How to apply non-discrimination commitments Standardization and Competition Policy

Transcription

How to apply non-discrimination commitments Standardization and Competition Policy
How to apply non-discrimination commitments
A case study on FRAND Patent Licensing,
Standardization and Competition Policy
This article explores the implications for Competition Policy of non-discrimination commitments
in standardization procedures. We operate an economic analysis of non-discrimination
commitments, exploring why rational agents commit, what these commitments do imply and how
they could be used efficiently. In a case study we assess complaints about alleged discriminatory
and exclusionary pricing from optical disc replicators against patent pools licensing the patents
for MPEG and DVD standards. We evaluate these licensing policies against the growing
literature on FRAND commitments and against economic theory of discriminatory pricing.
Technological standards are everywhere in our daily life. The http standard makes sure that our
PC can open any site in the web, GSM and UMTS standards ensure that all cell phones can
communicate, and CD and DVD standards guarantee that players and decoders can read all discs
we can find in the retail. These are only few examples from the ten thousands of standards that
govern technology-intensive industries. As little aware as consumers are of the importance of
standards, as little do they usually know about the complex corporate and public policy questions
relating to standard setting. Only in recent years an upsurge in litigation about licensing of
Intellectual Property Rights protecting components of standards has started to attract wide
attention.
Standardization is an important challenge especially for Antitrust Policy. Standard setting is a
cooperative agreement between competing firms. Many standards are open, i.e. free of
Intellectual Property Rights (IP), but most standards cover essential patents. Essential patents are
patents that are necessarily infringed by any user of the standard. Any firm that holds an essential
patent can block the adoption of the whole standard by any user. Standardization is thus not only
of concern to Competition Policy because it implies direct coordination between competitors on
sensitive issues, but also because it leverages market power of Intellectual Property Rights and
may entail exclusionary effects1.
Whereas Intellectual Property is intended to allow its owner excluding others from using the
protected invention, the main objective of standards is to encourage the spread and wide
implementation of the standardized technology. This apparent conflict is resolved by licensing.
Standardization participants are expected to allow others the use of their technology, but they can
require adequate royalties. Standard setting organizations (SSO) often require firms participating
1
Francois Lévêque, Yann Ménière: Technology Standards, Patents, and Antitrust; Competition and Regulation in
Network Industries, Volume 9 (2008), No. 1
at standardization exercises to disclose any patent that might turn out to be essential for the
standard that is to be commonly set. Furthermore holders of such patents have to submit a
declaration whether they accept to commit on fair, reasonable and non-discriminatory terms of
licensing for these patents (FRAND commitments). If a firm discloses a patent and refuses to
commit on such licensing terms, the SSO will usually set the standard such as not to include the
protected technology.2 Even though standardization very often goes along with a massive amount
of licensing, the rules for licensing of complementary patents essential for a common standard are
often unclear and can be subject to complex discussions. Nevertheless, FRAND commitments are
commonly seen as important instrument to curb anticompetitive and abusive strategies. Also
antitrust authorities have referred to FRAND commitments as remedy to the potential
competitive risks of standardization.
In view of the importance of FRAND commitments for Competition Policy, the attention given
so far to the non-discrimination part of FRAND seems to be insufficient. An important body of
academic research interprets FRAND commitments as restricting the maximum height of royalty
requests and attaches little attention to examining whether pricing is non-discriminatory among
the licensees. This general line of interpretation is somehow uncomfortable for antitrust scholars,
as Competition Policy, especially when dealing with patents, should not aim at restricting profits,
but at guaranteeing a level playing field for competition. Indeed we are little convinced of
theoretical arguments why the non-discrimination part of FRAND commitments should not be
taken seriously. We run a case study of optical disc replication, where antitrust litigation and
complaints issued by business association point to alleged discriminatory patent pricing. There
appears to be an important potential for improving the conditions of IP licensing and
guaranteeing pro-competitive standardization by a more far-reaching interpretation of nondiscrimination commitments. We find that in order to be of help, these commitments should be
understood as meaningful instruments of choice, not being subject to a general requirement in
standardization, but effectively enforceable and costly, i.e. restricting the freedom of the
committing firm beyond general antitrust law.
1.
Traditional antitrust concerns with standardization focused on the risk of anticompetitive
collusion, authorities being afraid that standardization participants restrict competition among
2
See e.g. ETSI Rules of Procedure, 26 November 2008; § 6.1 and 6.3:
6.1 When an ESSENTIAL IPR relating to a particular STANDARD or TECHNICAL SPE CIFICATION is brought to
the attention of ETSI, the Director-General of ETSI shall immediately request theowner to give within three months
an irrevocable undertaking in writing that it is prepared to grant irrevocable licences on fair, reasonable and nondiscriminatory terms and conditions under such IPR to at least the following extent:
● MANUFACTURE, including the right to make or have made customized components and sub-systems to the
licensee's own design for use in MANUFACTURE;
● sell, lease, or otherwise dispose of EQUIPMENT so MANUFACTURED;
● repair, use, or operate EQUIPMENT; and
● use METHODS.
6.3 As long as the requested undertaking of the IPR owner is not granted, the COMMITTEE Chairmen should, if
appropriate, in consultation with the ETSI Secretariat use their judgment as to whether or not the COMMITTEE
should suspend work on the relevant parts of the STANDARD or TECHNICAL SPECIFICATION until the matter has
been resolved and/or submit for approval any relevant STANDARD or TECHNICAL SPECIFICATION.
them or market entry by outsiders.3
In order to deal with these concerns, antitrust authorities have long exercised pressure to restrict
coordination inside SSO, excluding especially agreements on remuneration of intellectual
property rights. Such restrictions have proven to generate a new kind of problem. As patents
incorporated into a standard are strictly complementary, any single patent user, no matter how
small his contribution to the standard may be, can block implementation of the standard by any
user. In the absence of binding rules and coordination mechanisms, a non-cooperative patent
holder can thus sack exorbitant royalties out of any proportion with the value of his own
technological contribution to the standard. Even more worrisome for Competition Policy, a patent
holder who is also active on the downstream production market can effectively exclude his
competitors from the use of the standard by refusing to grant viable licensing conditions.
