les Nouvelles November 2017 Article of the Month
The Practicalities And Pitfalls Of The Smallest Saleable Patent Practicing Unit
Doctrine: A Review Of Teece And Sherry
Charles River Associates,
Vice-President, and an
Adjunct Professor at
Northwestern University School of Law, Chicago, IL
In early 2016, David Teece and Edward Sherry released a new paper assessing the economics of the "Smallest Saleable Patent Practicing Unit" (SSPPU) doctrine.1 The doctrine was first espoused in 2009 by Judge Randall Rader in Cornell v. Hewlett Packard.2 In the simplest terms, the SSPPU doctrine calls for setting the revenue base for reasonable royalty patent infringement damages at the smallest possible product level that still reflects the patented invention. For example, if patented feature A is sold in two products, ABCDE and ABC, then product ABC (not including features D and E) would likely form the SSPPU (it is a rare instance that product A would be sold on a standalone basis).
Since Cornell, two courts have explicitly required the SSPPU approach in the assessment of damages.3 In 2014, in Ericsson v. D-Link, the Federal Circuit upheld a damages award based on end product revenues, establishing that following an SSPPU approach is not mandatory nor is it the only means of achieving reasonable damages tied to the value of the asserted patents. Rather, the Ericsson ruling clarified that SSPPU is "an important evidentiary principle" used "to help our jury system reliably implement the substantive statutory requirement of apportionment of royalty damages to the invention's value."4 (The "statutory requirement of apportionment" referred to here is Garretson, the seminal case establishing the need to apportion revenues to reflect the accused technology in damages calculation.)5 The following year, in its ruling for CSIRO v. Cisco, the Federal Circuit reaffirmed its stance, arguing that "[t]he rule Cisco advances— which would require all damages models to begin with the smallest salable patent-practicing unit—is untenable. It conflicts with our prior approvals of a methodology that values the asserted patent based on comparable licenses."6 Thus far, however, the Federal Circuit has not expressed any views on use of the SSPPU doctrine outside of jury cases.
In their new paper, Teece and Sherry walk through the justifications expressed in support of applying the SSPPU doctrine and discuss the assumptions embedded within those justifications. The authors also explain a number of the limitations of the doctrine, both logical and practical. In this brief review, I summarize the key findings reported in the Teece and Sherry paper and highlight the policy implications.
My review proceeds as follows. Section II provides a succinct summary of each argument made in support of the application of SSPPU and evaluates their logical underpinnings. Section III then covers some of the practical difficulties that arise in applying the SSPPU doctrine. Section IV concludes with the policy implications. The bottom line: Teece and Sherry argue that a compelling case has not been made in support of automatic or broad application of the SSPPU evidentiary rule in damages calculations, and the rule's limitations imply that in some cases it could do more harm than good to social welfare.7
II. The Pros and Cons of SSPPU
Three justifications appear to support the emergence and development of the SSPPU doctrine: correcting cognitive bias among juries, facilitating revenue apportionment, and (relatedly) restricting damages to the footprint of the patented technology. While each of these arguments has its appeal, each also has its shortcomings.
A. Does Cognitive Bias Run Only in One Direction?
The most frequent argument made in support of a broad application of the SSPPU doctrine has a basis in juries' psychological limitations, referred to as cognitive biases. Two such biases are of particular interest in assessing damages: "framing" and "anchoring." Framing refers to the tendency of people to respond to a particular choice in differing ways depending on how that choice is presented. Anchoring refers to the tendency of people to rely too heavily on the first piece of information they receive (the "anchor") when making decisions. Thus, the concern in patent litigation is that if a jury is presented with revenues on a complex (multi-component) end product when the patented technology at issue in the case reads only on a narrow subset of that product's features, then the jury is likely to award damages that are "too large," exceeding the value that the patented technology provides to the infringer. Put differently, juries may anchor damages on the infringer's overall revenues, pushing damages upwards and presumably past any reasonable level given the patents in suit.
