Some economists have argued that the choice of the royalty base should be largely irrelevant as long as the royalty rate is conditioned on the choice of royalty base so that it reflects the economic value of the patented technology. To support their contention, they have pointed to the decision in Lucent Technologies, Inc. v. Gateway, Inc., which stated:
There is nothing inherently wrong with using the market value of the entire product [as the royalty base], especially when there is no established market value for the infringing component or feature, so long as the multiplier [i.e., the royalty rate] accounts for the proportion of the base represented by the infringing component or feature.
580 F.3d 1301 (Fed. Cir. 2009).
However, positing a "low enough" royalty rate (to account for the relative contribution of the patented invention) is no longer permissible, and the Uniloc case was quite clear on this point: "The Supreme Court and this court's precedents do not allow consideration of the entire market value of accused products for minor patent improvements simply by asserting a low enough royalty rate." 632 F.3d at 1320.
Application of a Strict Standard
Several post-Uniloc courts appear to have applied a strict standard vis-à-vis entitlement to the entire market value (EMV). In Lucent v. Gateway, the infringement allegations were related to a certain feature in Microsoft's Outlook program. The jury awarded Lucent a lump-sum award of $358 million that was later vacated by the Federal Circuit. 580 F.3d at 1336. In the second trial, Lucent introduced a modified damages calculation that first identified the percentage of Outlook customers who use the patented feature and then limited the revenue base to Outlook revenue stemming from these customers. The district court rejected this approach, stating that "Lucent fail[ed] to properly apportion its damages calculation to separate between patented features and unpatented features of Microsoft Outlook," and ordered Lucent to either perform additional apportionment or justify its use of the EMV. In its supplemental report, Lucent introduced a per-unit calculation based on the whole unit price of Outlook and argued that this is consistent with Microsoft's own license practices. The district court decided that while such Microsoft licenses may be relevant to the factors identified in Georgia-Pacific Corp. v. U.S. Plywood Corp., 318 F. Supp. 1116 (S.D.N.Y. 1970), aff'd, 446 F.2d 295 (2d Cir. 1971), an approach based on the whole unit price still failed to account for other unpatented features and that Lucent needed to further apportion its damages calculations.
In Versata Software v. SAP America, No. 2:07-CV-153, 2011 U.S. Dist. LEXIS 102333 (E.D. Tex. Sept. 9, 2011), the district court excluded the reasonable royalty analysis of all three of the plaintiff's experts because their analysis relied on "speculation and guesswork, and was divorced from the factual and economic realities of the case." The district court stated that while the plaintiff's experts did not apply an explicitly stated royalty rate to the EMV of the accused products, the plaintiff's experts "attempted to play mathematical games," and in the end, their methodology amounted to applying "an unsupported percentage to SAP's total revenue in violation of Federal Circuit precedent regarding EMV." Inventio AG v. Otis Elevator, No. 06-CV-5377, 2011 U.S. Dist. LEXIS 88965 (S.D.N.Y. June 22, 2011), is another example in which, despite evidence that the feature was a "substantial basis for demand" for the infringing product and that its absence would put a competitor at a disadvantage, the court struck any entitlement to damages based on the EMV:
It is not enough to present evidence that the patented feature was desirable, or that it played some role—even a substantial role—in the customer's decision to purchase a system containing the infringing product. If the patented aspect of a system containing both patented and unpatented elements creates a "substantial basis for demand," that would tend to support the reasonableness of a higher royalty rate. But as long as other features of a product contributed to the customer's decision, Supreme Court precedent (which the Federal Circuit is powerless to overrule) demands that there be an apportionment of the defendant's profits and patentee's damages between the patented features and the various unpatented features of the "whole machine."
