Is there a difference between a reasonable royalty and a FRAND royalty? Put another way, should a reasonable royalty awarded for the infringement of an SEP owned by a non-SSO member be different from a FRAND royalty awarded for the infringement of the same SEP if it were owned by an SSO member that had made a FRAND commitment?
At the outset, I’d like to thank Professor Norman Siebrasse for his two previous posts, which have set the stage for this question. My own views are largely in accord with Professor Siebrasse’s, and therefore somewhat at odds with Professor Gregory Sidak’s concerns, as expressed in his recent article, that the ex ante valuation framework threatens to undercompensate owners of standard essential patents (SEPs). Nevertheless, Professor Sidak’s point about the interdependent value of SEPs is an important one that until now has not received adequate attention in the literature.
First, a brief recap of what I believe the correct approach to be, having read Sidak’s paper and Siebrasse’s initial comments on it. Suppose that an SSO is just about to decide which of two competing standards to adopt, X or Y. X will incorporate patents A, B, C, D, and E; Y will incorporate a different bundle of patents. The “ex ante incremental value” of A to Implementer Z is the expected incremental contribution of A to Z’s use of Standard X (which might not get adopted at all). The “ex ante contingent incremental value” is the expected contribution of A to Z's use of standard X, on the assumption that standard X is adopted. Now let's assume the SSO does indeed adopt standard X. The more realistic ex ante state of the world is the one in which Z and the owner of patent A bargain to a license that is contingent on standard X actually being adopted. If so, then the theoretically correct FRAND royalty is the one based on ex ante contingent incremental value. This royalty allows the patentee to benefit from the fact that the patent has been incorporated into the standard, which is a necessary contingency for any royalty to be owing; but it is still an ex ante measure, not ex post Sidak proposes. The problems with ex post valuations—that they make the patentee better off than it would have been but for the infringement, and exacerbate holdup and royalty stacking risks—remain and should be avoided.
Sidak’s proposal appears to rest on his characterization of the SSO as a joint venture, and his concern that if the FRAND royalty is not based on ex post value there will be an insufficient incentive for firms to join the venture. That would be true, however, only if the patentee could expect to recover higher royalties if it was not a member of the SSO and thus not subject to a FRAND commitment; but I don’t see why that should be the case. In other words, I don’t see any reason why a reasonable royalty awarded for the infringement of an SEP owned by a non-SSO member should be any higher than a FRAND royalty awarded for the infringement of an SEP owned by an SSO member; both amounts should be calculated on the basis of ex ante contingent incremental value and therefore should be the same. (That should apply both for royalties awarded for past infringement and for ongoing royalties awarded in lieu of an injunction, but that’s a subject for another post.) In both cases, the correct royalty is the ex ante contingent incremental value: what the parties would have agreed to ex ante, on the assumption that the patent would be incorporated into standard X. On this interpretation, the only thing a FRAND commitment adds to the mix is that it is a commitment and therefore may give rise to some sort of contractual or equitable duty to bargain toward and accept a reasonable royalty. Additionally, being a member of an SSO may confer other benefits, among them the ability to participate in setting the standard and a commitment from other SSO members to license their patents on a FRAND basis (the latter being of concern to operating companies, though not to NPEs). (Sidak mentions, at p.990, that an SEP holder also must take into account “the fixed costs of participation in the SSO,” which might include, for example, the possibility that “failure to disclose a potentially essential patent may subject a firm to antitrust scrutiny, and the costs of defending a potential suit.” Based on the outcome of the Rambus case in the U.S., this may be a fairly remote risk, however. And non-SSO members may get hit with low-probability-of-success antitrust suits for any number of reasons too.) But the basic point is that it should not be necessary to equate FRAND royalties with ex post values in order to induce participation in the SSO.
Going back to the initial hypothetical, another point to note is that the parties could agree ex ante to a license (1) that is contingent on standard X being adopted, and (2) the amount of which is contingent on how much Z actually uses the invention. This second condition is the very definition of a running royalty, as Siebrasse points out, where the rate and the definition of the base are fixed but the value of the base depends on use. I suppose we could call this the “ex ante double contingent incremental value,” though that’s a bit wordy. In theory, it's a matter of no difference whether you use the ex ante contingent incremental value or the ex ante double contingent incremental value. (Although I had long forgotten it, it appears that Roger Blair and I, citing a paper by Sherry and Teece, made something like this point in Rethinking Patent Damages, 10 Tex. Intell. Prop. L.J. at 41-42 & n.203 (2001).) In reality, however, parties often prefer to use running royalties because of the difficulty of accurately predicting the amount of use ex ante. Moreover, as Blair and I discussed in another paper (The Elusive Logic of Standing Doctrine in Intellectual Property Law, 74 Tulane L. Rev. 1323, 1397-1401 (2000)), there may be some consequences in terms of efficiency of production as well.
A final observation is that, as Sidak notes, it is often unrealistic in practice to actually apply an ante framework--too many variables are unquantifiable in reality--and that we need to resort to appropriate heuristics. Based on my initial reading, I think that much of Sidak's analysis of what those heuristics should be makes sense. I plan to reread this portion of his article, however, and may address the matter a little more thoughtfully in my next post.