As Teece and Sherry point out, however, framing and anchoring work in both directions; a jury can be biased upwards, toward excessive damages that overcompensate the patent holder, or biased downwards, toward too low damages, so that the patent holder is undercompensated for its technical contribution. An experiment published in the cognitive science and behavioral economics literature illuminates this point: One group of high school students were given five seconds to assess the value of 1 × 2 × 3 × 4 × 5 × 6 × 7 × 8, while a second group of high school students was given five seconds to assess the value of the same expression, but in reverse order (8 × 7 × 6 × 5 × 4 × 3 × 2 × 1).8 The median estimate for the ascending sequence was 512 and the median estimate for the descending sequence was over four times larger at 2,250, but the correct answer is 40,320. In other words, the reported results reflected framing/anchoring—but both estimates fell well below the correct answer.
Since framing and anchoring can work to either increase or decrease an individual's damages determination, Teece and Sherry question why proponents of an SSPPU approach have only considered one side of the issue. The SSPPU approach deals only with upward bias and may, in fact, exacerbate downward bias. But why should we, as a general matter, be worried only about overcompensating patent holders, and not about undercompensating them? While the courts certainly had concerns regarding overcompensation in the cases explicitly calling for the use of SSPPU, there is no empirical evidence establishing that patent holders are commonly or systematically overcompensated (to the best of Teece and Sherry's, or my, knowledge).9 Without such evidence, there is no justification for courts to take regular (instead of case-by-case) corrective action to lower damages on the whole. Recognizing that framing and anchoring biases matter does not lead to support of an automatic application of SSPPU; rather, recognition of cognitive biases suggests that courts should carefully choose which frame to use in any given patent damages matter, a question that needs to be addressed on a case-by-case basis.
B. Can the Smallest Component Fully Capture or Reflect Patent Value?
The second argument that appears to buttress the use of an SSPPU approach for damages relates to apportioning the accused product's selling price to the patented features in relation to other features not covered by the patent but that contribute to product value. Using a component, as opposed to the end product, can narrow the feature set to more closely tailor the accused revenues to the patented invention. For instance, in the Cornell case, the patented technology contributed to computer processors, which are inputs into "CPU bricks," which themselves are inputs into computers. When extraneous features are removed from the damages base, apportioning the remaining revenues may be easier or more straightforward.
This argument is appealing, but incomplete. The first complication is what the asserted patent claims. If the asserted patent covers system claims (and those claims are found to be valid and infringed), then sub-system components will not infringe the patent (at best, there might be contributory infringement). In this case, the system revenues would be most closely tied to the footprint of the technology, and any concerns about overcompensation due to the narrow contribution of the patented technology could be dealt with through the royalty rate applied to the base, as opposed to the base itself. Teece and Sherry observe that, at least mathematically, "one can perform the apportionment required under Garretson by either adjusting the base or adjusting the rate (or some combination of the two)."10 Keeping potential jury bias in mind, it is nevertheless crucial to understand that both the base and rate work together to determine appropriate damages, tied to the footprint of the technology. As such, both the rate and the base should be set in conjunction with one another; a relatively small base might call for a relatively higher rate, and vice versa, in order to arrive at appropriate compensation for the use of the patented technology, as different combinations of rates and bases can yield the same damages payment.11
The second complication is the value of the particular use of the patented technology at issue. In short, value depends on context. Teece and Sherry explain that entities at different levels in the production chain will capture different portions of the overall value of using a given patented technology, as ultimately reflected in the end product: "The farther 'downstream' in the value chain one goes, the greater the cumulative fraction of the total value of using the patented technology that is captured in sales at that level and 'higher' levels in the value chain." As a result, relying on the price of some small upstream component as the damages base may distort the calculation of damages for use of the patented technology further down the production chain.
To illustrate this point, Teece and Sherry discuss semiconductor chipsets within end user mobile communication products, like smartphones and tablets. The chipset market is highly competitive and commoditized, and few chipset makers pay royalties for the intellectual property incorporated into their chips.12
These market features mean that chipset prices and revenues will not reflect the value of the patented technology used within the chipset, let alone the value of using the patented technology further downstream. As Teece and Sherry explain, "the value that the chipset supplier puts on being able to use the patented technology is likely to be only a small fraction of either (a) the cumulative value, at all levels of the value chain, of being able to use the patented technology rather than the available alternatives; or (b) the value 'downstream' from the chipset supplier, at the cellular device/cellular service/end-user level."