It is important to stress the interest that patent owners have themselves in credible rules for IP
licensing. Non-cooperative strategies slow down the diffusion of the standard and reduce the
overall income that the other patent holders can perceive on royalties or product sales.
Furthermore patent owners may find it difficult to get their patents included into a standard or to
encourage potential users to rapidly implement the standard if they can not credibly promise to
refrain from non-cooperative strategies.
Therefore SSOs have come up with rules and procedures including commitment policies, and
most importantly requirements of FRAND commitments. These commitment policies have often
been endorsed by patent holders, as they appear to be flexible while providing potential adopters
with sufficient guarantees to encourage standard adoption before licensing policies are set.
But problems remain, and important conflicts have ended up in formal litigation and enquiries by
antitrust authorities. At least since the Qualcomm and Rambus cases, the issue of IP licensing in
standardization contexts has been recognized as independent problem for Competition Policy.
Antitrust authorities have since very much softened their approach to far-reaching business
coordination among patent owners inside and outside standard setting bodies. The favorable
business reviews of the MPEGLA patent pool by the DoJ and the EC in 1997 have set a
precedent for a joint licensing model that has since been adopted for many other industry
standards without ever giving rise to charges for illicit price fixing. Whereas negotiations on joint
licensing very generally take place outside standard setting bodies and come to an end up to
several years after standards are set, standard setting bodies no longer refrain from dealing
actively with Intellectual Property issues. To give only some examples: the European DVB
project at ETSI has a policy of actively promoting the creation of IP pools, at least one of which
is managed by ETSI itself. The American IEEE has signed a general partnership agreement with
the patent pool manager Via Licensing that should help fostering rapid and smooth patent pool
creation.
It is a main source of problems that standards are set before the concrete licensing terms for
patents are disclosed. To deal with these problems, explicit commitments are a promising and
logical strategy. Standard setting bodies such as VITA and once again IEEE have come up with
3
For an analysis of these „early“ antitrust concerns with setting of compatibility standards, see Carl Shapiro; Setting
Compatibility Standards: Cooperation or Collusion?; June 2000
far-reaching policies for early commitments, coming close to ex ante price setting4. Antitrust
authorities seem to monitor this evolution closely, while viewing overall positively such industry
coordination.5 It appears that these policies of more explicit ex ante commitments have so far
concentrated on price levels and neglected the thorny issues of price structure.
But FRAND or more explicit commitments are not only a matter for Competition Policy in the
sense that they had to be acknowledged as licit under Competition Law. As they are crucial for
the success of eventual coordination, they have been especially endorsed by antitrust authorities.
In several business review or comfort letters directed to patent pools, European and American
competition authorities conditioned favorable business reviews or comfort letters for specific
licensing programs on patent holders’ FRAND commitments. This can be seen very clearly from
the Business Review letter sent in June 1999 to Toshiba as the joint licensor of the DVD 6C
patent pool. Referring to RAND commitments Toshiba submitted as a patent holder participating
in standardization, the DoJ states that “because Toshiba, the joint licensor, must license on a nondiscriminatory basis to all interested parties, it cannot impose disadvantageous terms on
competitors, let alone refuse to license to them altogether.” Furthermore, “although the meaning
of "reasonable" is open to various interpretations, each Licensor's commitment to license its
"essential" patents independently of the pool on reasonable, non-discriminatory terms may
further ensure that the proposed program facilitates, rather than forecloses, access.”6
For some authors, the requirement of FRAND licensing of patents reading on standards evolves
to a principle of common law7. It is not straightforward that FRAND should be a generally
applicable principle for licensing in all types of standardization exercises, but where a firm
commits on FRAND licensing, the obligation arising thereof can very well be a matter of
Competition Policy. Especially where antitrust authorities took note of FRAND commitments
while reviewing problematic business practices, they are well placed to require execution of the
commitments.
Diverging interpretations of FRAND commitments are an important source of conflict. Very
important cases of alleged breach of FRAND commitments have motivated a considerable
amount of academic research, intended to give a concrete meaning especially to the terms “fair”
and “reasonable”.8 The academic interest has concentrated on definitions of reasonable terms of
4
See VITA IPR Policy at
http://www.vita.com/disclosure/VITA%20Patent%20Policy%20section%2010%20draft.pdf; and IEEE patent bylaws
at http://standards.ieee.org/guides/bylaws/sect6-7.html#6
5
« Standards bodies can very often require disclosure [of ex ante royalty rate] without fear of competition law
intervention » Neelie Kroes, European Commissioner for Competition Policy Being Open about Standards, Open Forum Europe, Brussels, 10th June 2008, available at
http://europa.eu/rapid/pressReleasesAction.do?reference=SPEECH/08/317&format=HTML&aged=0&language=EN
&guiLangu(s ; Per Hellstrom, DG COMP official responsible for standardization issues, evoked speaking in personal
capacity the idea that „prices would be announced and or set before the standard is agreed, for example by auction.”
