The
remedy of lost profits has always seemed to me to be, conceptually, the
simplest of the various damages remedies.
You calculate how many infringing sales the infringer made; estimate how
many of these infringing sales the patentee would have made, but for the
infringement (a number which could range from "zero" to
"all"); and you calculate the profit the patentee would have earned
on those lost sales. To be sure, there
are both underlying policy issues and difficult practical issues surrounding
this simple concept. From a policy
perspective, a recovery of lost profits (coupled with a reasonable royalty on
infringing sales the patentee would not have made, but for the infringement) is
intended to render the patentee no worse off as a result of the infringement,
and therefore to preserve the patent incentive.
I generally think this is a sound policy, though reasonable minds may
differ--for example, see Ted Sichelman's interesting paper in a forthcoming
issue of the Texas Law Review (to which I will be contributing a response in
the review's online supplement), or Norman Siebrasse's thoughtful argument that
deterrence sometimes should play a greater, or at least somewhat different,
role than my work advocates (a point to which I hope to respond in a
forthcoming post). And there certainly
are a number of practical difficulties, as there are with any damages remedy;
there are inevitably tradeoffs between accuracy and administrative
efficiency. One of these difficulties
involves the accurate calculation of the number of sales the patentee would have
made, but for the infringement. This
requires one to determine, among other things, what the infringer would have
done if it had not infringed. Resolving
this issue, in turn, requires an inquiry into the next-best noninfringing
alternative. Suppose, for example, that
the infringer made 100 sales with the use of patented invention A; and that it
could have substituted nonpatented invention B for patented invention A, but in
doing so would have lost 10 sales to the patentee. In other words, 90 consumers were indifferent
between A and B, but the 10 who preferred A to B would have purchased from the
patentee, but for the infringement. The patentee should recover its lost sales
on the 10 and a reasonable royalty on the other 90.
This
is hardly cutting-edge economic reasoning, and it's been part of U.S. law for a
long time. One can also find application
of this reasoning in French case law and, more recently, in cases from
Canada. As I discuss in my book,
however, the U.K. still follows an outdated nineteenth century precedent
holding that courts may not take noninfringing alternatives into account in
calculating lost profits (or in awarding defendant's profits). And Japanese courts have denied the patentee
the opportunity to recover both lost profits on sales it lost to the infringer
and a reasonable royalty on infringing sales it would not have made. Under these precedents, the patentee must
choose one remedy over the other, even though this necessarily renders the
patentee worse off than it would have been, but for the infringement.
Two
recent U.S. opinions (from the Federal Circuit), both authored by Chief Judge
Rader, on the topic of lost profits also raise some questions, at least in my
mind. The first, Presidio Components,
Inc. v. American Tech. Ceramics Corp., 702 F.3d 1351 (Fed. Cir. 2012), was a
case in which the patentee apparently thought it was selling a product that
embodied its patent, but (as it turned out) it wasn't: it was selling a product that was not covered
by the patent in suit. The defendant's
infringing product competed with the patentee's unpatented product, and the
defendant's infringing sales caused the patentee to lose sales on its
unpatented product. The court affirmed
an award of lost profits on those lost sales.
The principle that the patentee can recover lost profits on sales of
unpatented goods that compete with infringing goods was established in
Rite-Hite Corp. v. Kelley Co., 56 F.3d 1538 (Fed. Cir. 1995) (en banc). As I discuss in my book, I think the
principle is probably correct, if the goal is to restore the patentee to the
position it would have occupied, but for the infringement. In Rite-Hite, however, there was (according
to the majority) evidence that the "unpatented" Rite-Hite products
actually were covered by other patents not in suit, and thus were not
noninfringing alternatives available to Kelly.
In Presidio, there is nothing in the Federal Circuit opinion that
indicates whether Presidio's unpatented products were covered by some other
patents or other IP rights. If not, then
it seems to me that those products themselves may have incorporated
noninfringing alternative technology that the defendant could have used, and
thus that a lost profits recovery may not have been appropriate. Perhaps there was good reason not to
consider the unpatented products as an adequate substitute technology, but I
think the court should have said something about this issue.
The
other, more recent opinion, is Versata Software, Inc. v. SAP America, Inc.,
Nos. 2012-1029, -1049 (Fed. Cir. May 1, 2013).
The court wound up affirming a lost profits award of $260 million (and a
reasonable royalty award of $85 million).
For all I know, these numbers may be the right ones, but there are two
aspects of the opinion that I find perplexing.
