United States: Antitrust And Intellectual Property Rights: Assessing The Link Between Standards And Market Power

Previously published in Antitrust, Vol. 21, No. 3, Summer 2007. © 2007 by the American Bar Association. Reproduced with permission. All rights reserved. This information or any portion thereof may not be copied or disseminated in any form or by any means or stored in an electronic database or retrieval system without the express written consent of the American Bar Association.

Tensions over intellectual property rights (IPRs) and antitrust law are certainly nothing new. What is new is the current brand of tension—not over whether a patent grants a legally-sanctioned "monopoly,"¹ but whether standards create market power for patent holders that can be abused even within competitive markets. A number of cases reflect this market-power-standard tension, including Microsoft v. European Commission, Dell, Rambus, Broadcom v. Qualcomm, and Nokia v. Qualcomm.² Each of these cases covers distinct claims, but they all have one common thread: claims that standards, either de facto or cooperative, augment the market power associated with IPRs and that market power can be abused.

This new strain of cases does not lend itself to cookie-cutter antitrust analysis. In some instances, applying traditional antitrust remedies to intellectual property problems may actually chill innovation or at least reduce incentives to participate in welfare-enhancing cooperative standardization efforts. That does not mean a whole new tool kit for antitrust cases with intellectual property elements, but it does mean considering how the traditional tools might work in the very different settings that involve IP and standards.

In this article, I consider the link between standards and market power and discuss the implications in light of the ever-growing trend for standardization within high technology industries.

The Cases

The European Commission’s 2004 decision against Microsoft (now on appeal before the Court of First Instance) turned significantly on intellectual property issues. The Commission found that Microsoft had abused its dominant position by refusing to supply information to competitors concerning certain proprietary communications protocols its Windows Server operating system uses to communicate among Windows-based clients and servers on a local or wide-area network.³ The protocols at issue essentially define how Windows interoperates with other programs, including such items as file and print services. Here the standard (Windows) is a de facto one, as recognized by the Commission, and the intellectual property is, according to Microsoft, embedded in the protocols—some protected by patent and others by trade secret. The Commission found that this information was indispensable to compete in what it calls the market for "work group server operating systems." Thus, by refusing to supply the information, the Commission found that Microsoft foreclosed competition.

The Commission’s remedy directed Microsoft "to disclose complete and accurate interface documentation which would allow non-Microsoft work group servers to achieve full interoperability with Windows PCs and servers."⁴ The Commission permitted Microsoft to charge reasonable and non-discriminatory (RAND) royalties for licensing the protocols, so long as the royalties did not reflect any "strategic value" arising from Microsoft’s dominant position in client or server operating systems. Since the 2004 decision, however, the Commission has demanded that Microsoft demonstrate that the interoperability protocols embody "innovation" with some intrinsic value and are not valuable simply because they have been kept secret. In its March 2007 Statement of Objections, the Commission reached the preliminary conclusion that Microsoft had not demonstrated innovation, and therefore it found that Microsoft’s proposed licensing plan was not RAND. Microsoft has countered that its protocols are innovative and its licensing proposal is RAND.

The remaining four cases involve standards set by voluntary standard-setting organizations (SSOs) instead of through non-cooperative competition. With formal standard setting through an SSO, firms choose to cooperate to define a standard or technical specification in order to meet some industry perceived need.

Member firms often propose their own proprietary IP for cooperative standards, and patented inventions are frequently implicated. As a result, the vast majority of formal SSOs, like IEEE (Institute of Electrical and Electronics Engineers) and ETSI (European Telecommunications Standards Institute), request that their members report their patents and other IP that might be interpreted as "essential" for a standard. That is, the standards body asks to be notified if its members hold any IP that they feel might need to be licensed in order to implement the developed standard. SSOs also usually call for all disclosed IP to be licensed on RAND terms to all firms wishing to implement the standard. No SSO appears to have defined what it means by the terms reasonable or non-discriminatory. Actual licensing terms are then left to bilateral negotiations outside of SSO activities, with the understanding that IP holders cannot offer exclusive licenses or refuse to negotiate a license.

