Friday, June 13, 2014

Reflections on Holdup and Royalty Stacking, Part 2

On Wednesday I published Part 1 of my current thinking on the topic of patent holdup and royalty stacking. Today, Part 2 focuses on royalty stacking in particular, and on the question of whether holdup and royalty stacking should be addressed, if at all, under the law of patent remedies or antitrust.

*          *         *

My analysis of royalty stacking is similar in some ways to my analysis of holdup.  Whenever a device incorporates multiple patents owned by multiple owners, economic analysis suggests there may be a Cournot complements problem.  As I’ve written before:
. . . when separate owners of complementary inputs each demand what is (for them) the individually profit-maximizing price, in exchange for permission to include those inputs in an end product, the cost of producing the end product will result in a price for the end product that is higher than the social optimum. As a result, the producers of the inputs are collectively worse off, due to the effect on the quantity demanded of the final product, than if they were able to coordinate the prices they charge.  Thus in the patent context in particular, as Farrell et al. note, “when multiple firms engage in complementary innovation, it is not possible . . . for each innovator's reward to equal its invention's incremental contribution.”  In the worst-case scenario, the aggregation of multiple royalties (sometimes referred to as “royalty stacking”) may result in an aggregate licensing fee that exceeds the value of the end product.
Thomas F. Cotter, Patent Holdup, Patent Remedies, and Antitrust Responses, 34 J. Corp. L. 1151, 1169 (2009).  As with holdup, however, there may be many situations in which a mere risk of a Cournot complements problem shouldn’t lead to any special response; and empirically, it may be very difficult to prove that royalty stacking is actually present.  (See, e.g., Judge Davis’s analysis in Ericsson Inc. v. D-Link Sys., Case No. 6:10-CV-473, 2013 WL 4046225,at *18 (E.D. Tex. Aug. 6, 2013).)  The fact that one patent owner charges, for example, 5% of the value of an end product doesn’t necessarily mean that all of the other patent owners would be entitled, much less able, to extract as much; some patents surely are worth more than others.  Moreover, patentees and implementers often agree to cross-licenses, sometimes on a one-for-one basis, simply to avoid the difficulties of calculating more accurate fees.  Thus, just because patent owners’ aggregate royalty demands amount to (say) $120 of the average $400 price of a smartphone (as argued in Ann Armstrong et al., Surveying Royalty Demands for the Components within the Modern Smartphone, available at, doesn’t necessarily mean that royalty stacking of a type we ought to be concerned about is present.  At the same time, however, just because aggregate royalty demands are not decimating an industry doesn’t prove that the Cournot complements phenomenon is not present at all, or that we shouldn’t take the risk of royalty stacking into account in setting FRAND royalties.  To be sure, reasonable minds may differ concerning what it should mean for a court to "take the risk of royalty stacking into account."  Maybe all it can amount to, as a practical matter, is a check on sanity--as in Microsoft v. Motorola, where Judge Robart concluded that the rate Motorola was seeking for certain patents was excessive, in part because (1) if every patent owner sought royalties similar to those Motorola was seeking, the price of the end product would be untenable, and (2) those particular patents constituted only minor contributions to the standard (paras. 456-57).  See also Judge Holderman's opinion in Innovatio IP Ventures LLC, p.18.  As with holdup, royalty stacking is a matter of degree, and judgment is required to determine at which point the potential harm is serious enough to do anything about it.

In summary, even if holdup is not causing “manufacturers [to] reduce the introduction of new products,” Alexander Galetovic et al., Patent Holdup:  Do Patent Holders Hold Up Innovation?, Hoover IP2 Working Paper Series No. 14011, at 10 (May 2014), and firms are not charging royalties that in the aggregate threaten to exceed the value of end products, this does not imply that policymakers should be unconcerned about the risks of opportunism that are inherent to the ownership of standard-essential technologies.  In my view, those risks counsel against awarding injunctions in some cases; and, regardless of whether one perceives those risks are substantial or not, I should think that we all can agree that it is worthwhile to work on improving the methodology for awarding damages so that those damages more accurately reflect the value of patented technology.