This “would allow competition on both price and the technology.” Quoted from MLex, EC’s Hellstrom flags benefits
of ex-ante standards setting regime, 11 September 2009
6
Joel Klein, US Department of Justice: Letter to Carey Ramos in response to request on behalf of Hitachi,
Matsushita, Mitsubishi, Time Warner, Toshiba and Victor; June 10, 1999
7
Daniel Crane; Patent Pools, RAND Commitments, and the Problematics of Price Discrimination; Benjamin N.
Cardozo School of Law Working Paper; April 2008
8
For an overview over the Rambus and Qualcomm cases and the academic literature originating from them, see
www.iprstrust.org
licenses, understood as the maximum amount of royalties a patent holder can ask. But many
complaints about anticompetitive effects of standardization and cases of antitrust litigation focus
on the structure of pricing, i.e. on allegations that specific licensing schemes put some licensees
at a competitive disadvantage with respect to their direct competitors. In several industries, but
especially for consumer electronics and mobile telephony, holders of essential patents compete
themselves on the production market against independent producers who do not hold patents.
Independent producers accuse patent holders to take profit of standardization and cross licensing
or joined licensing schemes to exclude their downstream competitors.
Two industries are characterized by especially complicated relationships between integrated
patent owners and independent licensees who wish to compete with them on the downstream
manufacturing market. In the mobile telephony industry, the setting of the GSM standard has
allegedly had exclusionary effects on the industry of handset manufacturers9. Business
associations and academic scholars have pointed to the patent owners’ cross licensing policies as
factor that has contributed to such exclusionary effects. In the optical disc industry, complaints
have been raised and several of them have ended in formal litigation and antitrust inquiries at the
occasion of the setting of CD, DVD or BluRay standards10. While the competitive effects of IP
licensing are an important issue for disc or decoder manufacturers, the share of IP royalties in
production costs is so high in the optical disc replication industry that the effects of IP pricing
structures on replicators are direct and drastic. In both the mobile telephony industry, antitrust
authorities and courts have been asked to evaluate ex post the lawfulness of licensing schemes, as
ex ante commitments appear to have failed in yielding pro-competitive conditions. Especially the
use of FRAND commitments could not avoid eventual litigation on alleged breaches of antitrust
non-discrimination rules.
We argue that this failure is due to the unclear status of the non-discrimination part of FRAND
commitments. Standard users selling products reading on the standard to their customers are
vulnerable to discriminatory patent pricing much more than to generally excessive royalties. High
or even excessive royalties can be passed on to the consumer as long as all competitors have to
bear the same IP costs and the demand is sufficiently incompressible. A generally high level of
prices is only problematic if conditions are unequal among manufacturers, as it is then a levy of
the problems generated by differential pricing. Contrary to what is sometimes asserted, the notion
of non-discriminatory or licensing is not self-explaining, and gives rise to as many divergent
interpretations and conflicts as the notion of reasonable licensing.
Many authors indeed suggest that a very plain definition of non-discrimination is possible and
sufficient. J.S. Miller quotes what he refers to as consensual view stating that “the
nondiscrimination part of the (FRAND) promise is straightforward, requiring that participants
9
Position Paper of the International Telecommunications Standards User Group : The GSM Standards, IPR and
Licensing (An Example of the Restrictive Effects of Standardization), December 1998 ; for a balanced academic
analysis see Rudi Bekkers, Bart Verspagen and Jan Smits : Intellectual property rights and standardization: the case
of GSM ; Telecommunications Policy 26, 2002
10
See e.g MPEGLA vs ODS, Landgericht Düsseldorf, Urteil vom 11. Januar 2007, Az. 4a O 343/05. The court
ruling can be consulted following the link http://www.duesseldorfer-entscheidungen.de/?q=book/export/html/509.
We have drawn many insights in the complaints of independent optical disc replicators that have given rise to the
formal complaint of the Independent Optical Disc Replicator Association (IODRA) before the European
Commission in 2005 (still pending) from direkt discussion with Guy Marriott, chairman of IODRA.
license similarly situated adopters on the same terms”. While this definition is intended to
quickly resolve the issue of non-discrimination, it rather underlines how problematic the notion
of non-discrimination is. The whole problem of interpretation lies in the meaning one gives to the
notion of “similarly situated licensees”. Pricing can be discriminatory under US Robinson
Patman Act or EC Treaty Article 82 when different licensees are offered different terms of
contract or when general terms of contract are “functionally available” to some, but not to all
licensees. In our case study and in a general comparison of licensing practices of patent pools
managing essential patents we have revealed many examples of pricing schemes that result in
different licensees having a different production cost due to differential royalty requirements.
A related though different strand of reasoning refuses to get into interpreting what the nondiscrimination promise exactly means, as discriminatory IP licensing is seen to make either no
business sense to patent holders, or to systematically enhance efficiency and consumer welfare.
The first argument refers to a general lesson from textbook Industrial Organization that “two
monopolies are no better than one”. As a patent holder holds a lawful monopoly on his intangible
input, he should be able to capture the whole monopoly rent without having recourse to affiliates
on the manufacture market that is strictly complementary to the technology input. Patent holders
should therefore have no interest in interfering with competition on the manufacturing market, as
vigorous competition drives down manufacturing prices to production costs of the most efficient
and thereby maximizes the monopoly rent of the patent holder. The second argument states that
any business practice that increases the returns on patents enhances efficiency, as it provides for
further innovation incentives. Third Degree Price discrimination (differential pricing as to
recognizable groups of customers) is particularly well seen by economists, as it can allow
increasing returns on patents and encouraging the spread of the technology in the same time.
