Note: I thought I published this post on Friday, and was dismayed to discover today that it wasn't up. Sorry for the technical glitch!
Today, Professor Siebrasse and I offer some further comments to conclude this series on Gregory Sidak’s recent article on FRAND.
Cotter: I could probably write several more posts on various aspects of Professor Sidak's paper, but to wrap matters up I’ll focus on just one: the use of appropriate heuristics as a proxy for the ex ante incremental value of a patent.
As I mentioned in my previous post, I agree with Sidak—and have stated elsewhere—that it is often impractical to quantify ex ante incremental value (that is, the value that a user would expect to derive from the use of patent X in comparison with the next-best available noninfringing alternative). Litigation occurs months or years after the relevant date (typically the date on which infringement began, though in the SEP context the date just before the standard at issue was adopted is preferable). It may be difficult to determine what the next-best alternative was; what it would have cost to acquire it (and here I agree with Sidak that this should be a relevant consideration, and have so stated in previous work); what its expected net benefits would have been; and so on. Matters are only compounded if the value of the patent depends in large part on the adoption of other complementary technologies, as in the SEP context. Nevertheless, I think it is important to recognize the ex ante incremental value (or, in the present context, the “ex ante contingent incremental value,” as discussed in the two previous posts) as the ideal starting point: it is the standard that renders neither the patentee nor the infringer any better or worse off than they would have been, but for the infringement. (Further modifications can then be made to provide for adequate deterrence or to satisfy other goals, if that is desirable.) Recognizing that it may be difficult or impossible to meet the ideal, one can then try to find alternative measures that might be reasonable substitutes. Here, the point is to use substitutes that come as close as we think we reasonably can to approximating ex ante contingent incremental value, and to avoid substitutes that would inflate FRAND (or reasonable) royalties to capture ex post value.
Oddly enough, then, I agree with much of what Sidak has to say about such substitute measures as comparable licenses (pages 1002-07) and (at pages 1011-14) Greg Leonard’s “top-down” approach as employed in the Innovatio IP Ventures LLC litigation (which I blogged about here). I need to think more about Sidak’s endorsement of “approved contributions” as a better way of estimating the value accruing to an individual SEP owner, over a methodology that applies Marc Shankerman’s 1998 estimate that 10% of all electronics patents accounted for 84% of the value of all such patents. I understand his objections to the latter (the study is relatively old and doesn’t involve SEPs), but I’m not sure I would sign off on the proposition that all of the patents contributed by a firm with a large number of approved contributions (e.g., Ericsson) should be classified as falling into the top 10%, rather than considering them individually (see p.1039). On the other hand, Sidak’s critiques of some of the other possible approaches (e.g., Shapley values, reliance on patent pools that simply divide up royalties by counting patents rather than by estimating their strength) seem at first blush to be pretty sound.
Bottom line: while I disagree strongly with much of Sidak’s theoretical analysis, I concur with a fair amount of his practical advice. Perhaps that puts me in company with Judge Robart, whom Sidak castigates time and again for assuming that the ex ante approach is theoretically correct—only to concede (at p.986) that Judge Robart also concluded that the ex ante approach “lack[s] . . . real-world applicability.” At the end of the day, I still don’t find a huge difference between what Judge Holderman (whom Sidak praises) and Judge Robart (whom he criticizes) were trying to do. Compare In re Innovation IP Ventures LLC Patent Litig., 2013 WL 5593609, at *4-8 (N.D. Ill. Oct. 3, 2013) (adopting Judge Robart’s methodology, with modifications reflecting the differing postures of the two cases, i.e., breach of contract versus patent infringement), with Microsoft Corp. v. Motorola, Inc., 2013 WL 2111217, at *20 para. 113 (W.D. Wash Apr. 23, 2103) (“reasonable parties in search of a reasonable royalty rate under the RAND commitment would consider the fact that, to induce the creation of valuable standards, the RAND commitment must guarantee that holders of valuable intellectual property will receive reasonable royalties on that property”).
Siebrasse: I also agree with much of what Sidak says about specific heuristics, but in this final comment I will look at one point where I disagree with Sidak, and perhaps also with Professor Cotter. The question is whether the cost of the next-best alternative should be taken into account in determining a reasonable royalty.
Sidak says (936):
to measure a licensee’s willingness to pay for the patent accurately, the profit generated using A must be compared with the profit from using B, including the costs of acquiring the rights to technology B.
He provides the following example (937):
Suppose that, relative to using a non-patented alternative, the use of patent A leads to increased revenue of $300, patent B leads to increased revenue of $200, and patent C leads to increased revenue of $100 [but] the licensing cost of patent B is $150, whereas the licensing cost of patent C is only $25.
Sidak concludes that Patent C, not B, is the next-based alternative to A, because C is more profitable net of acquisition costs, and “[i]f one neglects to include the costs of acquiring the lawful rights to use the next-best alternative in the calculation of the incremental value, then the analysis could understate or overstate the prospective licensee’s actual willingness to pay” (937-38).
But why is B asking $150? Suppose A does not exist, and the choice is between B and C. B asks $150, and the licensee (L) says no because C is more profitable net of the acquisition costs of B. But that means A walks away, and B gets $0. In principle, B would revise its offer downwards to incrementally less than $125, so that L would licence from B rather than C, making both L and B better off than if B licensed from C for $25. C would presumably respond by lowering its price, but in the end C would not be able to match B, and L would licence from B for $100. More fundamentally, the point of considering the non-infringing alternative is to capture the economic value of the patented invention, which is the economic value of the invention over the next best alternative. This requires ignoring the cost of licensing the alternative in determining which technology is the alternative.
Consequently, my view is that in principle the cost of acquiring the next-best alternative is irrelevant in assessing the patentee’s willingness to pay. But what difference does this make to the reasonable royalty determination? As Sidak points out, it affects the licensee’s maximum willingness to pay, which sets the upper bound on the bargaining range in the hypothetical negotiation. If the reasonable royalty determination turns on a mechanical split between the maximum willingness to pay and the minimum willingness to accept, as in a 50/50 split proposed as a Nash bargaining solution by some experts post-Uniloc, then whether the cost of the non-infringing alternative is included does directly impact the reasonable royalty. But a Nash bargaining solution is very artificial (and has been rejected for that reason by at least some courts). There is no particular reason, other than the lack of better evidence, for any particular split between the licensee’s maximum willingness to pay and the licensor’s minimum willingness to accept. It may be that on the facts, B charged $150 for transaction cost reasons, the costs of negotiating an individual royalty with L are greater than the benefits (if L is using only a small volume), so that in fact L’s next-best alternative is to licence from C – and yet A has little bargaining power, so that in fact A would have ended up with a royalty of only $30, and L would have captured the vast majority of the incremental value of the patent. For example, perhaps it is clear that A routinely licensed to all comers for $30. That would be the reasonable royalty, regardless of L’s theoretical maximum willingness to pay, and regardless of what a licence from B or C might have cost.
The larger point is that there is an important difference between principles, heuristics and facts. If we have little or no factual evidence, and must decide a reasonable royalty on the basis of principles alone, then my view is that the cost in fact of the next-best alternative should not be taken into account, because in principle, that cost might be bargained down. On the other hand, if we have a rich set of facts, then the principles become little more than a sanity check, and it probably doesn’t matter whether or we take the cost of the non-infringing alternative into account.