*                     *                    *

Notice, however, that nothing I‘ve argued so far suggests that antitrust law is the optimal way to combat whatever level of holdup and royalty stacking one might think is worth combating.  In other work, I’ve argued that it is preferable simply to deny injunctions in patent infringement cases in which the risk of holdup is substantial, and to calculate damages based on the familiar ex ante bargaining framework.  These results are feasible under U.S. law after eBay, and could be adopted in Europe under the Enforcement Directive.  Nevertheless, many European states still view injunctions as more or less the successful patentee’s entitlement, unless some other body of law (antitrust, contract, abuse of right) mandates a different result.  So what about competition law as a source for denying injunctions in SEP cases?

As I’ve stated before, I’m skeptical of this approach both as a matter of law and as a matter of policy. Under U.S. law, while it is possible, given the right set of facts, for the act of seeking an injunction for the infringement of a FRAND-encumbered SEP to be a violation of Sherman Act § 2 (monopolization or attempted monopolization), as a general matter there are many problems, among them (1) even if it is possible to characterize the patentee’s behavior as a unilateral refusal to deal, such behavior is rarely if ever actionable under U.S. law unless the patent itself was acquired fraudulently or the litigation is a sham; and (2) even if, notwithstanding the D.C. Circuit’s decision in Rambus v. FTC, 522 F.3d 456 (D.C. Cir. 2008), it can be a violation to obtain market power by deceiving an SSO into adopting your patented technology, it is not at all clear that reneging on a post- standard adoption commitment to license an SEP on FRAND terms amounts to the willful acquisition or maintenance of monopoly power in violation of § 2.  (There also may be problems under the Noerr-Pennington doctrine, or in proving monopoly power in a well-defined market, though personally I am inclined to think that these two potential weaknesses in the antitrust theory are exaggerated.)  Moreover, as a matter of policy, extending U.S. antitrust law to this domain could be problematic given the risk of private civil suits for treble damages (potentially resulting in overdeterrence).  On the other hand, under European competition law the same set of facts could, in the opinion of some but certainly not all commentators, amount to an abuse of dominant position.  I hesitate to offer my own opinion on the likely course of European competition law on this issue, but I have suggested that, given the absence of treble damages and the relative paucity of private antitrust actions in Europe, the competition law approach might be an acceptable second-best solution to the patent holdup problem there, even if it is not a desirable approach to use in the U.S.  For further discussion, see Thomas F. Cotter, The Comparative Law and Economics of Standard-Essential Patents and FRAND Royalties, 22 Tex. Intell. Prop. L.J. __ (forthcoming 2014), available here.


  1. As you note in your fifth point, Holderman J expressly stated that the rate should be based on the assumption that the patents are valid, and he rejects some comparators on the basis that they were negotiated under conditions of uncertainty. But I’m not sure the top-down methodology is consistent with that assumption. The top-down methodology caps the royalty at the manufacturer’s profit, on the view that otherwise the manufacturers “would leave the chip-making business altogether, rather than pay a royalty that would obliterate their profits.” Innovatio’s witness, Teece, makes the very obvious point that no they wouldn’t, they would just raise their prices. Holderman J responded to that by noting that there were three major manufacturers who had already licensed from Innovatio, and competition from those manufacturers would stop the defendant manufacturers from raising their prices. In effect, the licence negotiated by the three licensed manufacturers is being used as a comparator, albeit indirectly. But it seems to me that the licenses to those three manufacturers must have been negotiated under conditions of uncertainty regarding validity. So it seems to me that the top down approach indirectlly uses licenses negotiated under uncertainty as comparators, even though that was expressly rejected when it was tried directly.

    On another note, how does the assumption of validity interact with the non-discriminatory part of the FRAND commitment. If A negotiated prior to any litigation its royalty will be discounted by the probability of invalidity. If B then litigates, and the patent is held to be valid, essential and infringed, then the royalty should be calculated on the assumption that the parties would have known that. This implies that the royalty paid by B will necessarily be higher than that paid by A. That looks discriminatory to me.

  2. You seem to be correct as to the fifth point above. On the nondiscrimination issue, Carlton & Shampine address this issue in their paper "An Economic Interpretation of FRAND,", at p.14. In their view, the nondiscrimination condition is sufficiently important that B shouldn't be required to pay more than A. Absent the FRAND commitment, that wouldn't make any sense. If their point about nondiscrimination is correct, it highlights an important difference between FRAND royalties and other royalties, which perhaps we should think about some more. Alternatively, perhaps one could argue that the parties are no longer similarly situated and thus that the royalty is not discriminatory in effect. Of course, if B can't be required to pay more than A, there won't be any need to recalculate the royalty at all, which conserves on adjudication costs but perhaps at substantial accuracy cost.