1 Introduction

The search for balanced criteria in assessing the “fair, reasonable and non-discriminatory” (henceforth “FRAND”) licensing terms for standard-essential patents (henceforth “SEPs”) shall, first of all, focus on a number of guidance criteria which are consistent with the overall evolutionary and pro-competitive juris-political inspiration which has been recently witnessed in the EU, in the USA and also in India.Footnote 1 This is of pre-eminent importance, even vis-à-vis the precise, sophisticated calculations, such as those applied by Justice Robarts in the well-known Microsoft v. Motorola case.Footnote 2 In practice, indeed, the actual case-by-case fixation of royalties will mostly occur upon private agreements or “arbitrations” (as suggested by Lemley and Shapiro)Footnote 3 or, in default, by Courts or Competition Authorities’ individual adjudication.Footnote 4

So, consistency with the rationale of the duty to licence on FRAND terms, requests that the actual end result of the negotiation reflects an ultimately pro-competitive balance of the conflicting interest at stake, namely the holders’/licensors’ right to an appropriate remuneration—appropriate, not maximized (we borrow, as an indication of general scope, from the ECJ in Premier League)—and the ‘licensees’ right to obtain access conditions allowing them effective competitiveness.Footnote 5 This requires, first of all, the elaboration and application of basic criteria functional thereto, preferably to be established by ad hoc Guidelines of general application, and even possibly incorporated in regulatory norms—national and/or stemming from international and/or regional agreements. This is to avoid that the pure remittance to private agreements or to adjudications of mere individual scope, might translate, on the one hand, into an harlequin dress obnoxious to the need of (reliability and transparency and) harmonization of the standard-setting context, international by nature and participated by SSOs from all over the world.Footnote 6 And, on the other hand, into the subjugation of willing licensees to the superior contractual might of a licensor interested to maximize royalties, thus raising prospective rivals’ costs. Hereafter, in this chapter we submit some criteria that we deem suitable to this purpose. They are four, expressed as progressive joint (cumulative) steps.Footnote 7

2 First Step: Precise Identification of, and Fees’ Strictly Proportional to, the Technology to Be Effectively Adopted by the Willing Licensees

At times, courts and scholars have postulated that every patent that is declared essential to a standard would be implemented—and thus potentially infringed—by every product that complies with said standard.Footnote 8 This assumption seems to be grounded on the following syllogism: because (i) the patent has been declared essential to a standard; and (ii) the product is compliant with the standardized technology; therefore (iii) the product must implement the patented technology.Footnote 9

However, this assumption seems flawed by an imprecise understanding of the standard-setting rules, resulting in a fundamentally false syllogism. Indeed, not all patents that have been disclosed as essential are necessarily infringed by all products that are compliant with the standard at stake. In other words, there is no automatic infringement.Footnote 10 This is for three reasons, relating to the patent’s validity, its essentiality for the standard and, most importantly, to the existence of optional features in many standards.

As a first consideration, it is quite obvious that the patent(s) at stake may be invalid or non-essential. On the one side, the number of effectively “strong” patents, meaning those that would survive an invalidity attack is notoriously low. Where some have suggested that, within some jurisdictions, the majority of granted patents would be at least partially invalid,Footnote 11 the available data on SEPs would confirm this to be an exacerbated problem in the standard-setting context.Footnote 12 On the other side, the essentiality disclosure that SEP holders file before SSOs is a wholly unilateral act, which in most cases is not substantively examined nor weighed in any way by the organization. For instance, the ETSI Guide on Intellectual Property Rights states that, “ETSI has not checked the validity of the information, nor the relevance of the identified patents/patent applications to the ETSI standards and cannot confirm, or deny, that the patents/patent applications are, in fact, essential, or potentially essential”.Footnote 13 Because of this, the so-called over-disclosure phenomenon—i.e. claiming essentiality for a non-essential patent—is considered to be wide-spread.Footnote 14

However, both arguments do not appear very useful for the purposes of determining a FRAND rate: granted patents are presumptively valid and, to our knowledge, no major SSO is involved in a substantive examination of the disclosed patent’s essentiality.

Conversely, in order to establish general principles on how FRAND royalties must be assessed, it is best to focus on the intrinsic characteristics of technological standards, especially in the ICT sector.