Only three articles, Baumol Swanson (2005)11, Géradin (2008)12 and Crane (2008)13, explore
explicitly what non-discrimination in FRAND means with respect to non-discrimination in
general antitrust law. Baumol and Swanson suggest the so-called Efficient Component Pricing
Rule (ECPR) as benchmark for establishing discriminatory royalties. According to this rule,
royalties are discriminatory if the licensee is asked to pay more than the difference between the
price the IP owner asks for his own manufactured products and the cumulative cost of all other
inputs than the IP. This rule in fact establishes a general cap on the price level and does therefore
not deal with the competitive effects of the price structure. It ignores discrimination among
licensees. It is further-on virtually inapplicable in high-technology industries, as any single
product relies upon many different protected technologies and other intangibles. It is therefore
impossible to derive from accounting figures the economic cost of using self-owned IP.
Géradin and Crane make an interpretation of non-discrimination that comes closer to our
analysis. Both acknowledge that non-discrimination is not clear-cut and that discrimination can
take place also between different licensees. Crane refutes the reasoning according to which an IP
owner has no rational interest in discriminating in favor of his downstream affiliates. This
11
William Baumol and Daniel Swanson: Reasonable and Non-Discriminatory (RAND) Royalties, Standards
Selection, and Control of Market Power, Antitrust Law Journal, 2005
12
Damien Géradin: Pricing abuses by essential patent holders in a standard-setting context: A view from Europe;
Paper prepared for the “The Remedies for Dominant Firm Misconduct” Conference June 4-5, 2008 – University of
Virginia
13
Crane, op.cit.
argument of course ignores the reality of technology-intensive industries, where patent owners
have to share market power with the owners of complementary patents and producers with
preferential access to consumer markets. He further-on refutes the reasoning as to which such
discriminatory pricing is necessarily innocuous. Nevertheless, neither Géradin nor Crane
undertake to give positive meaning to non-discrimination commitments. Géradin argues that such
interpretation of RAND commitments falls outside Competition Law and therefore outside his
scope of interest, Crane observes the ambiguous welfare effects of prohibiting discriminatory IP
licensing and therefore retreats to a position of “in doubt do no harm”.
Against the background of such weak guidance from academic literature in interpreting nondiscrimination commitments, it is not surprising that licensees have obtained little help from
courts or antitrust authorities against licensing terms they perceive as discriminatory.
2.
In 2005, the Independent Optical Disc Replicator Association IODRA has submitted a formal
complaint pursuant to Article 7 of the Council Regulation 1/2003 to the European Commission,
in which it complains about discriminatory joint licensing practices by the DVD and MPEG2
patent pools, which run counter to European Competition Law and the patent holders’ nondiscrimination promise. This formal complaint was the climax of tense relationships between
patent holders and independent licensees, characterized by repeated patent infringement claims
and litigation about licensing conditions. Also declarations from American disc replicators and
European Consumer Associations tried to draw antitrust authorities’ attention to licensing
conditions in the disc replication industry14. Even though the complaint is still pending, antitrust
enforcement action against the accused patent pools is no longer realistically to be expected. We
nevertheless chose the industry as case study for the sensitive issues of price discrimination and
especially the potential risks that independent manufacturers incur when implementing new
technology standards. Optical disc replication is an interesting case to study, as it relates to two of
the eldest and best-studied modern patent pools, which have set a precedent for an antitrust
analysis of the current wave of joint licensing schemes. It is intriguing how little attention the
fundamental differences between the two pools have hereby received. We argue that some
features of the optical disc industry point to weaknesses and deficiencies of the current rules on
IPR licensing after standardization. The crucial point is the vagueness of the currently practiced
non-discrimination promise.
Optical disc replication is an industrial process whereby content is pressed on large series of discs
using so-called stampers. Even though this process is not very intensive in sophisticated
technology, disc replicators have to contract licenses for the use of essential patents incorporated
in two standards important for replication: the DVD 3C and DVD 6C patent pools require
royalties for using the disc format standardized as Digital Versatile Disc; and members of the
Moving Picture Expert Group at ISO have founded the MPEG Licensing Association to manage
14
Michelle Childs on behalf of Which?, Letter to DG Comp re consumer concerns in the media industry; the open
letter can be consulted following the link http://lists.essential.org/pipermail/a2k/2006-October/001758.html; Tom
O’Reilly (former chairman of American Independent Media Manufactuers‘ association AIMMA): Speaking out for
replicators, oto-online December 2006
a patent pool for the MPEG 2 Video data compression technology. Optical disc replicators are
particularly vulnerable to discriminatory licensing costs, as the share of royalties in production
costs is unusually high. Antitrust authorities have during their review of the DVD and MPEG
patent pools expressed little concern that these agreements could significantly affect downstream
competition, because the suggested royalty rates made up for only small shares in production
costs on the downstream markets. Since 1997 production costs in disc replication have fallen
dramatically (ex-factory prices for replicated discs came down from above 2,50 USD in 1997 to
below 0,50 USD today), whereas royalty rates have been reduced only slightly, and therefore the
share of royalties in production costs has increased sharply15.
In our case study, we took note of different types of complaints and concerns raised by optical
disc replicator representatives. IODRA’s complaint points to outright preferential treatment
granted to some standard adopters that has been reported by licensees or business associations
and been subject to litigation16. These complaints are difficult to assess, as the concrete licensing
conditions are subject to confidentiality. The complaints and our case study further-on encompass
differential pricing openly practiced in the standard licensing agreements issued by the pools.
Even though these differential features can give rise to very considerable differences in
production costs between competing manufacturers, there are also efficiency reasons speaking in
favor of their practice that have to be born in mind. We will in the following distinguish between
discrimination taking the form of different action towards different potential licensees and
discriminating price menus, offering discounts functionally unavailable to some but not all
potential licensees.