The impetus for the disclosure and RAND rules are obvious— firms with downstream operations need to know what to license and with whom to negotiate if they are to implement the standard once it is approved. They also need to be assured that licenses will be available on affordable terms. Knowing these things before a standard is settled can influence which technologies are actually included in the standard.

Despite the presence of these IP policies within SSOs, disagreements have arisen. Dell is just such a dispute stemming from non-disclosure. In the early 1990s Dell participated in the Video Electronics Standards Association (VESA) effort to promulgate a standard for a VL-bus, a cable to carry information between a host computer and peripheral equipment. As part of that participation, Dell’s representative to VESA signed a document stating that "to the best of [his] knowledge" the company had no patents, copyrights, or trademarks that would be infringed by the VL-bus standard. But about eight months after the new standard was approved—and had achieved considerable commercial success—Dell informed VESA members that it did indeed have a patent that the standard was infringing and that it intended to seek compensation. The other members of VESA cried foul, and the FTC stepped in alleging "unfair methods of competition" in violation of Section 5 of the FTC Act. In late 1995, the FTC issued a proposed consent decree that precluded Dell from asserting its patent rights for the standard, but not for other applications. In essence, the remedy for nondisclosure was compulsory royalty-free licensing within the confines of the standard.

Rambus is similar to Dell, but the facts suggest that Rambus had more than one motive for participating in the standardsetting efforts.⁵ Rambus Inc., a firm that develops technologies for semiconductor memory devices, began participating in JEDEC (Joint Electronic Device Engineering Council) in 1991. Rambus did actually disclose one of its patents to JEDEC in 1993, but company email records apparently indicate that it knew it had other, more pertinent patents for the standards that JEDEC was working on that it nonetheless did not disclose. In fact, company documents evidently portray Rambus as attending SSO meetings in order to gather information about the standards, information that Rambus then incorporated into claims language for a patent application it was preparing at the time. After the Dell ruling, Rambus’s lawyers reportedly warned of antitrust concerns were Rambus to withhold relevant patents from the SSO. Instead of such disclosure, though, Rambus chose to officially leave JEDEC in 1996.

In 2000, Rambus sued Infineon in district court (EDVa) for allegedly infringing four of its patents in implementing the JEDEC SDRAM standards. Infineon counterclaimed fraud (under state law), based on Rambus’s participation in JEDEC during the development of the standards. The initial rulings in 2001 fell Infineon’s way: the district court found that Infineon had not infringed Rambus’s patents, and the jury found that Rambus had committed fraud by failing to properly disclose patent information to the SSO.

While all of this played out in district court, the FTC filed an administrative complaint against Rambus in 2002, alleging the company had violated Section 5 of the FTC Act, along the same lines as Dell.

Then, in 2003, the court of appeals (in a 2–1 decision) vacated the non-infringement judgment and remanded the issue to the district court. The majority opinion was far more skeptical of the anticompetitive implications of non-disclosure than either the FTC or the district court judge. In particular, the majority found that JEDEC’s disclosure rules were too vague to create any enforceable commitment for Rambus. In other words, Rambus’s behavior was excusable since JEDEC had not asked the firm for a more specific pledge.

Despite winning the appeal in the private case, Rambus still faced the FTC action. But in 2004, the FTC’s Chief Administrative Law Judge issued his initial decision, stating that the Complaint Counsel had failed to establish liability. In the wake of these two favorable rulings, Rambus settled with Infineon in early 2005. Infineon agreed to pay Rambus $5.85 million per quarter between November 2005 and November 2007 in exchange for a worldwide license to existing and future Rambus patents that Infineon would need to implement the JEDEC standards.