Standards are complex sets of rules. They are continuously updated and amended by working groups and dedicated experts. For instance, the well-known 3G/UMTS standard for telecommunications was improved by way of several subsequent “releases” and accounts for hundreds of technical specifications (i.e. analytical documents addressing one specific function of the system). In turn, technical specifications are often modified and published, usually more than once for every release of the standard.Footnote 15 As a result of such relentless activity, standards often end up comprising both progressively added and “optional” technical features.Footnote 16

The technical features (and specification) of a standard may be “optional” in relation to a specific product implementing the said standard in three different ways.Footnote 17 In the first place, technical features often concern just one of the elements of the standardized system. For instance, the portions of the standard relating to mobile devices may not regulate the functioning of mobile infrastructures, such as the network itself, and vice versa. In the second place, a technical feature which is added in a specific version of the standard, released after certain products have been placed on the market, will not be (necessarily) implemented by those earlier products. In the third place, a feature of the standard may be optional in the sense that the producer can freely decide, from the outset, whether to implement such technology or not, without affecting the product’s interoperability.

If a specific feature of the standard is indeed optional, any patent that is declared as essential for that feature is only optionally implemented by the competitor’s products placed on the market. Therefore, in all the above mentioned cases, the infringement of the SEP covering said optional technical features is merely potential. These considerations cannot be ignored in an attempt to determine a FRAND royalty. If a product does not implement a certain patented technology, why would it need to remunerate the patent owner for it? This nevertheless, the issue of optionality is often disregarded in real-world negotiations.

Besides, it is generally understood that a feature’s optionality, as of itself, does not hinder the possibility for the relative patent to be sensu stricto essential. The patent at stake may well be essential, but only in relation to such optional feature. This is expressly confirmed, for instance, in the Standard Board Bylaws of the IEEE which convey that: “‘Essential Patent Claim’ shall mean any Patent Claim the practice of which was necessary to implement either a mandatory or optional portion of a normative clause of the IEEE Standard”.Footnote 18 According to a 2013 study, seven of the most important SSOs include optional portions of the standard in the definition of essentiality.Footnote 19

All this considered, we suggest that a first step for determining any FRAND royalty is to abide by a principle of strict proportion between the patented subject-matter and the standardized technology that is used by the licensee.

Only when the interference is effective, so that the essential patent is or will be effectively infringed by the competitor’s product, the implementer shall be burdened by the obligation to pay FRAND royalties to the SEP holder. Symmetrically, it is only when the SEP is effectively implemented or infringed that the compensation for its use can be deemed fair and reasonable. This evaluation shall be made prior to any request for compensation by the SEP holder itself, as he is ostensibly the only subject with sufficient knowledge of the patents and standard at stake.

More specifically, at the beginning of the negotiations, the SEP holder should: (i) provide to the potential licensee claim-charts (or similar documentation), showing how the patent can be read on the technical specification of the standard; (ii) assess whether those technical specifications constitute mandatory or optional portions of the standard for the products at stake; and (iii) in case of optional portions, the SEP holder should also show that the products at stake effectively implement those portions of the standard.Footnote 20 In the latter case, however, the SEP holder shall not provide full evidence of the effective implementation for the purposes of negotiating a license, as this would likely constitute an excessive burden for him; prima facie evidence would amount to a sufficient basis for the negotiation stage. Potential licensee may then be in the position to rebut such evidence, showing that the patent(s) at stake would not be effectively implemented within its products.

3 Second Step: Royalties Determination Ex ante, i.e., Taking into Account the Value of the Patent Prior to the Standard Setting

Once the standard-essential technology to be licensed has been circumscribed, the value of said technology—more specifically: the value of the patents covering said technology—must be assessed for the purposes of determining the FRAND royalty.

The standard selection process plays a fundamental role in the creation of value. Prior to the adoption of a standard, several technological alternatives for the same functionality usually compete as to which will be elected within the standard. In the commonly typified situation, different technological solutions will be weighed against each other by the SSO’s members on the basis of their quality, their price and their added-value to the standard, in order to determine which will better fit the industry’s needs.

However, when the standard is finally adopted, the owners of the IP rights covering the chosen technologies will see a dramatic increase of their patent’s market power. This is because the SEPs will be implemented—within the optionality limits explained above—by every manufacturer within that sector, potentially allowing the SEP holder to extract revenues and even to “hold-up” competitors with requests for supra-competitive fees, paired up with the threat of an injunction.Footnote 21 Conversely, those patents that have not been adopted may turn out to be worthless, especially if there is no market for those technologies other than the standardized one.Footnote 22

Against this background, the vast majority of scholars, economists and courts seem to agree on that a “reasonable” royalty should reflect only the value of the patent qua patent, and not the value potentially associated to its inclusion in the standard.Footnote 23 Swanson and Baumol were among the first scholars to suggest, in an influential article published in 2005, that “the concept of a ‘reasonable’ royalty for purposes of RAND licensing must be defined and implemented by reference to ex ante competition, i.e. competition in advance of the standard selection”.Footnote 24 A few years later, Lemley and Shapiro conveyed that “[b]y construction, the reasonable royalty rate does not include the value attaching to the creation and adoption of the standard itself. To allow the patentees to capture that value, which flows from the collective adoption decisions of the group rather than from the underlying value of the technology chosen, would undermine the goals of the FRAND commitment”.Footnote 25 Similar ex ante approaches have been endorsed by the Federal Trade Commission,Footnote 26 the European CommissionFootnote 27 and, most recently, within the IEEE Bylaws.Footnote 28