First, the court says (slip op. pp. 14-15) that it won't consider
defendant SAP's arguments relating to the expert's methodology because those
arguments should be resolved under Daubert (U.S. case law relating to the
conditions for the the admissibility of expert testimony), and SAP didn't
appeal the Daubert ruling. But SAP did
appeal from the final judgment, which should have preserved SAP's objection to
the expert's testimony. So does the
court mean to say that SAP didn't object to the expert's methodology at trial
and therefore is precluded from raising this argument on appeal, or is there
some fine point of civil procedure that I am missing here? Second, the court affirms a reasonable
royalty award, even though "the district court precluded Versata's expert
from presenting a reasonable royalty analysis, and the only evidence for a
royalty award came from SAP's expert."
But Versata, the plaintiff, had the burden of proof, so why is there any
reasonable royalty award at all if it failed to carry its burden? At least, I think Judge Posner would wonder
about that, as evidenced by his opinion last June in Apple v. Motorola . . .
Tom,
ReplyDeleteThis is really a comment on your “Initial Reactions” post, but I don’t have a Wordpress account so I can’t leave my comment there.
I have a more basic problem with footnote 23. If everyone who holds a declared SEP patent is part of the pool, which is the scenario (b) that Robart J accepted as the basis for his calculation (¶ 520), then why shouldn’t the RAND rate be equal to the pool rate, instead of three times the pool rate?
My initial reaction is that the only sound reason for the non-pool RAND rate to be higher than the pool rate is if the pool rate is depressed because membership in pool offers external benefits, E; but I agree with your point that Robart J conflated E with E + IP and I also agree that there is no good basis in the evidence for assuming E = P+.
If we ignore E for the moment, this brings us down to a factor of two, or since the analysis is additive, the addition of an extra P+. This comes primarily from the treatment of A- in the following passage from fn 23 in which Robart J asserts that A- = 1.5xP-:
"As for A-, the court has heard no testimony whatsoever about how much a company like Motorola would have to pay for the pool's H.264 SEP collection if it abstained from the pool. Faced with this lack of evidence, the court infers that the rate Motorola would have to pay would be higher than the pool rate, but not twice as high because some, if not all, of the companies holding SEPs would be subject to the RAND commitment."
If I understand this correctly, Robart J is saying the non-pool rate would be higher than the pool rate (he seems to be saying it would be twice as high, though it is not clear to me where the factor of two comes from), but the non-pool rate would be limited because _some_ of the non-pool SEPs will be subject to a RAND commitment. But I thought the assumption of scenario (b) is that _all_ SEP holders are subject to the RAND commitment: “(b) if Motorola received royalties equivalent to what it would have received if it and the other holders of other readily identifiable H.264 SEPs were _all_ added to the pool with the current pool rate structure” (my emphasis.) Note that the final result is sensitive to the exact factor used. Because A- = 1.5P-, the existing P- in the equation is only partially offset, leaving 0.5P-, which he concludes on the evidence is equal to P+. But if all SEP holders are pool members, then A- = P-, then the two P- terms cancel, and the RAND rate equals the pool rate.
(Continued - my original post was too long.)
(Continued from above) I don’t think this is a technical point. It seems to me that the view that the non-pool rate is higher than the pool rate implicitly reflects some hold-up value to the patents, which is exactly what the RAND commitment is intended to eliminate. This is why Robart J was right to choose scenario (b), in which all SEP holders are pool members, over scenario (a), in which they are not, as reflecting more accurately the RAND commitment.
ReplyDeleteThis passage also has some odd consequences. It implies that the more parties that are subject to the RAND commitment, the lower the RAND rate will be, and conversely, the fewer parties subject to the RAND commitment, the higher the RAND rate will be. The reason for this is that having assumed VP = VA, as A- goes up, A+ must also go up to compensate. But it seems to me that this misses the point, emphasized by Robart J as a “basic principle” that an important goal of the RAND commitment is to ensure wide adoption and that the RAND rate must take into account royalty stacking (¶70-72). The RAND commitment is a solution to the collective action problem that individually each patentee wants to a high royalty, but the overall royalty has to be reasonable or the standard will not be adopted and even if an individual rate is high, the patentee will not make any money because no one will licence. The assumption that as A- goes up, A+ goes up to compensate, instead incorporates the vicious circle that the RAND commitment seeks to avoid.
Am I off base here? If so, can someone explain intuitively why the RAND rate should be higher than the pool rate for a pool in which everyone is participating?
I don't think Judge Robart is assuming that all of the owners of SEPs (assumed to be members of the pool under scenario (b)) are necessarily subject to a RAND obligation. Not every SEP owner is necessarily a member of the relevant standard-setting organization. If all SEP owners are pool members and are subject to a RAND obligation, then I think you are right that the pool rate and the RAND rate should be exactly the same. His assumption that the nonpool rate is 1.5 times the pool rate seems arbitrary, though he says it's based on his weighing of the evidence.
