Wednesday, June 11, 2014

Reflections on Holdup and Royalty Stacking, Part 1

In connection with some of my pending scholarship, and in preparation for the upcoming Antitrust & IP conference at which I will be speaking in Brussels, I've been giving some more thought to the question of what the terms "patent holdup" and "royalty stacking" should be understood to mean, and how (if at all) they should be addressed.  Here is Part 1 of my current thinking on the matter.  I'll publish another installment later this week, and would welcome feedback.

I begin with the premise that the "ex ante" bargaining framework--that is, attempting to reconstruct the hypothetical bargain the willing licensor and licensee would have made ex ante, before the infringer incurred costs in reliance on the infringing technology--is in principle the correct one for estimating reasonable royalties in general and FRAND royalties in particular, because it awards the patentee for the contribution of its technology to the state of the art, i.e., the value of the technology over the next-best available noninfringing alternative.  By contrast, an "ex post" bargain, conducted after the date on which the infringer has begun infringing, would in addition reward the patentee in part based upon what the infringer would be willing to pay, ex post, to avoid (potentially substantial) switching costs.   The higher royalties or damages that are the consequence of an ex post approach are social costs to the extent they lead to higher prices and lower output, unless on balance they stimulate more benefits (more invention or disclosure) than they do costs; but it’s hard to see why we should assume that the patentee’s ability to extract a portion of the ex post switching costs would generate more social benefits than costs.  (Indeed, in previous work I have expressed the view that, for purposes of calculating damages, we should assume that the substantive patent law strikes the correct balance between incentives and access.  The law of patent damages should support, and not alter, that balance.)  In addition, those higher costs could inhibit some inventive applications of patented technology.

A brief aside:  note that the "ex ante" approach could take one of two forms, both subject to administrability constraints:  a "pure" ex ante approach, in which we try to reconstruct the ex ante bargain the parties would have made, based only on information that would have been available to them ex ante; and an alternative approach in which we try to reconstruct the ex ante bargain the parties would have made based on all relevant information that is known at trial (ex post), such as the amount and duration of the infringing use.  The German courts phrase the hypothetical bargain in the latter manner, that is, what the parties would have agreed to before the infringement began, if they had foreseen the future development and the duration and amount of use of the patent.  (See my book, p.266.)  I used to think the German approach was doubtful as an economic matter, but Norman Siebrasse and I are now working on a paper in which we argue that it is, basically, correct.  Readers will hear more about that in due course.

In any event, in my view the term "patent holdup" can refer to any situation in which the patentee is able, in effect, to extract a license based on an ex post, as opposed to ex ante, bargain.  This may occur if courts automatically award injunctions for patent infringement, because ex post the infringer will pay more than it would have paid ex ante simply to avoid switching costs.  (It also would arise whenever courts award royalties on the basis of an ex post, as opposed to an ex ante, bargain, as they do in the U.S. when awarding postjudgment royalties in lieu of an injunction.  I've explained before why I think this is wrong.)  So defined, patent holdup to some degree could arise virtually anytime a court awards an injunction in a patent case, though I think it would be a mistake to draw any policy implications on this basis alone.  Injunctions have the virtue of encouraging the parties to negotiate over the value of patent (preferably, though inevitably not always, ex ante), and this is a good thing if we believe that the parties generally have an advantage (in comparison with a court) in determining what that value is.  Nevertheless, there may be cases in which those advantages are outweighed by the social costs of holdup.  In particular, where the potential negative consequences of holdup are both large and systematic—as I think they are in the case of SEPs, because of the large number of interdependent patents that are needed to market a standard-compliant device and, relatedly, the difficulty of designing around the standard ex post—a deviation from the general preference for injunctions is, in my view, justified.  In such cases, the potential harm if injunctions are granted may be unusually large and may outweigh the costs emanating from courts’ informational disadvantage in comparison with the parties with respect to valuing patented technology. Moreover, the owner of a FRAND-encumbered SEP is, by definition, willing to license rather than exclude, and benefits from the widespread adoption of its technology resulting from standardization.  (Some though not all of these factors also may weigh in favor of generally denying injunctions to PAEs, as under current U.S. practice.)  For further discussion, see, e.g., Thomas F. Cotter, Comparative Patent Remedies: A Legal and Economic Analysis 59-60 (2013); Thomas F. Cotter, Make No Little Plans:  Response to Ted Sichelman, Purging Patent Law of “Private Law” Remedies, 92 Tex. L. Rev. See Also 25, 27 (2014) (describing the conditions giving rise to patent holdup as “e.g., when the patent reads on a component of a multicomponent device, the ex ante value of the patent is substantially less than the ex post cost of designing around it, and the infringement is inadvertent”); Thomas F. Cotter, Patent Holdup, Patent Remedies, and Antitrust Responses, 34 J. Corp. L. 1151 (2009) (tracing the evolution of the application of the holdup concept to patent law).