Not all courts have adopted a strict standard, though. The court in Mondis Technology Ltd. v. LG Electronics Inc., No. 2:07-CV-565, 2011 U.S. Dist. LEXIS 78482 (E.D. Tex. June 14, 2011),adopted what may be called a "non-absolutist" approach to the Uniloc court's directive. In Mondis, the district court reasoned that the Uniloc v. Microsoft "basis for customer demand" test could not be "absolute." If it were, on these facts, the patentee would be between Scylla and Charybdis: Ignore the licenses of the patents-in-suit, even though they may constitute "the most reliable evidence" from which to calculate a royalty or try to recast the licenses by inferring the relative contribution of the patented feature in those licensed products, making the "previously comparable licenses non-comparable." However, using the infringer's licensing history to untether the proscription on EMV damages under Uniloc, and thereby render it permissible, has not found favor in other courts.
Successful EMV Approaches
Courts have allowed an EMV damages analysis to go to a jury when evidence shows that the patented feature "substantially creates the value" of the accused product. In Funai Electric Co. v. Daewoo Electronics Corp., the plaintiff employed what may be called a funnel approach that passed muster with the Federal Circuit vis-à-vis entitlement to EMV damages. First, and at the broadest level, Funai's expert presented evidence that there was general industry demand for smaller, faster, more reliable videocassette records. Second, specific evidence was provided that Funai's patented technology satisfied consumer demand for these requirements and that there were no available, noninfringing substitutes. Third, Funai was able to satisfy the Federal Circuit that its technology was the basis for consumer demand. Such a funnel approach is attractive but works only if the evidence at the narrowest point of the funnel actually meets the "patented technology creates the value" requirement. In ActiveVideo Networks, Inc. v. Verizon Communications, No. 2:10-CV-248, 2011 U.S. Dist. LEXIS 91722 (E.D. Va. July 29, 2011), the district court refused to exclude damages testimony based on the EMV, finding that the plaintiff had presented substantial evidence that the patented feature is "the basis for consumer demand for [the infringing system] or substantially contributed to the value of the system."
ActiveVideo is an interesting decision because there may be an underlying tension between "the basis for demand" and "substantially contributed to the value" prongs of the test used by the district court there. The former prong is closer in spirit to the "substantially creates the value" principle; the economic notion that the value of the accused product is a number close to zero absent the patented invention is implicit; whereas, implicit in the latter prong is the relevance of other factors that influence demand, something that the Inventio court used to strike the use of EMV damages.
Damages Approaches after Uniloc
Despite variation in how the courts are applying the Federal Circuit's EMV damages precedents, some tea leaves may be cautiously read: The Uniloc standard is not insurmountable; proving that the patented feature is the basis for consumer demand probably means being closer to "substantially creates" than "substantially contributes"; the more complex and multifaceted a product, the less likely that an economic justification of "substantially creates" will carry the day; the infringer's licensing history is unlikely to be a free pass to an EMV damages entitlement, even in an evidence-poor environment; and there is increased demand from the courts for the real-world data that can readily be found through survey evidence.
Interestingly, some recent damages approaches have departed completely from base and rate determinations and undertaken reasonable royalty analysis under the aegis of bargaining theory. Such approaches essentially circumvent the issue of apportioning the royalty base by using a profit-splitting approach. Such profit-splitting approaches have ranged from being predicated on very well established economics principles, such as Nash bargaining theory, to ipse dixit determinations based on implicit rules of thumb. In the case of Nash bargaining theory, while the economic concept is sound enough to have garnered its originator a Nobel Prize in economics, anchoring the approach to "a record of actual transactions" in the patent damages context has been problematic. Indeed, the logic of the court in Oracle v. Google, No. 3:10-CV-03561, 2011 U.S. Dist. LEXIS 80280 (N.D. Cal. July 22, 2011), is precisely that "the Nash solution cannot describe real-world behavior unless the conditions on which it is premised are satisfied in the real world."