Outright preferential treatment of individual standard adopters by patent holders can take various
forms. The most clear-cut form is granting preferential licensing contracts to some but not to all
independent standard adopters wishing to contract a license. As all three patent pools issue
standard contracts to all potential licensees, this concern appears to be appeased. Nevertheless,
IODRA and disc replicators such as ODS report the grant of preferential side-letters to some
licensees. Furthermore patent holders are free to negotiate individual licenses with standard
adopters, and it can not be excluded that some standard adopters manage to obtain better
conditions from individual licensors than those detailed in the standard license from the joint
licensor.
Whereas the concrete provisions included in licensing contracts are subject to speculation,
safeguards against this type of discrimination are easier to observe and reveal important
differences between the DVD patent pools and the MPEG2 pool. Whereas MPEG LA has
committed to provide “the same terms and conditions to all licensees”, the DVD patent pools’
commitments “include only a “most favorable conditions” clause entitling licensees to the
benefit of any lower royalty “rate” granted licensees under “otherwise similar and substantially
15
“The patent royalties, which have been set at $ 0,168 at the early stage of the introduction of DVDs in 1997, when
a DVD (without packaging) has cost Euro 4,50 and therefore represented about 6% of the ex factory price, have not
come down proportionately to the price of a DVD in today's market. Factories owned by some of the patent owners
in this lawsuit offer DVDs for well below € 0,20. For this price level the royalties set by the same parties amount to a
minimum of US$ 0,1155, which is between 45 and 50% of today's ex factory price. This is a royalty share without
precedent in any other industry producing consumer products.”, quoted from
http://www.emedialive.com/Articles/ReadArticle.aspx?ArticleID=12165
16
See e.g. MPEGLA vs. ODS, op.cit.
the same conditions.” [...] In practice, the DVD pools are now offering different royalty rates to
different licensees, depending upon when prospective licensees sign licenses. Even when offering
the same royalties, the DVD pools are offering different terms to different licensees. Given the
potential for significant differences in effective price through non-price terms, such
discrimination may swallow the prohibition if allowed.”17
Another delicate issue is discriminatory patent enforcement. Disc replicators repeatedly complain
about distortions of competition by costless use of patents by IP infringing replicators. Patent
litigation is costly and lengthy, and if the private return of such an action does not exceed its
private cost, the welfare-maximizing strategy for the patent owner is rational apathy, a conscious
decision to renounce on enforcing his right. This is indeed a worrisome scenario for optical disc
replicators, allegedly facing competition from many production plants running without the
appropriate licenses. Patent pool representatives decline to have knowledge of such production
plants. Given the obscure situation of European patent protection, such allegations are
nevertheless plausible. The appointment of a third entity, such as MPEG Licensing Association
for the MPEG2 pool, could offer some relief to lawful licensees, as the third entity has an explicit
legal obligation of patent enforcement towards the patent owners.
Finally, the playing field for competition is crucially determined by the provisions of patent pools
concerning the use of the technology by pool members. Two different models are practiced: in
the MPEGLA model, pool members have to contract the same licenses as independent
manufacturers, if they wish to make use of the technology. This means that pool members pay
money into the pool from which they receive their share when royalties are distributed. Even
though these sums of money can neutralize themselves, such a scheme means that a patent holder
incurs nearly the same opportunity cost per unit of production as independent licensees. The
DVD patent pools on the other hand include cross licensing schemes. This means that pool
members are free to use the technology protected by patents of other pool members free of charge
under condition of reciprocity. Under such a scheme patent holders run with an important
competitive advantage over independent licensees on the downstream manufacturing market.
Even though there are important and valid efficiency reasons speaking in favor of such a cross
licensing agreement, it is striking that this fundamental feature has not been addressed when the
DoJ has reviewed the DVD patent pools by clear and explicit reference to the precedent set by the
MPEG2 patent pool.
Besides these forms of differential treatment granted to individual standard adopters, the standard
licensing contracts of the pools contain many general provisions that crucially shape the
conditions under which downstream manufacturers compete. Striking examples are the choice of
per unit or percentage royalty fees, the grant of quantity rebates and the possibility of special
discounts for early adopters or long-term subscribers.
In the MPEGLA vs. ODS case, the defendant disc replicator alleged that MPEGLA’s practice of
per unit fees is abusive, as it entails an automatic progression of overall royalty payments and of
the share of royalties in production costs over time as the standard spreads throughout the
17
Prepared Testimony of Howard Morse, Drinker Biddle & Reath LLP before the U.S. Department of Justice
Antitrust Division and Federal Trade Commission Hearings on Competition and Intellectual Property Law and
Policy in the Knowledge-Based Economy: Cross Licensing and Patent Pools; April 17, 2002
industry and production costs decrease. Per unit fees are further-on alleged to be discriminatory
against replicators specializing on low value-added sectors such as covermount replication
(gadget DVDs intended for publicity purposes only). In a different litigation, Chinese DVD
player manufacturer Wuxi claimed that the percentage royalty scheme practiced by the DVD 3C
patent pool is discriminatory as it entails different per unit production costs for different
producers depending on the price of the sold product18. In both cases the courts rejected the
charges of discrimination.
From an empirical overview over the pricing practices of over 50 pools, we can infer that both
per unit and percentage royalty schemes are widely practiced, even though the vast majority of
pools practices per unit fees. Even though percentage royalties have the advantage to provide
subscribers with the guarantee not to be driven out of the market, it is impossible to conclude that
one of the pricing schemes is illegal or generally detrimental. In our review we furthermore
revealed that a majority of pools practice quantity rebates, two part tariffs or royalty caps, all
having the effect of providing large licensees with a lower per unit cost. The ratio can go up to
above 1:719, which is clearly incompatible with a narrow interpretation of the non-discrimination
promise. Also early-subscriber discounts, shielding licensees from potential competition by
entrants, are practiced by at least 5 pools we have knowledge of.