An alternative approach would be to define holdup as occurring only in the (perhaps rare) cases in which the patentee‘s demands measurably retard follow-up innovation.  If so, whether or not hold-up is present arguably becomes an empirical, not merely a theoretical or definitional matter; and as an empirical matter, it doesn’t appear that the contemporary smartphone industry is marked by increasing prices or diminishing innovation.  See Alexander Galetovic et al., Patent Holdup:  Do Patent Holders Hold Up Innovation?, Hoover IP2 Working Paper Series No. 14011 (May 2014), available here.  The problem with this narrower definition of holdup, however, is that it seems to assume that just because prices are falling and innovation is increasing there is no reason to be concerned about patentees extracting royalties that reflect, in part, implementers' ex post switching costs.  It seems to me that this is still a social cost, potentially a substantial one in the case of SEPs, even if at the same time technological progress is, fortunately, mitigating some portion of that cost. Put another way, to assume that price reductions and innovation increases necessarily mean that holdup is exacting no social costs is a little like arguing that, because the price and cost of kerosene was falling throughout the late nineteenth century (as it was according to Dominick Armentano, Antitrust and Monopoly:  Anatomy of a Policy Failure 60 (1982)), there was no need for Congress to enact antitrust laws to remedy monopolistic conduct--or, to use a very different analogy, because we are now living in what may be the most peaceful period in human history, see Steven Pinker, The Better Angels of Our Nature:  Why Violence Has Declined (2011), we can eliminate our armed forces and police.  In my view, whether a response to patentee opportunism is desirable or not depends upon how much better, if at all, conditions would have been in the presence of greater competition.  We can’t just assume that conditions are optimal because prices are falling.  But see Galetovic et al., supra, at 20 (“. . . the performance of SEP industries is remarkable by any realistic standard.  So the ‘without holdup it could be even better’ is apparently saying that it could be even better than anything that is normally observed.”).       

To be sure, there are countervailing concerns that one should take into account as well.  First is the possibility of “reverse holdup,” where implementers abuse their lack of susceptibility to injunctions to negotiate a license that is too low to sustain a healthy rate of innovation (though it seems to me that courts can address such gamesmanship by holding out the possibility of awarding injunctions at least in these types of cases, or by awarding enhanced damages and fees).  Second, perhaps the courts’ posited informational disadvantage is more acute than I am assuming it to be; maybe courts systematically err by awarding damages that are less than those that would have resulted from the ex ante bargain.  If so, however, the better response might be to educate them on how to do better, rather than to counsel them to award injunctions in SEP cases.  Note also that if, as I happen to believe is correct, reasonable royalties and FRAND royalties are (or at least should be) calculated in largely the same manner--and injunctions awarded except in cases in which the risk of holdup threatens substantial social costs--there shouldn’t be much of an incentive on the part of SEP owners or potential owners to stop participating in SSOs or to under-declare.  All that a FRAND commitment adds to the mix is that it’s a commitment, which may give rise to additional legal responses such as breach of contract or promissory estoppel for its breach, but I am inclined to think that the amount of infringement damages (and the availability, or not, of injunctive relief) should be the same regardless of whether there is a FRAND commitment in place or not.  As far as injunctions and damages are concerned, consequences should turn on likely economic realities and not on the existence or not of a FRAND commitment.