Some commentators have questioned the use of profit splitting by asking whether the approach is justified in the first place. According to this school of thought, the difficulty in apportioning between the fraction of profits attributable to the patent holder's invention and the remainder attributable to the infringer's own contributions (at least in part) led Congress to eliminate the infringer's profits as a remedy for utility patent infringement in 1946. M. Gooding et al., "Reasonable Royalty Patent Infringement Damages after Lucent v. Gateway and Uniloc v. Microsoft: Reports of the Dearth of Patent Infringement Damages Are Greatly Exaggerated," 83 Pat., Trademark & Copyright J. 235 (BNA) (Dec. 16, 2011). Therefore, a profit-splitting approach turns the clock back, as it were, in these commentators' minds; instead, they posit that the proper place to start for a damages expert is a royalty rate and base calculation with the base being apportioned ("rate/base with apportionment") instead of a "naked" profit-splitting approach. Of course, the rate/base with apportionment approach has its own critics, and some have expressed the view that "the apportionment approach of reducing the royalty base treats the symptom (an overly large damages award), without addressing the underlying cause." E. Bailey et al., "Making Sense of 'Apportionment' in Patent Damages," 12 Colum. Sci. & Tech. L. Rev. 255 (June 2, 2011). In their view, the underlying cause of excessive damages calculations is the approach sometimes taken by damages experts of choosing the royalty rate and royalty base independently of each other and without reference to the economic value of the patented technology.
Nevertheless, there are practical issues surrounding a profit-splitting approach if it is to be done correctly. When so-called synergies are present between the patented invention and other features (something that critics of the "rate base with apportionment" approach point to), how does a profit-splitting approach reflect the incremental value of the patented technology to the infringer as compared with the next best alternative? To grapple with this in practical terms, and create a methodology that does not substantively reduce to an unsupported rule-of-thumb approach, is a challenge that is likely to be played out in the courts with varying degrees of success.
On the other hand, market-based evidence—such as statistical or regression analysis, customer surveys, customer interviews, and "marketplace-wide evidence of demand sensitivities"—has been unambiguously sought by the courts. How successfully experts implement market-based approaches, however, will be a function of context as well as the availability of data. A relatively inexpensive customer interview approach in a "simple" product market will not work for a complex, high-tech product for which demand depends on a multiplicity of product attributes, some more important than others. Thus, the use of economists who study consumer demand and marketing experts is likely to grow to calculate patent damages. Surveys—as an affirmative club or as a defensive shield—also are likely to become more important in calculating patent damages. By definition, surveys are market-based evidence; therefore, they can be a powerful tool that is responsive to context-specific treatment that many courts are demanding.
A recent example of a survey-based evidence informing damages calculations is Lucent Technologies, Inc. v. Microsoft Corp., No. 07-CV-2000, ECF No. 1478, at 14–19 (S.D. Cal. Nov. 10, 2011). The Lucent survey approach was roughly based on the following scheme: A set of respondents is surveyed, and the critical question is something like, "If the product did not have feature X, would you still buy the product?" (the assumption being that the patented technology maps onto feature X); a subset of respondents, say Y percent, reveal that they would not buy the product absent the feature X; the patentee then argues that the infringer would have conducted the hypothetical negotiation knowing that if it did not license the patent, it would lose some portion of its revenue, which is a function of Y percent; and the damages base is adjusted accordingly. This approach has the virtue of simplicity and straightforward interpretation. Interestingly though, the Lucent court, while allowing the survey into evidence, implicitly recognized that the "must have" feature (X) may have been one of several "must have" features, even for the small percentage of respondents who would not buy the product without the "must have" feature.
There are, of course, many other approaches to surveys that can capture awareness, incidence of usage, or the relative importance of a feature in a multifeature product. In the latter case, eliciting responses from survey respondents can accurately capture to what extent the feature or features implicated by the patented technology are (or are not) important drivers of demand. However, in all but the simplest of surveys, there may be challenges in properly selecting the population to survey and accurately representing the claim or claims of the patent to the respondents in language that the respondents will understand. Notwithstanding these challenges, surveys, when conducted properly and interpreted correctly, are likely to become increasingly important in aiding the fact finder.
Treating Comparable Licenses
In the wake of Lucent v. Gateway, a patentee that wishes to rely on other license agreements to support its reasonable royalty damages claim has a clear obligation to ensure it can prove that "the licenses relied on . . . in proving damages are sufficiently comparable to the hypothetical license at issue in suit." 580 F.3d.