The ex ante approach provides a first benchmark for what constitutes a FRAND royalty. In simple terms, the ex ante approach points to the reasonable royalty that the SEP holder could have obtained in an arms-length hypothetical negotiation with the prospective implementer just before the standard was adopted, whereas the adoption of the standard flags the ex post moment, when the patent has been included in the standard, thus potentially gaining substantial added-value on the market (when not a dominant one). The underlying assumption is, of course, that only an ex ante royalty would reflect the intrinsic value of the patent in a competitive environment.Footnote 29

From our perspective, the ex ante approach constitutes a fundamental criteria for determining a “reasonable” licensing rate. Given the underlying principle that, in order to be pro-competitive, the FRAND royalty must also be appropriate and not maximised, it seems rather logical to cap the reasonable royalty to the intrinsic value of the patent before it acquires inevitable (and possibly “hold-up”) power following its inclusion in the standard.

However, this rule is to be applied ultimately when and if, given the specific circumstances of the case, there are (or were) available alternative and comparable technologies providing similar added-value to the standard, which may be used for determining the effective ex ante value of the patent. For the same purposes, one may also consider whether licensees for that same specific technology had been stipulated prior to the election within the standard.

4 Third Step: Looking at the Overall Licensing Scenario and Royalty Stacking Issues

The determination of the license fees should also take into account the overall licenses’ scenario that may encumber potential licensees. This should lead to reckon with the fact that in the standard-setting context, in particular within the ICT sector, hundreds of patents may insist on a single final product, so that the implementers are normally obliged to pay royalties to multiple SEP holders. Ignoring this problem might determine a disproportionate “royalty stacking”, potentially exceeding a reasonable portion of the product’s value and price—hence crippling its market and/or discouraging the producer from the very adoption of the standard.

As a matter of fact, given that the sheer number of SEPs in the ICT sector and, more generally, in the electronics industry is very high,Footnote 30 many have argued there are serious chances for implementers to be burdened by a burdensome “stack” of royalty demands. For instance, an empirical research conducted in 2014 came to the conclusion that the potential royalty demand over a $400 smartphone would amount to a $120 stack, i.e. to 30% of the end product’s price and almost equal to the cost of the device’s components.Footnote 31

An occurrence of “royalty stacking” clashes with the very concept of FRAND, as it cannot be deemed to be fair, nor reasonable, for the aggregate licensing fees to make “commercialisation of products compliant to the standard uneconomical or unprofitable”.Footnote 32 Lemley and Shapiro add that the “royalty stacking” existence would exacerbate the hold-up problem, by multiplying the chances for an implementer to face infringement claims and supra-competitive price demands.Footnote 33 Obviously, the possibility of a “royalty stacking” is not dependent from the licence stipulated by the individual SEP holder, but rather from the aggregate amount of all SEP-related licenses. A number of decisions rendered in the US have recognized the potential harm caused by royalty stacking issues. In the 2013 Microsoft v. Motorola case, Judge Robart famously held that “a proper methodology for determining a RAND royalty should address the risk of royalty stacking by considering the aggregate royalties that would apply if other SEP holders made royalty demands of the implementer”, and that the potential for royalty stacking should be taken into account by SEP holders when setting prices.Footnote 34 Similar opinions have been rendered by the EU Commission,Footnote 35 the FTCFootnote 36 and, quite recently, by the Competition Commission of India (CCI), which pointed out to the fact that “FRAND licenses are primarily intended to prevent patent hold-up and royalty stacking”.Footnote 37

However, legal scholars and economists have raised substantial objections against the royalty stacking theory. Many argued, for instance, that it fails to duly take into account the “reductionist” impact of common practices such as cross-licensing, the possible non-exertion of patent rights and the real-world market dynamics (which would punish those companies that set excessive royalties). What is more, several authors underlined that there would be no empirical evidence of a royalty stacking phenomenon, at least enough to cause any kind of serious public concern.Footnote 38 Taking a stance in this debate, the Federal Circuit in Ericsson v. D-Link assessed that, while “royalty stacking” may be a potential problem posed by SEPs, in order for it to be weighed the defendants had to present evidence of an “actual” royalty stack, which could not be simply presumed.Footnote 39

All in all, the existence and the issues possibly raised by royalty stacking, though much discussed, did not gather substantial consensus among academics and practitioners. On the one side, common sense would suggest that if hundreds, if not thousands of patents, insist on a single product, the royalty stacking is—to say the least—likely. On the other side, there may be cases where the potential for a royalty stack is neutralized (by way of cross-licensing, lowered prices, patent exhaustion, and so on) or absent.

