This article, which can be found here, proposes a novel approach for calculating FRAND royalties, based upon average total cost per patent plus a reasonable return for the patent holder. Unlike the methods discussed in the paper above – which focus on the value of a patent – this method is cost-based.
The paper is structured as follows:
An introductory section explains why standards are important and why FRAND obligations are imposed.
A significant increase in the relevance of standards can be predicted in the near future. Industry 4.0 will greatly increase the degree to which industrial processes will depend upon the exchange of information not only between people, but also between toolkits, that is, between “hardware.” The same holds true for a number of new technologies such as autonomous driving, data compression, or 3D printing.
Standard-setting organizations (SSOs) are tasked with the development and creation of standards by identifying and selecting the most suitable technologies for the standard. It goes without saying that the most suitable technologies tend to be patented. As a result of this, those patents that underlie a standard can become essential – in that these patents now do not just control access to a certain technology, but rather to the standard itself – and, thus, to important markets. This essentiality allows patent holders to act opportunistically by either refusing to license their patent altogether (thereby preventing access to the standard by implementers) or by using that increased leverage in order to charge excessive royalties.
In order to address these risks, SSOs commonly condition the incorporation of a patent into a standard on an obligation to license that patent on Fair, Reasonable and Non-Discriminatory (FRAND) terms. Perhaps unsurprisingly, an issue that has led to increasing litigation concerns is the identification of whether a licence is FRAND or not. One dimension of this question is how to calculate whether a royalty is FRAND – which is the subject of this paper.
A second section describes the main methods proposed to calculate FRAND royalties.
In the literature, one can find extensive discussions of how to determine each of the three limbs of FRAND terms (fair, reasonable, and non-discriminatory). The terms “fair” and “reasonable” are often dealt together – which may explain the absence of differences in EU and US approaches, despite SSOs in the US referring to RAND (reasonable and non-discriminatory) terms without any reference to fairness.
A number of different methods have been developed to calculate the economic value of a patent. These can be divided into main types, depending on the starting point for their calculations: i.e., output-oriented or input-oriented variables. While output-oriented variables start from the benefit derived from the patent, input-oriented variables focus on the costs associated with the patent’s creation.
Output-oriented methodologies are usually thought to be preferable whenever information on the benefit derived from the patent is available, such as net revenue achieved with a product using the patented technology. A number of different output-oriented methods to determine SEP values have been proposed by economists. Swanson and Baumol propose setting the value of a patent before the standard is set. A similar method requires patent holders to commit to price caps when declaring their patents to be standard essential. Another method requires one to fairly apportion the total amount of royalties among a variety of patent holders by reference to the relative incremental benefit contribution of individual patents to the standard; the problem with this approach is that it has such large information requirements that it is not feasible in practice.
Input-oriented methodologies focus on the cost of producing an invention and patenting it. They build upon the idea that royalties need to ensure that SEP-holders are reimbursed for their costs and receive a reasonable return on their investments, thereby ensuring that there are incentives to invent and to patent such inventions. The authors argue that this is a more practical approach, since it allows royalties to be assessed by reference to their cost. While there can be significant difficulties with identifying the adequate measure of cost, these are thought to be easier to quantify than the incremental benefit of a patent used in a product as a result of that patent having been incorporated into a standard.
A third section looks at the main challenges with deploying output-oriented methodologies in practice.
This section discusses two main limitations of the most prevalent basis for calculating FRAND royalties, net sales of a product using the patent. The first limitation stems from the complexity of technology products. If a product contains a multiplicity of patents, net sales will have to be broken down by patented functions. This can represent a major challenge, because products tend to be made up of many different components – in the case of cell phones, thousands of patented components. In such situations, the incremental contribution from an individual patent needs to be determined in order to arrive at a royalty that is based on the benefits that that patent provides. In the context of SEPs, an additional difficulty will be to quantify which parts of the benefit flow from the patent and which from the standard itself, while also identifying the contribution brought by the quality of downstream products using the standard to the final price.
