1.1 Preliminary Matters
This section will briefly describe (a) the extent to which reasonable royalties are awarded in the major jurisdictions for which descriptive statistics are available; (b) the principal theoretical justifications for awarding them; and (c) at a very general level, the principal methods for calculating them.
1.1.1 Empirical Literature
The empirical literature on reasonable royalties consists largely of descriptive statistics reporting median, average, or largest-ever patent damages awards for selected countries. These statistics provide insight into different jurisdictions’ approaches and priorities related to awarding damages.
The most extensive literature on this subject pertains to the United States. According to a 2014 Lex Machina Patent Litigation Damages Report, for example, in 708 U.S. patent cases filed and terminated from 2000 to 2013, district courts awarded over $8 billion in reasonable royalties, slightly less than $3 billion in lost profits, and slightly more than $2 billion in compensatory lump sum damages for which “the specific sub-type (reasonable royalties or lost profits) is not specified or the apportionment of the award between sub-types is not specified.” Lex Machina’s list of median reasonable royalty, lost profit, and compensatory lump sum awards from 2000 to 2013 indicates that reasonable royalty awards are more common than lost profits awards, but that in some years the median lost profit award exceeded the median reasonable royalty award. Lex Machina’s Patent Litigation Year in Review 2016 reports median reasonable royalty damages in 2016 of $3,552,600, based on thirty-six cases; median lost profits damages of $1,631,231, based on eight cases; and median “Other/Mixed Damages” of $67,785, based on eighteen cases.
PricewaterhouseCoopers (PwC) also publishes annual patent litigation reports. However, PwC reports median patent damages awards in the United States (excluding summary and default judgments) without separately accounting for lost profits and reasonable royalties. Interestingly, PwC’s reported median award for 1997–2016 ($5.8 million in 2016 dollars) is considerably higher than the medians reported by Lex Machina for 2000–2015, most likely due to methodological differences between the two studies. PwC also reports that in 80 percent of the cases in which courts awarded damages to practicing entities from 2007 to 2016 they awarded reasonable royalties. (Courts awarded lost profits in 40 percent of these cases; the percentages exceed 100 percent because courts sometimes award lost profits on a portion of infringing sales and reasonable royalties on the remainder.) Further, although nonpracticing entities (NPEs) had a lower win rate than practicing entities during the time period studied, the median award to NPEs that prevailed at trial from 2012 to 2016 was almost four times the median award to practicing entities ($15.7 million versus $4.1 million). Awards to NPEs almost always consist of reasonable royalties, rather than lost profits.
For other countries, less data is available, and the data that is available is generally less precise. Studies of Japanese damages awards indicate that reasonable royalty awards make up a plurality of all such awards but that the amounts awarded tend to be low by U.S. standards. For example, according to a 2014 study of all sixty-eight cases from January 1, 1999, to March 5, 2013, in which Japanese courts awarded reasonable royalties, in only five cases did the award exceed ¥200,000,000 (equal to about U.S. $1.7 million). The royalty rate was 5 percent in 28 percent of cases, 3 percent in 22 percent, and 10 percent in 16 percent (based on the value of the infringer’s sales revenue from the infringing product). Like the PwC studies of U.S. damages, the reports of which we are aware on average or median damages awards in France do not distinguish between royalty and lost profits awards. In China, statutory damages predominate and awards of reasonable royalties are comparatively rare.
1.1.2 Theoretical Justifications
As noted in the Introduction, for purposes of this project we take the substantive law of patents as a given, and do not advise courts to use the law of remedies to correct for perceived flaws in the substantive law. It follows from this premise that, in general, the law of patent damages should work to preserve the patent incentive, such as it is, by restoring the patent owner to the position it would have occupied, but for the infringement. Consistent with this rationale, courts and other observers often view reasonable royalty awards as a substitute for the royalty the patent owner would have earned, and that the infringer would have paid, absent the infringement. Commentators nevertheless sometimes express concern that such a standard threatens to encourage infringement (and to discourage ex ante negotiation), since it leaves the infringer no worse off for having infringed. This concern is particularly applicable if the royalty award is exactly the same as the royalty the patent owner would have negotiated, if the infringement was intentional, and if the infringer rationally could expect to avoid detection some nonzero percent of the time. In addition, the infringer may avoid some of the risks that a real-world licensee would incur – though of course, if the infringer is sued, it may wind up incurring substantial attorney fees, which it otherwise could have avoided, to defend itself. To the extent that restorative damages risk underdeterring infringement, the law in the United States already ensures that royalty awards will not be exactly the same as what would have been negotiated, because the royalty awards must be calculated based on an assumption that the patent in question is valid and infringed, whereas in actual negotiations the parties commonly reduce the royalty based on the possibility of invalidity and noninfringement. In addition, to address the risk of underdeterrence due to nondetection, policymakers could authorize courts to (1) grant injunctions, (2) award the disgorgement of the infringer’s profits, (3) shift fees to the prevailing party, (4) impose criminal sanctions, or (5) award enhanced damages. Other chapters of this book discuss these alternatives in depth. However, as specifically discussed in Chapter 3, many countries decline to award enhanced or punitive damages on public policy grounds – though in some of these countries, courts occasionally award reasonable royalties above the “normal” rate to reflect the infringer’s avoidance of risks that a good-faith licensee would have incurred.