But the saga was not quite over. The FTC appealed the 2004 ruling, and in August 2006 the full Commission overruled the Administrative Judge. The Commission’s decision noted that Rambus engaged in exclusionary conduct that significantly contributed to its acquisition of monopoly power in four related markets. By hiding the potential that Rambus would be able to impose royalty obligations of its own choosing, and by silently using JEDEC to assemble a patent portfolio to cover the SDRAM and DDR SDRAM standards, Rambus’s conduct significantly contributed to JEDEC’s choice of Rambus’s technologies for incorporation in the JEDEC DRAM standards and to JEDEC’s failure to secure assurances regarding future royalty rates—which, in turn, significantly contributed to Rambus’s acquisition of monopoly power.⁶

Finally, in February of 2007, the FTC issued its remedy, which sets the maximum royalty rates that Rambus can charge for patents essential for JEDEC-compliant products. While the rates are not zero, as they were in Dell, they are quite low, ranging from .025 to .5 percent. Moreover, after three years the rates drop to zero. Rambus has appealed the FTC decision to the D.C. Circuit.

The two other cases that involve standards and market power issues hinge less on IP disclosure and more on what a RAND promise means in practice. In both Broadcom v. Qualcomm and Nokia v. Qualcomm, the plaintiffs argue that Qualcomm made a RAND promise to ETSI, but allege that the company is not living up to that promise.⁷ Again, at the core of these complaints is the presumption that being included in a standard, this time the standard for mobile telecom (so called 3G), confers market power on patent holders. Broadcom, in particular, argues that Qualcomm obtained monopoly power through its patents’ inclusion in the 3G standard, and that Qualcomm is exercising that power in the licensing terms it is asking ETSI members for, in violation of both the Sherman Act and the Clayton Act. The complaint in this case was dismissed because it did not support claims of antitrust violations, but Broadcom is appealing.⁸

One commonality between these two cases is concern over the "aggregate" royalty rate for the 3G standard. Both Broadcom and Nokia allege that RAND entails not just a reasonable royalty rate from each licensor as viewed in isolation, but a reasonable cumulative rate when all firms’ rates are stacked up, as would be required for any downstream firm to implement the standard. Since the elements of a standard are complementary, implementers must license all of them. Thus, downstream firms care about the total price, not necessarily any one element’s price. Broadcom and Nokia assert that a given firm’s rates can be deemed reasonable only in light of their relative place in the cumulative total. One key element in their complaints against Qualcomm, then, is that its rates are "excessive" in relation to its contributions to the standard.

In its defense, Qualcomm has argued that its IP contributions to the 3G standard are highly valuable and its rates are therefore justified. Moreover, Qualcomm points out that its patent portfolio was well recognized at the time ETSI 3G votes took place, and the firm’s royalty terms have not changed materially since that time. Qualcomm thus maintains that instead of any ex post abuse on its part, it is the mobile handset manufacturers buying the technology that are exerting market power in an effort to lower their licensing costs below reasonable levels.

The Economic Issues

The five cases summarized above cover a range of complaints that all stem from one underlying issue: the claim that intellectual property rights in the context of a market standard entail more than simple government-granted rights to exclusion; they can result in augmented market power that can be wielded in anticompetitive ways.

With Microsoft the logic underlying the Commission’s stance is that the firm’s position in Windows creates a de facto standard. Thus, the Commission maintains, as the definer of that standard, Microsoft has an obligation to enable competitors to reproduce Microsoft’s protocols in a way that allows them to compete directly with Microsoft’s work group server operating system products. Without doubt, interoperability issues are crucial, and are likely one of the key drivers behind the increased voluntary standardization activities in many high technology industries. The problem lies in how that interoperability is achieved.

Most worrisome is the effect that remedies equivalent to compulsory royalty-free licensing can have on industry innovation. It is well understood that a firm’s incentives to innovate and invest in risky R&D are driven to a large extent by the monetary rewards the firm expects to achieve in return, rewards that are negated by compulsory royalty-free licensing. ⁹ But more than economic theory cautions against compulsory licensing, as the Canadian Parliament’s "natural experiment" demonstrated in the 1970s through the early 1990s. The 1969 amendment to the Canadian Patent Act allowed for immediate compulsory patent licensing of pharmaceuticals. Economists studying the Canadian experience in the ensuing decades found that the compulsory licensing mandate did not increase innovation by licensees but instead reduced local Canadian commercial development of pharmaceuticals. ¹⁰ After the enactment of other amendments attempting to salvage compulsory licensing while still providing incentives for R&D, the Patent Act was amended again in 1993 to eliminate compulsory licensing altogether and instead extended patent protection to twenty years.