But we took note not only of the concerns of independent disc replicators, but also of economic
reasons that can justify the practice of differential prices. This is especially true of cross-licensing
agreements and quantity rebates.
Cross licensing agreements are heavily criticized by licensees as distorting competition on the
downstream manufacturing market. This understandable criticism should not veil the fact that
cross licenses are a widely used instrument in high technology industries that have generally been
seen as efficiency-enhancing and legitimate by antitrust authorities20. By comparison to a general
pool agreement, cross licensing agreements have the advantage of resolving the multiple margin
problem generated by several patent holders owning strictly complementary patents. As each
patent holder can implement the standard without incurring licensing costs, he can pass on this
reduced per unit price to his own customers. On the other hand, in the absence of cross licensing
agreements integrated patent holders have a stronger interest in viable licensing conditions,
leading to lower royalty rates for all potential standard adopters. Which effect dominates for
consumer welfare is a difficult question that would be worth the while to be investigated by
formal Industrial Economics21. In any case cross licensing is clearly an important feature of joint
licensing schemes and it is questionable that the fundamental difference in this respect between
the DVD pools and the MPEGLA scheme has not been addressed by the DoJ business reviews.
18
Wuxi Multimedia, Ltd. v. Koninklijke Philips Electronics., Case No. 08-1041 (Federal Circuit, June 5, 2008)
Royalties for MPEG Surround consumer products, see www.vialicensing.com
20
To quoate as an example from a DoJ statement: „The Agencies continue to recognize that most of the nonexclusive
cross-licensing agreements of the type discussed herein generally do not raise competition concerns”
www.usdoj.gov/atr/public/hearings/ip/chapter_3.pdf
21
A promising framework would be an extension to Kim’s paper comparing the pricing incentives of vertically
integrated pool members and innovation specialists having no downstream affiliates; Sung-Hwan Kim: Vertical
Structure and Patent Pools, Review of Industrial Organization, 2004
19
Pricing schemes benefiting large licensees are equally ambiguous. On the one side, they reduce
the competitive advantage of patent owners over independent licensees, as the declining per unit
cost of licensees is an incentive to produce larger quantities. On the other side, declining per unit
costs can drive small potential licensees out of the market. While licensees can thus either benefit
or incur losses from such pricing schemes, consumers tend to benefit, as quantity rebates
encourage manufacturers to exploit economies of scale.
While some independent replicators thus indeed seem to be harmed by the pricing policy of the
patent pools, it is impossible to conclude unambiguously on the existence of consumer harm.
This is especially true when one bears in mind the massive decline of ex-factory sale prices of
replicated discs. Even a detailed and careful analysis can not safely establish the effects of the
incriminated pricing schemes by comparison to what would have happened in their absence.
Enforcement of competition law against alleged discrimination requires evidence of consumer
harm. This bar could turn out to be very high for independent optical disc replicators. They have
furthermore an important strand of economic theory standing against them. Price discrimination
on IP licenses is generally seen as efficiency-enhancing, as it allows softening the dilemma
inherent to intellectual property: IPR allow their owner to exclude others from the use of the
technology in order to internalize some of the returns of their innovative activity. This is achieved
at the price of a deadweight loss resulting from the restricted use of the technology. Price
discrimination allows the IP owner to align royalty requirements with the user’s disposition to
pay, and thus to earn more while excluding fewer potential licensees. It is an often quoted
example that Microsoft’s students discounts for MS Software are efficiency-enhancing, as they
allow software-developer to reap higher profits while encouraging the rapid spread of new
software. Baumol generalizes the argument to state that in industries where fix costs are
important price discrimination not only is not an illicit exercise of market power, but can be
indicator of vigorous and efficient competition22.
It is fundamental to understand that this reasoning applies to consumer markets only. On input
markets, this story does not hold anymore. While a consumer is not hurt if another consumer is
granted a preferential price, an intermediary faces a competitive disadvantage and risks to be
driven out of the market if his competitors receive more favorable contract conditions. Therefore
also the welfare effects of price discrimination on input markets are different. An important and
still growing literature, going back at least to Katz (1987), explores the welfare effects of
discriminatory pricing practices on input markets. These effects vary and depend upon factors
which are often very difficult to observe23.
22
Baumol: Regulation Misled by Misread Theory: Perfect Competition and Competition-Imposed Price
Discrimination, 2005 Distinguished Lecture AEI Brookings Joint Center for Regulatory Studies
23
Michael Katz: The Welfare Effects of Third-Degree Price Discrimination in Intermediate Good Markets,
American Economic Review, 1987; see also Patrick DeGraba: Input Market Price Discrimination and the Choice of
Technology, American Economic Review; 1990; Daniel O’Brian and Greg Shaffer: The Welfare Effects of
Forbidding Discriminatory Discounts: A Secondary Line Analysis of Robinson Patman, Journal of Law, Economics
and Organization, 1994; Yoshihiro Yoshida: Third-Degree Price Discrimination in Input Markets: Output and
Welfare; American Economic Review, 2000 and Roman Inderst, Tommaso Valletti: Price Discrimination in Input
Markets, Working Paper Series, 2007
In light of this ambiguity, Crane recalls in mind the principle for antitrust enforcement “In doubt,
do no harm”. It is a consensual view that a general ban on differential IP pricing would not be
efficient. It rather seems warranted to follow Geradin’s reasoning: if the observed cases of
discriminatory licensing do not justify antitrust enforcement on the ground of Competition rules
banning discrimination, that should not mean that non-discrimination commitments as contractual
instruments can not be enforced. Unfortunately Geradin has not carried further this idea, leaving
business practitioners and authorities with their doubts how to apply and how to enforce nondiscrimination commitments. It should be sufficiently clear that generic arguments drawn from
economic theory of price discrimination can not be used to exclude a more far-reaching use and
interpretation of this instrument. The ambiguous welfare effects of price discrimination are
exactly the reason why a commitment-based flexible policy is superior to a general antitrust
enforcement practice.