C. Do Royalties Set on Product Prices Necessarily Overreach the Value of the Patent?
A third argument in support of taking the SSPPU approach to damages is a variation of the overcompensation concern described above. In brief, the idea is that patent holders should not charge different royalties for the "same amount of use" of the patented technology. Percentage running royalties lead to higher damages for higher priced products, even though the higher price may be driven by features unrelated to the patented technology. Sticking with the Teece and Sherry smartphone example, Apple sells its iPhone 6s with several different levels of memory, with higher memory costing more: a 16GB handset has sold for $649, 64GB for $749, and 128GB for $849. Suppose the patent in suit covers mobile radio connectivity technology unrelated to storage and memory technologies. If the patent holder calculates damages using a 4 percent royalty rate applied to the product price, the accused infringer would pay $4 more for the 64GB handset as compared to an otherwise identical 16GB handset, despite the fact that the connectivity technology remained constant across the two products. Suppose, on the other hand, that a court applying the SSPPU doctrine set the royalty base as the price of the 16GB handset as applied to the total number of units (regardless of those unit's memory capacity). This approach would strip out at least some variation in price due to unrelated features (e.g., memory differences).
Again, this argument has much appeal and could make sense under certain circumstances, but it is incomplete. What the argument misses is the possibility of synergies across features. Even though a feature may be technically distinct from the patented technology, commercial links might lead to synergies that imply a higher value for the patents in the presence of the other features. Teece and Sherry again use smartphones to illustrate the point:
…having more memory in a cellphone enables the cellphone to handle larger files in a way that enhances the value of cellular connectivity, and having cellular connectivity likewise enhances the value of having additional memory, because the user can do more with the information contained in the memory if the device is capable of cellular connectivity than if it is not.
In another example showing the potential for relationships across what might appear to be even more disparate technologies, Teece and Sherry observe that:
…adding a camera to a cellphone increases the range of ways that the owner can make use of the cellphone; the owner can now take photos and share them with others over cellular connections in a way that the owner of a camera-less cellphone cannot. This increases the value to the owner of having cellular connectivity. Similarly, the ability to share photos with others over a cellular network enhances the value of the camera functionality, as compared to the value of a stand-alone camera without cellular connectivity that is not capable of such sharing. In other words, even though the two features are technologically unrelated, adding the camera functionality enhances the value of cellular connectivity, and vice versa.
Because the patented feature does not "drive demand" in either of these examples, neither would clear the bar set by the entire market value rule (EMVR),13 even though synergies are present. Since it ignores commercial interrelationships, limiting the royalty base to a narrow component, like the chipset, is likely to bias the damages calculation downward whenever synergies are present. Chipset revenues will not reflect the enhanced value the smartphone maker may receive through the combination of the patented technology with the other features that it affects. Thus, applying the SSPPU approach when the patented technology in suit involves synergies, like those illustrated above, will likely lead to under-compensation for the patent holder.
III. Practical Complications in Applying SSPPU
In addition to the above inherent limitations, Teece and Sherry also highlight a number of problems in interpreting the SSPPU doctrine. These ambiguities in turn lead to difficulties in applying the SSPPU approach in practice.
A. What is the "Smallest" Unit?
As an initial problem with implementing the SSPPU, Teece and Sherry point out that the courts have not provided any guidance on how to identify a "smallest" unit. In some cases there may be multiple choices for units "smaller" than the end product. For example, suppose that a patent holder is asserting a patent that covers feature A in the complex end product ABCDE. Further suppose that the maker of ABCDE also supplies less complex products or components in the form of ABD, ACD, ACE, ABE, and ADE. If the value A contributes differs in each of these smaller units, which should we select as the basis for determining the value that A contributes to the larger product ABCDE? Should some average across products/components be used instead? Moreover, with any of these smaller units, the task of apportionment will still remain; SSPPU does not necessarily avoid apportionment, nor necessarily makes it easier.
Alternatively, a proposed smaller unit may be an input for the accused infringer's end product, in which case it will buy the component for use in production but will not price it for sale. In this case, the smaller unit "price" reflects the cost to the end product maker of using the input (and the value the input maker obtains from the patented technology, if it pays for the patent), but we cannot assume that the input maker's price fully captures the value of the patented technology for the end product maker. The Wi-Fi chipsets relied upon by the district court in In Re Innovatio fall into this input category. As explained above, the value contributed by a patented technology is likely to differ along the production chain, so it is unclear why an input cost for the accused infringer would form a more relevant damages base than an output price, particularly when damages are traditionally understood to be based on some portion of the accused infringer's sales.