In the Microsoft case, the Commission appears to believe that the overall outcome will not be harmful to industry innovation. As expressed in the 2004 decision, the Commission maintains that on balance, the possible negative impact of an order to supply on Microsoft’s incentives to innovate is outweighed by its positive impact on the level of innovation of the whole industry (including Microsoft). As such, the need to protect Microsoft’s incentives to innovate cannot constitute an objective justification that would offset the exceptional circumstances identified.

It is too soon to tell whether Microsoft will innovate and patent less as a result of the Commission’s remedy, but history suggests it is a strong possibility. It is also too soon to tell whether Microsoft’s rivals will in fact fill in the void.

The disclosure and compulsory licensing issues examined under the broad umbrella of Section 5 of the FTC Act in Dell and Rambus may be less controversial than those at play in Microsoft, but they are far from clear cut. While a few SSO non-disclosure cases had been tried before under various theories (namely, equitable estoppel), Dell marks the first time that a firm’s violation of an SSO IP rule implied a violation of antitrust law as well. As a result, the case has serious implications for the standard-setting process, namely the extent of a member firm’s obligation to search for relevant IP. As dissenting Commissioner Azcuenaga highlighted in her statement, the FTC never alleged that Dell willfully mislead VESA in any way.¹¹ It appears most likely that Dell’s lack of disclosure was an honest oversight rather than an intentional deception. Given that firms typically send engineers and not patent lawyers as their representatives to an SSO, inadvertent nondisclosure is bound to happen again. The upshot of this distinction is that the Dell consent decree translates into a duty to disclose well beyond what VESA actually required of its members: it effectively imposes a duty for a thorough search of all IP, with royalty-free licensing as the price for incomplete reporting, regardless of the circumstances.

Rambus represents the other side of the non-disclosure coin. The record in this case suggests that Rambus did intend to withhold relevant patents from JEDEC members. Evidently, to avoid the legal consequences of that decision the firm quit the SSO just before the official vote on the standard took place. The FTC’s Rambus ruling indicates that dropping out does not provide a safe harbor: a firm cannot profit from knowledge gained during cooperative standard setting efforts without abiding by SSO rules.

Rambus also highlights the dangers arising from ill-defined SSO disclosure policies. Most disclosure rules are actually rather vague. For example, one SSO requests that members make "reasonable efforts" to identify IP that "might" read on a standard, and do so in a "timely fashion." This ambiguity has been intentional, reflecting a number of concerns. The first is related to antitrust law. SSOs are, after all, organizations of firms, many of which are direct rivals in various downstream markets, that meet to coordinate on product offerings. The procompetitive benefits of such efforts are widely recognized— as witnessed by the increasing role that such standards are playing in high technology industries—but the potential for cartel-like behavior remains. Thus SSOs are generally careful to avoid any rules that even suggest member coercion. The second concern is more pragmatic: a voluntary standard does little good if no firms volunteer to participate. Thus, SSOs are careful to avoid rules that place too great a burden on members. Assessing each patent to determine its relevance for a potential standard—as suggested by the Dell ruling—would be extremely burdensome, and especially so for firms with large patent portfolios. The third driver of ambiguity is related to the second: SSOs are typically comprised of firms with widely diverging interests. A consensus agreement on precise language for an IPR policy may well be an unattainable goal.