Within this context, we hold that an occurrence of royalty stacking is nevertheless irreconcilable with FRAND requirements in any case where the expected outcome is a licensing fee that minimizes or eliminates any profit on the standard-implementing product. In order to neutralize such risk, negotiating parties must take the overall licensing scenario into account and rely on the available data when setting the royalty fees: SEP negotiations do not take place in a vacuum. Besides, most standards have been in place for years now and even when upgraded, they are usually improved gradually, within time. Therefore, parties attempting to negotiate a FRAND license will likely have some degree of understanding of the relevant licensing scenario, which—may that be the case—could be enough to weigh in a certain degree of stacking. On the contrary, if no information on the overall context is available, the parties may discuss a prima facie FRAND license, possibly coupled with a re-adjustment clause over the fee in case of changes in circumstances.Footnote 40

These first considerations shall be backed up by regulatory and/or normative provisions, whereby the importance of referring to the overall scenario for the purposes of assessing FRAND royalties would be acknowledged.

Once again, the IEEE bylaws constitute a guidance example in this regard, as they suggest that the reasonable rate’s determination shall include the consideration of “[t]he value that the Essential Patent Claim contributes […] in light of the value contributed by all Essential Patent Claims for the same IEEE standard practiced in that Compliant Implementation”.Footnote 41 Similar instances should be implemented by other SSOs, under the Competition Authorities’ guidelines and/or, eventually, by amending multilateral international treaties (such as: TRIPs Agreement).

5 Fourth Step: Dynamic Approach to FRAND Royalties’ Determination

Finally, we consider that FRAND royalty rates should also be accounted for in a dynamic and innovation-oriented perspective. Innovation is, by definition, rooted into the ICT sector. Where new products are constantly introduced into the market, the standardization process is characterized by an ever-evolving nature. This is clearly reflected also by the incessant patenting activity which takes place on most technical features relating to each standard in the sector.

Therefore, it is here suggested that any license negotiation concerning SEPs needs to take into account the dynamic evolution of ICT products, standards and patents in order to be compliant with FRAND terms. The underlying idea is that FRAND terms are to be determined in relation to the overall licensing scenario, as expressed in the previous part. Moreover, it is herein submitted that, because the licensing scenario is subject to rapid changes, the line between what is FRAND and what is non-FRAND may shift in time.

For instance, if “new” SEPs are discovered as being implemented by certain product(s)—whereby the parties did not take into account said rights when assessing the FRAND licensing rate either because these rights were not published or otherwise ignoredFootnote 42—it may be necessary to modify the negotiated royalties in the light the value and FRAND fees of the “new” SEPs.Footnote 43

Besides, if the licensing agreement is drafted in broad terms, generally encompassing all the licensee’s products (as it is usually the case for FRAND agreements), the FRAND rates may need to be adjusted to take into account the evolution of the standard. A license that the parties negotiated for all implementing products in 2014 may be compliant to FRAND obligations and yet, in consideration of major evolution both in the standard and patenting landscape which took place in 2016, become excessively onerous for all newly released products.Footnote 44

Furthermore, the overall licensing scenario may also be appreciated from a “subjective” point of view. As a matter of fact, if the SEP holder has established a FRAND rate that adequately remunerates its inventive efforts in relation to a certain number of licensees (say, 5), the same rate may end up over-compensating him, in case the number of licensees grows (say, 25).

In conclusion, it is here submitted that FRAND terms shall be read as providing for a permanent adjustment mechanism, whereby changes and modifications in the technological and proprietary scenario are duly taken into account and calibrated on the industry dynamics. The said mechanism might well be contractually agreed upon—thus foreseeable from the start of negotiations. It might also involve in default, ADRs or formally arbitral resolution under pre-definite criteria.

6 Conclusion

Might a wide consensus on these principles, and others of corresponding proportionate pro-competitive inspiration, be reached, the optimal seat for their embodiment would be by international or regional agreements (e.g. by addenda to Article 31 TRIPs or to the ASEAN treaty) or by ad hoc EU Horizontal Directive or EU Commission Guidelines. By default, each country should incorporate them in ad hoc Guidelines, entrusting their application to Judiciary Courts or Competition Authorities—whichever deemed more experienced and sophisticated in dealing with IP licensing disputes.