B. Does Saleable Mean Actually Sold?
The second "s" in SSPPU stands for "saleable." What, exactly, does that mean? Must the smaller product or component be sold separately in a market so that objective prices are available? Or is it enough that the smaller unit could be sold, even though it is not actually sold? A hybrid scenario held in the Cornell case: the district court chose to use computer processors as the damages base, though of the eight processors accused of infringing the patented technology, only three had market prices and were sold independently, so "prices" for the other five models that were not sold had to be imputed. Teece and Sherry observe that in this case, "reducing the 'framing bias' comes only at a cost in accuracy." If the goal of employing an SSPPU approach is to arrive at a damages calculation that better reflects the value the patented technology brings to the accused infringer, is pushing that calculation to imputed, as opposed to actual, price and revenue data really achieving the desired end goal?
As a further problem, Teece and Sherry argue that pushing the SSPPU doctrine and at the same time maintaining the "hypothetical negotiation" damages framework established under Georgia Pacific has led to "internal schizophrenia" in patent damages law.14 The hypothetical negotiation framework calls for paying close attention to real-world licensing considerations and assessing what the parties could actually have agreed to, had they negotiated willingly prior to any infringement. But in practice, the vast majority (if not all) available licensing agreements are likely to be on the basis of the end product prices for the goods that licensees sell incorporating the patented technologies, not on components that the licensees purchase as inputs and not on sub-products that are never sold in any marketplace. Furthermore, in the real-world, the parties care not only about reasonableness but also about administrability, a concern that precludes the use of imputed prices or the use of any royalty base with no objectively reported data.15 In short, real-world negotiations tend to limit royalty bases to products tracked in the normal course of business, but these practical concerns have thus far gained little to no traction among proponents of the use of the SSPPU doctrine, leading to the schism Teece and Sherry point out.
C. What Does "Practicing" Mean?
As a third and final pragmatic concern, how should "patent practicing" be interpreted? At least two alternatives are possible: the literal interpretation, meaning the smallest unit that satisfies all of the asserted claims of the patents in suit, or a less stringent interpretation, such as the smallest unit that simply contains the "inventive element" of the asserted patent claims.
In the In Re Innovatio case, the district court followed the latter course. More specifically, the court used the Wi-Fi chip selling price as the damages base for the accused wireless network access, though Claim 1 of the asserted patent (U.S. Patent 5,844,893) lists a method claim applicable to "a data communication network having a plurality of mobile transceiver units selectively communicative with a plurality of base transceiver units." Since a chip is not a system and does not have either a plurality of mobile transceivers or a plurality of base transceiver units, a Wi-Fi chip will not satisfy Claim 1 of the asserted patent. Instead, the court reasoned that "Because the purpose of a Wi-Fi chip is, by definition, to provide 802.11 functionality, determining the importance of Innovatio's patents to the 802.11 standard also determines the importance of those patents to the Wi-Fi chip."16
Teece and Sherry note that patent claims can be written in a myriad of ways, so requiring the SSPPU to infringe each of the claims could lead to patent holders "gaming the system" through creative claims drafting. That being said, the further the SSPPU doctrine strays from the precise claims language in the asserted patent, the less guidance we will have on identifying and applying appropriate SSPPUs for calculating reasonable damages.
IV. Concluding Remarks
Teece and Sherry (2016) calls into question both the fundamental underpinnings of the SSPPU doctrine and the practical application of the doctrine in real-world litigation cases. They clarify that cognitive biases, like framing and anchoring, can run in both directions so courts should be considering which frame or anchor makes the most sense for a given set of case facts, rather than single-mindedly focusing on overcompensation. Teece and Sherry also calls into question the use of input costs as a royalty base, in light of the fact that the value of patented technology will differ along a production chain. When commercial synergies exist, the use of component costs, as opposed to output prices, can be even more problematic. Finally, they discuss a number of practical concerns that make implementing the SSPPU doctrine in a sensible way a challenging endeavor. The many limitations raised by Teece and Sherry caution against a broad, automatic application of the SSPPU doctrine, and suggest a more careful use of the doctrine when it is applied. ■
Available at Social Science Research Network (SSRN): https://ssrn.com/abstract=2855148
- David J. Teece, and Edward F. Sherry, "On the 'Smallest Saleable Patent Practicing Unit': An Economic and Public Policy Analysis," Tusher Center, U.C. Berkeley Working Paper, January 2016, available online at http://innovation-archives.berkeley.edu/businessinnovation/documents/Tusher-Center-Working-Paper-11.pdf.