Prior thoughts of antitrust scrutiny and member resistance aside, the lack of specificity in SSO IPR policies has clearly left room for firms to game the system (or at least attempt to). In fact, the district court ruling in the Rambus case observed that JEDEC’s disclosure rules were too vague to constitute an actionable commitment and thus Rambus had no obligation to disclose all of its relevant IP. On the other hand, the FTC decisions suggest that any non-disclosure— willful or otherwise—could lead to severe antitrust remedies. Neither of these two extremes seems ideal, suggesting that SSOs should at least consider revising their IP policies.

One apparent effect of the FTC rulings has been to counteract firms’ natural tendencies for under-disclosure. That is, some firms have been reluctant to disclose all of their patents for fear of providing their rivals within the standard with too much information about the direction of their research program and where they see fit to expend resources on patenting. Plus, on a pragmatic front, even if a firm’s representatives to an SSO know the company’s IPR portfolio inside and out, determining which patents are genuinely essential for a standard is rarely a clear-cut matter. Instead, finding relevant patents can require a good deal of analysis and interpretation. Finally, standards evolve over time, making IP disclosure something of a moving target. But in light of Dell and Rambus, some IP disclosures to SSOs now appear to fall under the insurance category—patents that may or may not be relevant are disclosed anyway to avoid accusations of hold out. This might be a good thing, in that the rulings encourage more complete disclosure, but only insofar as the additional disclosures are genuinely relevant. Excessive irrelevant disclosures could weigh a standard down, incorrectly signaling a patent thicket of rights to be cleared before implementation could proceed.

The FTC rulings might also have a far more detrimental effect than simply populating standards with irrelevant patents. It might lead to fewer SSO participants, although this effect is necessarily more difficult to discern. That is, if abdication of all patent rights is the cost of mistakenly failing to disclose some IP that turns out in retrospect to be relevant for the implementation of a standard, some firms could choose to opt out of the voluntary standard-setting arena altogether—although this route is not always an option, even for firms that would prefer to leave. And, too, SSOs might respond to the recent cases with more stringent search and disclosure rules, meant as protection measures for members in the future, but which could have the same effect of reducing overall participation.

While it might appear democratic on its face, IPR rules that lower firm participation could have serious negative consequences. Cooperative standards rely on broad participation to ensure sufficient industry representation and raise the odds that the best technology available is included in the standard. Moreover, within most SSOs IP users outnumber IP owners by a considerable margin, so care must be taken that one group’s interests do not supersede the other’s. Reducing the number of firms with valuable IP would harm the standardization process and the consumers reliant on those standards. This effect, if appreciable, seems a disproportionate price to pay for inadvertent non-disclosure.

Nor are disclosure rules the only aspect of SSO IP policies under attack. The two mobile telecom cases call for more specific rules regarding RAND licensing commitments. Clearly licensing is of paramount importance for a standard’s success. After all, what good is a standard if downstream firms cannot access the IP needed to implement it? But the reforms proposed in the Broadcom and Nokia cases have severe limitations. For example, the RAND benchmark put forth by Broadcom and Nokia is something called numeric proportionality. If a standard is comprised of 100 patents, and company X holds 10 of those patents, company X is entitled to 10 percent of the aggregate royalty. Of course, that aggregate rate must be agreed to first (which would be difficult in light of the usual diversity of interests in any given SSO), but after that the formula is simple to implement.

The problem with this seemingly straightforward approach is that patents are not created equal, as decades of empirical work in the economics literature has well established.¹² One patent might cover a crucial aspect of the standard, while another covers a relatively minor backward compatibility feature or some optional element of the standard. Compensating these two patents equally is neither reasonable nor nondiscriminatory, let alone economically efficient. As a result, numeric proportionality only works under very special circumstances, such as acknowledged symmetry in patent contributions to a standard across contributing firms.¹³

But if simple patent counts do not work, what does? Here’s the rub. To truly assess a patent’s worth a detailed patent by patent review is necessary, entailing technical as well as legal analysis. Moreover, even this kind of in-depth review is subjective, at least in part. One set of reviewers might provide a dramatically different rank ordering of patent values than would another set of reviewers. The literature has developed a number of well-accepted proxies for patent value based on observable patent measures, but different proxies can suggest different rankings as well. A silver bullet solution is not readily apparent.