3.
The potential for a commitment-based approach can be understood studying the case of optical
disc replication. Even if the effects of discriminatory licensing schemes on consumer welfare are
not clear-cut, their effects on the profitability of those who have to pay more are unambiguously
negative. This is particularly problematic because standardization and agreements on licensing
policies are not occurring at the same time. The DVD format was standardized in 1995, but joint
licenses from the DVD pools were available only after the business reviews in 1999. The BluRay
standard won its rivalry against HD DVD in spring 2007, and still there is no comprehensive joint
license available.
As standards are set long time before negotiations on joint licensing schemes have been
concluded and all IP holders have made public their decisions on licensing policies, potential
standard users have to make their adoption decisions without knowing the royalties they will
have to pay for essential IP. Royalties on discs already sold have to be paid retroactively, and
retroactive pricing is often different from the general pricing scheme.24 By action for incursive
24
As an example, here is an extract from the DVD 6C standard license (http://www.dvd6cla.com/royaltyrate.html):
relief patent holders can immediately drive out of the market any standard adopter unwilling to
accept the licensing conditions. In this case the patent infringement damages can easily outweigh
the profits made on sales, and sunk investment costs in replication lines (reportedly exceeding 2
million USD for a BluRay replication line) are lost.
It is hereby not the central concern how high the royalty per disc will be. In any case this amount
will be a tiny fraction of the price the consumer pays on replicated discs and will not significantly
reduce overall demand (in case of disc replication the ex-factory sales price of a replicated DVD
is below 0,50 USD and thus a tiny fraction of the retail prices of DVDs including content). If
high royalties have to be incurred by all competitors, they do therefore not represent a threat to
standard adopters. But while representing a tiny fraction of the output price, royalties represent an
important share of the production costs of disc replicators (as we have seen, this share can reach
50 %). If cross licensing, discounts unavailable to small replicators or insufficient patent
enforcement provide a replicator´s competitor with lower or no costs for use of IP, this
competitive disadvantage can easily drive the replicator out of the market and make his
investment in standard adoption a costly adventure.
This situation is not unknown to economic theory and has been analyzed e.g. in the context of
franchising25. As in the case of a franchisee, the profitability for a manufacturer to implement a
standard depends upon future action of the licensor: it is only profitable for the manufacturer to
implement the standard if the licensing conditions that will eventually be set do not discriminate
against him. But the licensor may have incentives to favor some licensees that are critical for the
success of the standard, or to discriminate in favor of his affiliates. As the costs of standard
implementation incurred by the independent manufacturer are sunk, patent holders can hold up
standard adopters without fearing loss of potential licensees. Against the background of this
threat, early standard adoption is unreasonable in the absence of strong commitments.
FRAND commitments provide potential licensees with no information on very essential features
of the licensing scheme. In order to undertake a well-informed adoption decision, manufacturers
need to know the licensors’ intentions as to cross licensing. If patent owners decide to have
recourse to cross licensing, their affiliates will benefit from lower production costs, making
standard adoption less attractive for independent licensees. Furthermore potential licensees
should be able to make their adoption decision depend upon the pricing structure: if heavy
quantity discounts or two part tariffs are to be expected, standard adoption is not attractive for
small independent manufacturers. Finally, the appointment of an independent third entity that
manages a pool can be an element of interest for potential standard adopters, as it might
encourage vigorous patent enforcement. None of this essential information seems to be included
in non-discrimination commitments. By fear of eventual discrimination against them,
independent disc replicators therefore refrain from heavy sunk investments in early standard
adoption. IODRA reports that only very few independent replicators have invested in BluRay
replication lines, leaving the lion share of the replication market to patent holders and their
affiliates. These market shares do in no way reflect market shares in DVD replication, on which
independent replicators stand for the majority of replicated discs. This shift of market shares can
be explained by still pending uncertainty about licensing terms for BluRay patents: If only one
25
Preston McAfee and Marius Schwartz: Opportunism in Multilateral Vertical Contracting: Non-Discrimination,
Exclusivity, and Uniformity; American Economic Review, 1994
out of all holders of essential IP refuses to grant non-discriminatory licenses, the concerned
licensees run with a competitive disadvantage.
These problems are real and likely to be significant. Disc replicator business associations report
strongly delayed publication of licensing terms after time-consuming negotiations on joint
licensing schemes, strong hesitation by independent licensees to adopt new standard generations,
and widespread complaints about weak patent enforcement. These problems can delay
implementation of a standard and weaken a standard in its rivalry against a competing
technology. If the fear of discrimination deters entry by efficient independent potential licensees,
overall returns on essential patents and consumer welfare are reduced.
Furthermore the possibility to use essential patents to leverage the market power of an own
affiliate on a downstream market could hold back patent holders from joining a coordinated
approach to licensing (such as a patent pool), increasing transaction costs, reducing transparency
and weakening patent enforcement.
In the case study of disc replication, it is striking that the only holder of DVD patents who never
joined a pool was Thomson, who controls one of the World’s biggest disc replicators. The
difficult negotiations on a joint licensing scheme for BluRay patents could further-on be an
indicator of diverging interests between firms integrated into different markets (disc replication,
disc manufacture, player and encoder manufacture). These facts suggest that the incentive to
grant favorable conditions to the own affiliate can lead to delayed or incomplete joint licensing
solutions. Possibly stronger non-discrimination policies could encourage smoother adoption of
cooperative licensing policies.