Comparability between the hypothetical negotiation of a reasonable royalty for the patent at issue and prior licensing agreements must be supported by evidence more detailed than simply being concerned with the same general industry or type of product. The licenses must be comparable economically or technologically, and where a patentee attempts to introduce licenses that are less closely linked economically or technologically to the facts at issue in a case, any differences must be carefully explained by the evidence before a court will allow their use in a reasonable royalty damages analysis. Absent such specific explanatory evidence, the most that the looser, more general license comparisons can be used for is to introduce evidence regarding the general structure of such license agreements, such as a preference for lump-sum royalties as opposed to running royalties.
In ResQNet.com Inc. v. Lansa Inc., 594 F.3d 860 (Fed. Cir. 2010), the Federal Circuit seemed more open to the use of settlement licenses in the calculation of hypothetical negotiation royalties, in the absence of alternative comparable licenses. In ResQNet, the Federal Circuit set aside a $12.5 million reasonable royalty verdict, criticizing the patentee's damages expert for using licenses "with no relationship to the claimed invention" in order to "drive the royalty rate up to unjustified double-digit levels." Five of the seven licenses relied on by the patentee had no relation to the claimed invention; they were re-branding or re-bundling licenses that included software products and source code as well as marketing, training, maintenance, and upgrade support, in exchange for ongoing revenue-based royalties. "Notably, none of these licenses even mentioned the patents in suit or showed any other discernible link to the claimed technology." Id. The two other licenses arose out of litigation over the patents in suit and were also flawed bases for a reasonable royalty analysis: One was a lump-sum agreement (which the patentee's expert could not even translate into a royalty rate), and the other called for a royalty rate substantially less than the 12.5 percent the expert was proposing. Significantly, however, the Federal Circuit remanded the case back to the trial court for reconsideration of the reasonable royalty calculation, observing that "the most reliable license in this record arose out of litigation" and noting that "the district court may also consider the panoply of 'events and facts that occurred thereafter and that could not have been known to or predicted by the hypothesized negotiators.'" Thus, ResQNet appeared to open the door for potential use of settlement licenses in calculating royalty damages.
Subsequently, however, in Wordtech Systems Inc. v. Integrated Networks Solutions, Inc., 609 F.3d 1308 (Fed. Cir. 2010), the Federal Circuit noted that several of the licenses used in a reasonable royalty calculation were signed "after initiating or threatening litigation against the licensees," and the court reiterated that "litigation itself can skew the results of the hypothetical negotiation." In Wordtech, the Federal Circuit held that of the 13 licenses relied on by the patentee at trial, none was sufficient to support the verdict. The Federal Circuit also found that past royalty rates of 3 to 12 percent could not explain the jury's rate of 26.3 percent. The Federal Circuit therefore found that the damages awarded by the jury were "clearly not supported by the evidence" and "based only on speculation or guesswork."
There is no uniform approach by the courts regarding whether litigation licenses may be used to calculate a reasonable royalty rate. Some courts have admitted litigation settlement agreements into evidence, while others have ruled such agreements inadmissible. Or, where the agreements are ruled admissible, they are also deemed to have minimal probative value on the reasonable royalty question. Courts have declared that the admissibility of such agreements must be determined on case by case and have criticized the selective use of a subset of litigation settlement agreements favorable to the claimant's analysis.
Given this variance, at some point, the Federal Circuit will undoubtedly provide additional direction on the issue of litigation licenses. In the interim, patentees who wish to use litigation licenses to calculate royalty rates should be prepared to demonstrate specific facts supporting the claim of comparability between those litigation licenses and the license at issue. Therefore, in a climate where there is, and correctly so, heightened scrutiny of licensing terms and comparability between licenses, licensing experts who have real-world licensing experience are likely to become increasingly important.
Conclusion
Notwithstanding several landmark decisions in the past few years, significant gray areas remain in the calculation of reasonable royalty patent damages. Damages experts will need to look to the Federal Circuit for clarity before they can adopt a stable set of heuristics for doing patent damages "correctly." In a real sense then (and borrowing from Charles Dickens), "we have everything before us, we have nothing before us."
Keywords: litigation, intellectual property, patent damages, entire market value, licenses