ReplyDeleteAs to whether his analysis allows Motorola to extract some holdup value: maybe, though I guess one consideration would be that Motorola, like any SEP owner, can choose whether or not to join the pool. Presumably it will choose not to join if it thinks that, on balance, it will do better, in terms of net licensing revenue, if it's not a member of a pool. If the pool rate is the RAND rate, then you may as well join the pool . . . but at what point is the pool rate considered the right one? The 802.11 pool was not as widely accepted, and as he points out in para. 558, the less a pool has widespread acceptance, the less relevant it is as an indicator of RAND value.
I suppose you are right, though, that if more firms are not subject to the RAND commitment, A+ goes up; though I'm not sure what to make of it. If there are SEPs out there that are not subject to a RAND obligation because their owners aren't members of the SSO, I can see how the rates they charge for comparable patents could have some bearing on what is considered reasonable--though maybe it shouldn't, since they are, by definition, going to be extracting some holdup value.
I see - he's saying that if everyone but Motorola was a member of the pool, they would not necessarily all be subject to the RAND obligation. Still, it seems to me that what is driving the wedge between the pool rate and the RAND rate is the assumption that those parties who are not subject to a RAND obligation wil charge a higher royalty. That is a reasonable assumption, but why can they charge a higher rate? I don't see any reason except holdup value. Is there an alternative intuitive explanation of why the RAND rate should be higher than the pool rate? I think your very last phrase gets at the core of my problem; we can't use holdup value to set the RAND rate, and that is what I think he is implicitly doing.
ReplyDelete(I should add that overall, and even with this quibble, I think this is a really excellent decision which provides a good framework. Between this and Posner J's decision, the FRAND caselaw has started out very well.)
My comments above were actually intended for your previous post on FRAND Remedies. This comment is on this post. You give the following example:
ReplyDelete“Suppose, for example, that the infringer made 100 sales with the use of patented invention A; and that it could have substituted nonpatented invention B for patented invention A, but in doing so would have lost 10 sales to the patentee. In other words, 90 consumers were indifferent between A and B, but the 10 who preferred A to B would have purchased from the patentee, but for the infringement. The patentee should recover its lost sales on the 10 and a reasonable royalty on the other 90.”
You state that this is well accepted in law and economic reasoning. However, I don’t see why the patentee should recover a reasonable royalty on the other 90. If, in the but-for world, the defendant would have sold the non-infringing product B, why should the defendant pay a royalty on a product which is not infringing? If the but-for world had been the real world, the defendant would not have paid a royalty on product B; why should it be presumed to do so for the purpose of damages calculation? If a reasonable royalty is awarded in this case, it will be supra-compensatory. In my view, reasonable royalty damages are properly awarded only when the defendant would have made infringing sales in the but-for world. This would happen when the defendant sold a patented product which the customers actually preferred into markets, or in individual sales, that were not available to the patentee, so that it would be rational for the patentee to license to the defendant and split the incremental profits. It is not justified when the defendant would have made non-infringing sales.
The result you describe statutorily mandated in US law, which requires “damages adequate to compensate for the infringement, but in no event less than a reasonable royalty for the use made of the invention by the infringer, together with interest and costs as fixed by the court.” 35 U.S.C. § 284. “In no even less than” as a qualifier on “compensate” implies that a reasonable royalty is required even if it is supra-compensatory. I think the US statute is an over-generalization from the award of a reasonable royalty in respect of sales that would have been infringing even in the but-for world, as that is the scenario in which the reasonable royalty award was originally developed. (For example, the early leading UK case of Penn v Jack. (1867) L R 5 Eq. 81 was one in which the patentee did not practice its invention at all, but exploited its patent entirely by licensing.) Neither Canada nor the UK has a similar statutory provision, and in my view, in those jurisdictions it would be wrong to award a reasonable royalty as compensatory damages in your example.
Interesting point, but I have to disagree on this one. Here's my thinking. In the but-for world, if the defendant had not infringed, it would have used (in my example) B instead of A. Defendant should pay market value for what it took, in this instance a license to use A instead of B. That's how I conceive of a reasonable royalty. The value of that license could be substantial, if for example A enables the defendant to produce its end product at much lower cost; or it could be minimal (even zero), if A is literally no better than B, which may be what you're assuming is the case in my hypo (in which case, I should have drafted the hypo more clearly).
ReplyDeleteYes, I took your example to mean that A is literally no better than B for the 90 customers. In that case, I think that US law still would require a reasonable royalty, and it is wrong to do so.
ReplyDeleteIt may be that U.S. law would require a royalty under that circumstance, in which case it be wrong (economically) to do so. To my knowledge there is no case law specifically addressing whether, in such a case, we have to interpret the statute literally and award something. Judge Posner's opinion in Apple v. Motorola, however, might be read to indicate that you can't go to trial without admissible evidence that you have suffered some harm, so perhaps the statute doesn't need to read quite so literally. But we'll see if the Federal Circuit agrees.
ReplyDeleteI'm really enjoying these discussion we're having!
Likewise!
ReplyDelete