The courts facing these new IP-antitrust cases do have some options, but none of them is easy.¹⁴ The most obvious approach would be to evaluate the IP in a standard using the same criteria as applied in reasonable royalty assessments for patent infringement cases—the Georgia-Pacific fifteen factors. ¹⁵ These include such items as the conditions and prices in previous licensing agreements involving the IP at issue, the market power of the IP and the products reliant on it, and profits attributable to it. It’s a case-by-case approach that can lead to dramatically different rulings from different judges, but it is a familiar approach as Georgia-Pacific has been guiding patent infringement damages for over thirty years.

Another option is one proposed by Daniel Swanson and William Baumol.¹⁶ They suggest applying the Efficient Component Pricing Rule (ECPR) as developed in the utilities regulation literature to give meaning to RAND. The basic idea here is that a vertically integrated firm (meaning it both has IP and implements the standard) should get paid enough to make it indifferent between licensing its IP and using the IP internally. The final good price thus constrains the royalty rate, addressing concerns over the aggregate rate. Swanson and Baumol’s approach only applies to vertically integrated firms, however, and their model assumes one patent held by one firm per standard, which is far from reality. Their second proposal is more problematic: an auction over technology and price. This would involve parties with diverse business models, though, which has well-known limitations—namely, vertically integrated firms have a clear advantage in winning an upstream auction because they can forgo royalty earnings in exchange for downstream profits. Nonetheless, their models can be extended to multiple firms and multiple patents, and these extensions could be useful in defining a safe harbor for the current RAND cases.¹⁷

The Broadcom case may not reach the stage where such an evaluation is necessary. Qualcomm moved for a motion to dismiss, and in August 2006 the New Jersey District Court obliged—although Broadcom has appealed. The court acknowledged that "The existence of SDOs [Standards Development Organizations] in the cell phone industry draws additional antitrust scrutiny to Qualcomm’s alleged conduct."¹⁸ That said, the judge went on to note that being included in a standard did not bestow any additional market power that holding a patent did not already provide. In other words, patents entail exclusionary rights by definition and design, even within an SSO. Moreover, the judge reasoned that "[w]hile Qualcomm’s behavior may have influenced how the SDO would eliminate competition, it is the SDO’s decision to set a standard for WCDMA [3G] technology, not Qualcomm’s "inducement," that results in the absence of competing WCDMA technologies.¹⁹

This latter statement echoes a point made by the dissenting Commissioner in the Dell case. Anticompetitive abuse of market power implies that a firm acquired the power to control price and output in some relevant antitrust market. That a standard makes a product more valuable, and thus enables firms to charge more for the necessary inputs, does not automatically square with this definition of price control and output reduction. Economic efficiency requires consideration for the firms with key IP for a standard—if standards enhance value, at least some of that enhanced value should go to the IP holders making that standard possible, especially since this kind of feedback increases the incentives to participate in voluntary standard setting organizations. Furthermore, if inclusion in a standard does in fact enhance market power for patent holders, and empirical evidence on this point is mixed, the abuse of that power must still be clearly delineated within a well-defined antitrust market. That means addressing all of the traditional competition factors, like barriers to entry and the availability of substitutes (and yes, even standards can have substitutes).

The Implications

As noted above, the new tension over IPRs in antitrust cases does not mean a whole new set of competition rules. What is clear, however, is that how the customary tools are applied needs to be sensitive to the unique aspects that IP issues bring to antitrust. A thorough understanding of intellectual property rights is crucial in evaluating claims of abuse of market power in these new competition cases, since IP has theory and stylized facts all its own.

It is reasonable to expect more antitrust cases to include intellectual property aspects—from alleged misbehavior in standard-setting organizations to anticompetitive crosslicensing practices to cartel-like patent pool cooperatives. Careful thinking on the implications of traditional antitrust remedies in these "new" settings will be important if courts and competition authorities hope to do more good than harm.