- Cornell University v. Hewlett-Packard Company, 609 F. Supp. 2d 279 (N.D.N.Y. 2009).
- In re. "Innovatio IP Ventures," LLC Patent Litigation, 1:11cv-09308 (N.D.Ill. 2012); and Apple Inc. v VirnetX Inc. et al, U.S. Federal Circuit Court of Appeals, No. 2013-1489.
- Ericsson, Inc. v. D-Link Sys., Inc., 773 F.3d 1201, 1226 (Fed. Cir. 2014).
- Garretson v. Clark, 111 U.S. 120, 121 (1884).
- Commonwealth Scientific and Industrial Research Organisation v. Cisco Systems, Inc., No. 6:11-cv-00343-LED (Fed. Cir. 2015), at 14.
- The social welfare shortcomings of the SSPPU doctrine is the focus of another recent publication as well; see Froeb, Luke and Mikhael Shor, "Innovators, Implementes, and Two-sided Hold-up," The Antitrust Source, August 2015.
- See A. Tversky, and D. Kahneman, "Judgment under Uncertainty: Heuristics and Biases," 185 Science, New Series, No. 4157 (September 27, 1974), 1124–1131, available at http://www.hss. caltech.edu/~camerer/Ec101/JudgementUncertainty.pdf.
- Overcompensation appears to be a key concern behind Judge Rader's creation of the SSPPU, but that one case says nothing of systemic overcompensation in patent infringement cases warranting wholesale application of the SSPPU doctrine. Moreover, it is unclear why the royalty base is the best tool for correcting risks of overcompensation when they arise (as discussed further below).
- See also the Federal Circuit's Ericsson ruling: "Making real world, relevant licenses inadmissible on the grounds D-Link urges (that they are based on multi-component products) would often make it impossible for a patentee to resort to license-based evidence. Such evidence is relevant and reliable, however, where the damages testimony regarding those licenses takes into account the very types of apportionment principles contemplated in Garretson. In short, where expert testimony explains to the jury the need to discount reliance on a given license to account only for the value attributed to the licensed technology, as it did here, the mere fact that licenses predicated on the value of a multi-component product are referenced in that analysis—and the district court exercises its discretion not to exclude such evidence—is not a reversible error." Ericsson, 2014, at 128.
- Teece and Sherry point out that the Cornell ruling "committed a significant conceptual error" in lowering the royalty base without considering whether any changes were needed in the royalty rate as well.
- Common industry practice in the high tech sector is for all royalties to be paid at the end product and thus as a general matter patent holders have not sought licenses from chipset makers. The result is that chipset prices are unlikely to incorporate or reflect the patented technologies they embody.
- According to the Rite-Hite decision, damages can be based on the value of unpatented features only if the "patent-related feature is the basis for consumer demand." Rite-Hite Corp. v. Kelley Co. Inc., 56 F.3d 1538, 1549-50 (Fed. Cir. 1995) at 1549. The more recent Ericsson ruling softens the Federal Circuit's EMVR stance: "When the accused infringing products have both patented and unpatented features, measuring this value requires a determination of the value added by such features… The essential requirement is that the ultimate reasonable royalty award must be based on the incremental value that the patented invention adds to the end product." Ericsson, 2014, at 1226. For an interesting discussion of the history and evolution of EMVR, see Fahrenkrog, Aaron R., John K. Harting, Christine Yun Sauer and Logan Drew, "Farewell, Entire Market Value Rule," Law360, April 16, 2015, available at http://www.law360.com/articles/634837/farewell-entire-market-value-rule.
- Georgia-Pacific Corp. v. United States Plywood Corp., 318 F. Supp. 1116, 166 U.S.P.Q. (BNA) 235 (S.D.N.Y. 1970).
- Basing damages on profits pose the same problem. Because profits are calculated by firms in many different ways, they can be manipulated, loaded with all manner of overhead in order to report lower profits as a means of reducing royalties.
- 16. In re. "Innovatio IP entures," LLC Patent Litigation, 1:11cv-09308 (N.D.Ill. 2012).