  1. Hi - thanks for this thoughtful and balanced review. I had three thoughts as I was reading it, and thought I'd jot them down.

    1. If your definition of holdup is when an ex post license imposes social costs that an ex ante license does not, then isn't virtually every ex post license holdup? I've seen several models of the ex ante and ex post problem, and while some explain why we might live with the social cost, I haven't seen any that say - on margin - ex post has less cost. What do we do with policy when every license is holdup?

    2. What is the role of discovery costs? I've seen one model that thoughtfully includes discovery in the mix of determining how to think about ex post licensing, but if we ignore them in our policy (like saying no injunctions ever) do we wind up disincentivizing discovery even if that would be the least cost option?

    3. Notwithstanding my point 1, I'm wondering whether there are times when the ex ante model imposes a higher social cost, perhaps due to uncertainty. I suppose if all licenses were royalty based, perhaps not, because the cost of unused license would be low (unless the license were exclusive). But if the ex ante license is lump sum, I wonder whether it would be better to see whether the patent is actually used prior to licensing.

    Thanks again - I look forward to the next installment.

  2. Good comments. Part of the problem, I think, is that the word "holdup" is kind of loaded--once you use it, it sounds like there's a problem that needs to be addressed, even though (on my view) that may not always be the case. The main idea I want to get across is that any time the licensor and licensee enter into a licensee ex post-that is, after the licensee has invested sunk costs in reliance upon its ability to use the allegedly infringing technology--the license the parties would voluntarily negotiate necessarily reflects switching costs. These are not related to the value of the invention and, in principle, I don't see why the patent owner is entitled to it. In the typical case, though, this probably isn't a good reason to deny an injunction, for reasons I discuss above and have discussed elsewhere (injunctions conserve on informational costs, etc.). It might suggest that, for purposes of calculating reasonable royalties, we should move the date of the hypothetical negotiation to just before the defendant incurred costs in reliance on its ability to use the patented invention, rather than using the date of first infringement, though again I think in the typical case the additional accuracy gained is unlikely to justify the cost. SEPs, though, are the exception that proves the rule: potentially very significant social costs if we grant injunctions and use the date of first infringement as opposed to the date of reliance (the date the standard is adopted), which in my view tip the scales the other way. Since none of this is really quantifiable, though, it comes down to a matter of judgment, not proof. Perhaps that responds to your second point as well; I tend to think that a case-by-case estimate of costs and benefits is not really a good idea, so that a broad presumption in favor of injunctions coupled with some broad presumptive exceptions (SEPs, maybe PAEs, maybe some others), is the best we can practically aspire to.
    As for your third point, I'm not sure I quite follow the very last part ("better to see whether the patent is actually used prior to licensing", but I think the article Norman Siebrasse and I are working on addresses some of what you're getting at. Our bottom line, we think, is that the ideal damages framework is to replicate the ex ante bargain the parties would have struck with ex post information. The ex ante part avoids holdup (in the switching costs sense) but the ex post part helps to ensure that the damages are commensurate with the social value of the invention. A typical running royalty using an ex ante rate and an ex post base does precisely this; the assumption that the parties presumed the patent valid and infringed is consistent with it too..

  3. Thanks for the reply. On point one: "These are not related to the value of the invention and, in principle, I don't see why the patent owner is entitled to it." If you haven't seen it, Scotchmer's book on innovation and incentives discusses the circumstances when you might see ex post licensing but still want to preserve payment back to the earlier inventor. The rate is necessarily higher in some cases.

    On the third point, I was speaking more of actual licensing than damages. Of course, I realize that this is a damages blog, so the "ex ante with ex post info" may work in that context. I was merely querying about ex ante licensing where the licensee does not know whether the patent will actually be practiced (e.g. the product might fail).