesised manner where every implementer will pay royalties to SEP owners. In the actual scenario, under-licensing is pervasive (P. 157-163 of the linked study). Another area is the lack of empirical evidence by the Commission regarding the royalty stacking problem. It says: “In any case, no matter if the problem of royalty stacking is weak or serious, an explicit aggregate royalty should help to mitigate it.” (P. 119). This is not a good basis for proposing a really radical reform. It also conflicts with the Commission’s Better Regulation Framework that promotes for transparency and suggests regulation to be carefully deployed.

Further a 2010 study highlighted in the Impact Assessment Report found that the sum of published maximum expected royalty rates, representing 60% of declared SEPs, amounted to 14.8% of the sales price (P. 45). However, in lieu to this there is no evidence that indicates that implementers actually paid the maximum expected royalty rate as indicated in the report. The question here is whether royalties for different SEP licenses actually stack up to an unreasonable royalty burden? In short, instances in literature that cite to the problem of excessive aggregate royalty burden on standards implementers provide little to no evidence for royalty stacking being a concerning problem.

Further, as part of their literature analysis, the Impact Assessment highlights the decision of the U.S. District Court in Microsoft Corp. v. Motorola, Inc. in 2012, and suggests it is “[a] clear illustration of the problem of royalty stacking (P. 118). The Impact Assessment Report quotes the court statement that “the royalty rate that implicates such clear stacking concerns cannot be a [FRAND] royalty rate” (P. 118). But a closer observation in the court case reveals that the court in the Microsoft case did not use an aggregate royalty rate to determine the FRAND rate, and adopted the comparable licenses approach. The court used aggregate royalty only as a check on FRAND rates determined in the evidence.

Similarly, in Unwired Planet vs. Huawei, the court used aggregate royalty rate determinations as a cross-check. They were overturned in TCL vs. Ericsson and dismissed in Interdigital vs. Lenovo. The primary indicator to establish whether a SEP owner has made a FRAND licensing offer has been through licensing terms of comparable licences. It is sufficient that a SEP licensor discloses the rates of comparable licenses to justify the FRAND character of a SEP licensing approach: as seen in Erricsson v D-Link.

Contrary to the approach decided, the Impact Assessment Report has itself identified numerous challenges in ascertaining a suitable combined royalty (P.90 and 119).

“[e]x-post aggregate royalty determination is difficult also because the implementing products began using the standard without knowing or taking into account the need to pay FRAND royalties,” and that for ex-ante aggregate royalty determinations “[i]t is . . . very difficult to establish a fair value for a standard before it is even developed, and to predict the success, nature and scope of implementations.”

Thus the option selected and the Commission’s assessment of it remain controversial.

Reasons for the regulation

The Commission has stated that the proposed regulation is needed because the number of SEP licensing disputes are only likely to increase. However, as identified, one study from the Commission itself suggests, that SEP licensing disputes are rare and have been decreasing over time. This also contradicts with the assessment’s assumption that any impediment to the development of a product may result in a licensing dispute. The study commissioned by the Commission and used as a basis for its impact assessment for the regulation found that SEP litigation has continued to decrease over time. It says: “In more recent years, the share of declared SEPs subject to litigation has decreased.” (P.109).

One aspect which the assessment has also failed to capture is determining whether FRAND rates will be economically efficient rates or not. The general question is whether mandating a FRAND rate determination by a conciliator panel, without any mode of appeal, at an institution which has little to no experience in SEP matters is the preferable solution? It is such a strange solution to the problems that the EU say they want to fix. If the problem is insufficient transparency and risks of delayed uses of standards, a conciliation process will not help. Indeed, as parties are still free to litigate, it is likely going to delay the introduction of new technology and products. Moreover, if the result of the conciliation process remains non-binding, a question on the viability of such an exercise or a regulation is imposed.  The Impact Assessment highlights the following (P.49):

“Reducing uncertainty about costs and legal exposure will have a positive effect on implementer-level innovation. Reasonable SEP licensing costs and reduction of negotiation delays will have a positive impact on SEP owner innovation. Both SEP owners and implementers may experience negative impact if a balance is not achieved in the determination of the FRAND terms.

However, this view is based on the assumption that licensees pay too much to obtain the SEPs license. How does the Commission know that this is correct – or that tilting the balance in favour of implementers will lead to better price efficiency? Are companies like Ericsson and Nokia running profits that are unsubstantiated and too large? Is the price of the Avanci patent pool a sign of inefficient pricing? The fact is that the Commission does not have the answers. Yet it still wants a process leading to reduced license costs.

Market players, on the other hand, seem to take a different view on the conciliation proposal. In a public consultation, 53% of the respondents considered arbitration a useful mechanism of dispute resolution and only 35% of the respondents considered mediation useful (P. 42). Still, the Impact Assessment interestingly acknowledges the following:

“Arbitration is also seen as unfit to contributing to a wider understanding of FRAND, including because the proceedings are confidential. Arbitration may not compel discovery of third parties. Arbitrators are faced with the absence of clear rules, which will result in unsatisfactory decisions that are not subject to appeal.” (P. 230)

“Unfit” according to whom, one has to ask? Despite a majority view on the issue of arbitration, the Commission proposes a solution that is less supported by market participants.

Policies that concern issues and sectors at the technological frontier should be based on firmer evidence – ideally also be data driven. The Impact Assessment and the regulation largely fail to capture that ethos: it is surprisingly short on firm evidence and data, yet still proposes a change of policy that can have extraordinary strong economic and technological consequences.

Others have noticed too. Recently, a letter was signed by many former leaders of U.S. government agencies under several presidents (Clinton, G.W. Bush, Obama, Trump, Biden) which expresses “shared concerns regarding an apparent pivot in the European Commission’s longstanding intellectual property (IP) policy that threatens European and American innovation leadership, and by extension, European and American economic success and security.” The letter says SEP policies “should be data-driven and should consider the practical impact on industry and relevant geopolitical realities.”

Indeed, the Commission should go back to the drawing board and come back with a better proposal – this time also with an Impact Assessment that is based on facts and evidence.