• Embed Doc
  • Copy Link
  • Readcast
  • Collections
  • CommentGo Back
Download
 
“Essential” Patents, FRAND Royalties andTechnological Standards
∗
Mathias Dewatripont and Patrick Legros
†
February 5, 2009
Abstract
We abandon the usual assumption that patents bring known benefits to the industry orthat their benefits are known to all parties. When royalty payments are increasing in one’spatentportfolio, asisimplicitlythecaseinFRANDagreements, privateinformationaboutthequalityofpatentsleadstoavarietyofdistortions, inparticulartheincentivesoffirmsto“pad” by contributing inessential patents. Three main results that emerge from the analysis are that:(i) the threat of court disputes reduces incentives to pad but at the cost of lower production of strong patents; (ii) mitigating this undesirable side-effect calls for simultaneously increasingthe cost of padding, that is, for better filtering of patent applications; (iii) upstream firmshave more incentives to pad than vertically-integrated firms which internalize the fact thatpatent proliferation raises the share of profits going to the upstream segment of the industry but at the expense of its downstream segment. This seems consistent with recent evidenceconcerning padding.
1 Introduction
While firms try to differentiate themselves from competitors, they also benefit from havingstandards established: this facilitates in particular the access of consumers to other providers’consumers and allows economies of scale in the production of various inputs (e.g. chipsets,other electronic parts, etc). At the same time, once a standard is established in the industry, it becomes very costly for a single firm to produce a non-standard good; it may then be vulnerableto ex-post opportunism by the holders of essential technologies embodied in the standard. This
∗
We thank Mathias Roende, Pekka Saaskilahti and Jean Tirole for useful comments on earlier drafts. We are alsograteful to the Nokia corporation for financial support and for useful discussions. The views expressed in this paperare however solely our own.
†
ECARES, Universit´e Libre de Bruxelles and CEPR.
1
 
2problem is particularly important when the creation of a standard requires the use of manydifferent innovations (in the case of mobile telephony for instance, a handset can require theuse of more than one thousand technologies protected by patents).Because the anticipation of opportunistic behavior may discourage participation in SSOs,the industry has designed rules of conduct to limit the possibilities of hold-up, and thereforeincrease the willingness to participate in the development of a standard. FRAND is a lead-ing example of such rules. In a FRAND agreement, firms agree to contribute to the standardall the patents that are essential and to settle on royalties that are “fair, reasonable and non-discriminatory” (note that FRAND is the European terminology, while in the US one talks aboutRAND, that is, about “reasonable” rather than “fair and reasonable”). Note that when a patentis deemed essential for a technology, it means that a firm cannot legally produce this technologywithout paying the royalties associated to this patent.Our objective in this paper is to analyze how the nondiscriminatory and fairness dimensionsof FRAND affect the quality of the patents in a standard. Because the literature on patents
1
fo-cusesmainlyontheirstrategicuseinmarketsettings,wecannotdirectlyrelyonittounderstandthe effects of cooperative agreeements like FRAND, especially if the focus is on the quality of the patents that will be considered essential for the standard.
2
A noted exception is the literature on patent pools.
3
Patent pools presume that firms agreeon a well defined set of patents to share (and verify carefully the legal validity of the patentsin the pool), and licensors can use the patents in the pool to develop their own technology. Bycontrast, in SSOs, the objective is to develop a common technology. The process of formationof the standard is dynamic and members (patent holders, producers of final goods and oftennetwork operators) define along the way which technologies are needed and identify whichexisting patents are essential for these technologies. In addition to the usual legal validity of the patent, there is also the issue of whether the patent is truly essential – from a technical pointof view – for developing the standard. These specificities of SSOs introduce three importantdimensions in the problem of licensing.
1
In the following we will use “patent” and “patent family” interchangeably and assume that patent families havefull geographical coverage.
2
For empirical work on SSOs, see the analysis of Chiao et al. (2006), or Rysmann and Simcoe (2007).
3
See for example Shapiro (2001), Aoki and Nagaoka (2004), Lerner and Tirole (2004) and Layne-Farrar and Lerner(2006).
 
3First, the process of verification of essentiality of the patents is more difficult than in patentpools since the technology against which essentiality is verified is evolving. Rysmann and Sim-coe (2007) show that between 1990 and 2005 the number of disclosures within some of the mainSSOs (ANSI, IEEE, IETF, ITU) has increased significantly (sometimes by a factor of five). Thesheer number of disclosures makes verifiability of essentiality claims difficult. In addition, thereis also a lot of uncertainty as to which patent applications will in fact be granted, due in par-ticular to randomness of the work of patent offices (see the book by Jaffe and Lerner (2004) onthe shortcomings of the US patent system; see also Guellec and van Pottelsberghe (2007) on theEuropean patent system), and to which patents will be upheld in courts.Second, the probability that a given patent is essential or not is endogenous. This is becausethe definition of the standard will make some patents technically essential while they wouldnot have been so in another standard. Or because firms are free to contribute patents to thestandard and claiming that they are essential, when they are not. For instance, in a recentpaper Goodman and Myers (2005) show that up to 80% of the patents that firms claimed to be “essential” for a mobile telephone standard were possibly not essential.
4
The tables belowsummarize their findings for two third-generation standards for mobile telephony
5
(‘D’ denotespatents that were declared essential by the firms and ‘J’ those which were judged essential bythe experts.)3GPP D JQualcomm 279 30Ericsson 129 34Nokia 94 40Motorola 38 113GPP2 D JQualcomm 340 54Ericsson 16 3Nokia 45 14Motorola 37 14Finally, since there is a lot of uncertainty about the quality of the standard that will be even-tually produced, many royalties are negotiated ex-post in SSOs. This makes the problem of hold-up even more crucial than for other licensing environments since once a standard is de-veloped and adopted by a fraction of the industry, the opportunity cost of not adopting the
4
This evaluation was made from a purely technical point of view, that is ignoring the possibility for these patents to be infringing other patents. On average, an expert spent one hour per patent to evaluate its essentiality.
5
These two standards refer to two “Third Generation Partnership Projects”. 3GPP is concerned with descendentsof the European GSM standard, while 3GPP2 is concerned with advanced versions of the original US-based CDMAcellular system.
of 00

Leave a Comment

You must be to leave a comment.
Submit
Characters: ...
You must be to leave a comment.
Submit
Characters: ...