It becomes clear why FRAND commitments are endorsed by many patent owners, as they
encourage standard adoption and enhance the conditions for licensing, as long as they are
perceived as credible. Arguably the recent cases of litigation have shakened the confidence of
stakeholders in the reliability of FRAND commitments.26 For this reason standard setting bodies
such as VITA or IEEE come up with policies of more explicit commitments, such as
commitments on maximum royalty requests or outright ex-ante auctions. Also the statements of
Commission officials in favor of explicit ex-ante regimes appear as sensitive. Nevertheless, there
is no apparent reason why these policies of early commitments or ex-ante auctions should
concentrate on price levels. As we have stressed earlier, licensees can pass on royalty costs if
these are non-discriminatory. The most important obstacle to standard adoption is the fear of
eventual discriminatory pricing.
In the light of these problems non-discrimination commitments have the potential to play an
important role, if they are meaningful. This requires (1) that non-discrimination in FRAND goes
beyond non-discrimination obligations arising from antitrust law, (2) that commitments can be
effectively enforced, and (3) that IP holders have a choice to commit, i.e. that there is no plain
and general requirement to commit. If non-discrimination commitments are not understood as
going beyond antitrust law, they are just cheap talk. If a patent owner is able to break antitrust
law, he is equally able to break his non-discrimination promise. If on the other hand non26
Francois Lévêque and Yann Ménière : Vagueness of RAND commitments is unreasonable for patent holders,
unpublished, 2009
discrimination commitments are understood as generating no legally enforceable obligation, they
have no economic value, as they won’t be taken seriously by standard adopters. Finally, the valid
economic arguments that can justify the use of differential prices and generic arguments taken
from economic theory of price discrimination make clear that committing on a far-reaching form
of non-discrimination should not be a general obligation for patent owners in standardization
exercises.
These conditions are not necessarily met at the current state of the art. Wide practice of all kind
of rebates and heavy discounts in spite of non-discrimination commitments suggests that these
commitments are currently not interpreted as significantly restricting the patent holder’s freedom
in setting pricing schemes. The numerous complaints and litigations about discriminatory
licensing of patents essential to standards have so far failed to meet a sympathetic response by
courts or antitrust authorities. Furthermore, the fact that FRAND commitments increasingly turn
into a mandatory exercise for patent holders in standardization exercises makes a differentiated
use and costly interpretation of non-discrimination commitments impossible.
If non-discrimination commitments were to become meaningful instruments of choice, they could
be used as efficient tools of ex ante competition. Participants in standardization procedures could
better anticipate the effects of including proprietary technology. Potential standard users could
make better informed standard adoption decisions, especially when they have the choice between
rivaling technologies.
For Competition Policy, this could mean that functioning commitment policies by SSOs could
make heavy ex post interventions unnecessary. But antitrust can contribute to the emergence of
such a self-regulation system. Two prominent cases provide examples for antitrust contributions
to the efficient use of binding commitments in standardization: First, as in the proceedings
against Qualcomm, antitrust authorities can contribute to make commitments enforceable. A
breech of a commitment can be an abuse of dominant position, especially if the commitment
helped the firm introducing its technology into a standard. Second, as in the proceedings against
Rambus, antitrust authorities can obtain binding commitments where the standardization
participants or standard adopters were unable to obtain a commitment because of an
anticompetitive strategy by the patent holder, such as patent ambush (the failure to disclose one’s
patents during standardization). The remedies imposed upon Rambus by the European
Commission reflect some of the concerns that disc replicators mentioned with the pricing
practices by DVD and MPEG2 patent pools: not only does Rambus commit on an explicit MostFavored Customer clause that should make discriminatory preferential treatment impossible, also
is the royalty cap on which Rambus has to commit expressed as a percentage of the product
price27. Both these conditions should provide some guarantees that Rambus is unable to drive any
licensee out of the market.
In general, antitrust authorities are certainly among the addressees of non-discrimination
commitments and should not hesitate to require their execution. Patent pools and cross licensing
schemes are forms of horizontal coordination between competitors that are difficult to evaluate ex
ante. Just as other types of commitments in merger control or antitrust, non-discrimination
commitments in standardization facilitate antitrust authorities their task while clearing specific
27
PROPOSED COMMITMENT RAMBUS INC. CASE C-3/38.636,
http://ec.europa.eu/competition/antitrust/cases/decisions/38636/proposed_commitments.pdf
licensing practices. Antitrust authorities should use their longstanding experience with
commitment policies to adopt a more proactive stance toward standardization. This also implies
that antitrust authorities push for strong non-discrimination commitments where they seem
necessary, but not as a general policy in every standardization exercise.
Conclusion:
We have evaluated the complaints of independent disc replicators against the pricing practices by
the DVD and MPEG2 patent pools. Based on an analysis of the allegedly discriminatory practices
and the general evolution of the industry, heavy ex post antitrust intervention does not seem
warranted. Rather, the case of optical disc replication sheds light on the insufficiencies of current
ex ante commitment policies. We have stressed the importance of giving a concrete meaning to
the non-discrimination promise included in FRAND commitments. In the future, standard setting
bodies may be able to make antitrust intervention unnecessary thanks to successful ex ante
policies based on explicit and reliable commitments. In the mean time, the problematic of
discriminatory patent pricing, especially in the context of standardization, is likely to remain high
on the agenda of Competition Law enforcers. The recent moves of the EC, especially in the
Rambus case, might sketch a possible strategy how antitrust enforcers can deal with the thorny
issues of standardization.