A second, and related limitation concerns the very large differences in royalties that flow from different products using the same patented technology. This is a consequence of differences in prices of these products. Under such an approach, manufacturers of high quality equipment will be charged disproportionately, while manufacturers of cheaper devices would have to pay only a nominal fee. This is said to breach FRAND principles by discriminating against manufacturers of expensive equipment.
At this point, the authors discuss a recently developed alternative approach, which focuses on the Smallest Saleable Patent Practicing Unit (SSPPU). Under the SSPPU approach, royalties are based upon the smallest saleable component that includes the patented technology’s relevant functionality, thereby avoiding the problems identified above. However, this would require one to break down complex products into individual components in such a way that individual components and the functions for which customers are willing to pay can be clearly identified. Furthermore, such an approach is unable to address downstream spillover effects. Thus, it is doubtful whether the uncertainty in determining a component’s true value will be offset by additional accuracy in the determination of the incremental value contribution that a patent or patent portfolio may deliver under the SSPPU approach.
Section four reviews whether cost can provide the basis for calculating royalties.
First, the authors note that a cost-based approach is adopted in regulated sectors to address situations of market power, like that of SEP holders. This cost-based approach serves more than one purpose. Cost-based approaches are designed to simultaneously control market power, ensure non-discriminatory access and incentivise innovation. ‘Therefore, it seems rather obvious to apply rules designed for network use in the telecommunications sector for determining the level of FRAND royalties in this very sector.’
Nonetheless, cost-based approaches have thus far barely been used – without good reason. It is true that the costs of creating a patented invention are not necessarily related to the economic benefit that the invention will provide; in most cases, market potential is largely dissociated from the R&D costs incurred in developing an invention. However, correlations between cost and benefit may arise more clearly from large numbers of patents. Given intense levels of competition in the market, there likely will be a correlation between the cost of creating the patent portfolio (R&D plus patenting cost) and the patent portfolio’s economic benefits, even if a similar correlation does not arise for individual patents. Similarly, while it may be hard to determine the individual cost of developing a patent, publicly available, audited R&D costs for companies provide reliable figures that allow the determination of average costs within patent portfolios.
In section five, the authors develop a cost-based method for calculating FRAND SEP royalties.
In short, they propose identifying average patent costs in light of the costs of developing patent portfolios. To the measure of cost thus identified should be added a reasonable return ratio.
Building on the theory of capital, the authors recommend that a risk-adjusted return on capital be adopted based on a Weighted Average Cost of Capital (WACC) measure. ‘The total cost (plus a reasonable return) for a SEP are covered charging royalties for its use. Therefore, an important input variable for the cost-based approach is the number of uses of the SEP. This number affects the total amount of royalties to be paid to the patent holder over the patent term or the term of the licensing contract. Therefore, estimating these numbers is critically important.’ To estimate usage figures, market research data can be utilised that provide information on past sales figures and include a forecast of expected future sales. The overall usage figures on the patent term would then determine annual averages. Existing data can then be used to determine a royalty that meets FRAND requirements.
The cost-based approach for calculating FRAND royalties fulfils all FRAND requirements. It results in royalties that are “fair” as well as “reasonable,” because it ensures full coverage of costs for the patent holder as well as an appropriate return on his investment. However, this approach has a few weaknesses too. Firstly, it requires a clear input–output relationship between development efforts and patented technologies. Sometimes, it is difficult to find such a strong and clear relationship between patented technologies and the corresponding R&D. Secondly, it requires a somewhat reliable calculation of cost. Since R&D costs play an important role, SSOs should impose a requirement to disclose these costs in sufficient detail.
This paper is well outside my comfort zone, so I will not comment on its substance. Nonetheless, I was struck by the fact that there are two types of sources used by the authors: economic papers on methodologies to determine FRAND royalties, and competition law papers. I expected the authors to mention some IP literature on the topic, but it seems that such literature does not exist. Naturally, this is to the greater merit of the competition scholars who venture into such murky waters.
A second point regards the paper’s observation that there have been very few instances of judicial determinations of FRAND royalties. There may be good reasons for that, but I think this has started to change, as such cases are now being brought (as is apparent from the paper above).