Alternatively, one could view reasonable royalties as a form of restitution, in the sense that the award forces the infringer to pay back the royalty it wrongfully withheld from the patentee. Whether the characterization of royalties as restorative or restitutionary makes any practical difference may depend on whether the focus is on awarding the royalty the parties would have negotiated absent the infringement, or the royalty the infringer should be required to pay in light of some normative criterion. The “hypothetical bargain” or “willing licensor-willing licensee” approach, as it is often applied in the United States, might seem to be an example of the former approach, insofar as it attempts to construct the terms of the bargain the parties themselves would have negotiated prior to the date of infringement. But even that approach does not construct the exact bargain the parties would have made, because the hypothetical negotiation assumes the patent in question was valid and infringed, as discussed above. Without these assumptions, there would appear to be little difference between characterizing reasonable royalties as restorative or restitutionary. The royalty the court believes the patent owner would have earned absent the infringement is identical to the infringer’s gain (i.e., the royalty it withheld, if not adjusted upward to reflect certainty as to validity and infringement). On the other hand, an approach that attempts to determine the royalty the infringer should be required to pay does not necessarily entail restoring the parties to the positions they actually would have occupied but for the infringement – though any such approach needs to specify just what the appropriate normative criterion is. Some recent scholarship recommends focusing more on the benefit the infringer actually derived from the use of the invention (as opposed to its expected benefits ex ante), so that the resulting award will more closely correlate with the invention’s contribution to the art. As discussed in Sections 1.2 and 1.3, one would then have to determine how to divide that benefit between the parties. In theory, the division could be based on what the parties likely would have negotiated ex ante, or on industry custom or other criteria. Other recent scholarship also suggests that a restitutionary approach to patent damages would provide courts with more flexibility to adjust the requisite level of proof based on factors such as the stakes involved and the extent to which the infringer was at fault.
1.1.3 Principal Approaches
Courts throughout the world often consider a range of factors in calculating reasonable royalties. One approach often used in the United States, the United Kingdom, and some other countries is to construct the hypothetical bargain to which the court believes the parties would have agreed to avoid infringement. As discussed above, the hypothetical bargain approach may be viewed as either restorative or restitutionary. If the resulting royalty reflects what the parties actually would have negotiated, the patentee is rendered no worse off, and the infringer no better off, compared to the positions they would have occupied had they actually negotiated a license. As discussed in the following Sections, however, among the issues courts may need to address in constructing such a hypothetical bargain are (1) the timing of the bargain, (2) the knowledge the court should impute to the parties (including knowledge of validity and infringement of the relevant patent, as discussed above), and (3) the relevant factors that are probative of the terms of the bargain. Alternatively, as suggested above, an approach that focuses on dividing the actual gain to the infringer could still be cast as a hypothetical bargain, albeit one in which the parties agree ex ante on how to divide the benefit the infringer actually derives ex post. This approach would be less concerned than the more common Georgia-Pacific approach with trying to accurately construct the terms the parties themselves actually would have negotiated ex ante.
Another option under U.S. law is the so-called analytical approach, which “focuses on the infringer’s projections of profit for the infringing product.” The leading case is TWM Mfg. Co. v. Dura Corp., in which the Federal Circuit approved a damages award that involved subtracting “the infringer’s usual or acceptable net profit from its anticipated net profit realized from sales of infringing devices.” Although courts sometimes permit the patentee who employs the analytical approach to use the infringer’s actual profits as a proxy for expected profits, the approach does not appear to be used very frequently. Critiques of the analytical approach argue, among other things, that the method is indistinguishable from disgorgement; that the concept of a “usual or acceptable net profit” is not very precise; that the approach does not account for various other factors that can explain a divergence from the normal rate of return, including the presence of other product features, or for the fact that different products can have different profit margins; and that the approach can unfairly penalize an infringer who has a higher profit rate due to efficiencies in production.
Where an established royalty rate exists, courts sometimes have used that rate rather than endeavoring to construct a hypothetical bargain or an appropriate division of the profits projected or earned from the use of the invention. Where no such established rate exists, courts nevertheless frequently turn to comparable license rates as an aid in constructing the hypothetical bargain. In some countries, courts also make extensive use of what are believed to be industry standard rates for various technologies. For example, in Japan courts often start with the standard royalty rate for a given technological field, as reported in publications of the Japanese Institute of Inventors and Innovation (Hatsumei Kyokai), and then adjust the rate up or down based on factors such as “the technical or economical value and importance of the invention,” the plaintiff’s own high profit margin, the contribution of the invention to the infringer’s profitability or to the value of the end product, the existence of alternatives, and the infringer’s sales volume.
A fourth possibility would be to employ some sort of “top-down” approach as in In re Innovatio IP Ventures, LLC Patent Litigation, whereby the court identifies an appropriate royalty base, decides how much of the revenue attributable to the base should be payable as aggregate royalties, and then determines what portion of those aggregate royalties should accrue to the patents in suit, based on their relative importance. Some form of “top-down” approach may be used in cases involving complex products, but the accuracy of the approach in estimating the value of the patents in suit depends upon obtaining a considerable amount of arguably difficult-to-obtain information. This approach is discussed further in Chapter 5 on the effect of FRAND commitments on patent remedies.
1.2 Reformulating Georgia-Pacific
Judicial systems throughout the world often permit the finder of fact to consider a range of factors of arguable relevance to the calculation of reasonable royalties. In the United States, for example, damages expert witnesses frequently base their opinions on the fifteen factors first compiled in Georgia-Pacific Co. v. U.S. Plywood Co. (set forth below). Courts in other countries, including Canada, Germany, and Japan, sometimes look to a similar range of factors.
Critics nevertheless have noted several potential problems with the Georgia-Pacific framework. First, depending on the facts of the case, some of the Georgia-Pacific factors may simply be irrelevant, thus potentially distracting the trier of fact from focusing on the economically relevant considerations. Second, the framework offers little or no guidance to either the trier of fact or the judge on how to weigh or prioritize the factors. Third, and following from the first two points, it is sometimes said that a clever expert can manipulate the factors in support of virtually any award. As a consequence, it can be very difficult for the parties to predict how the trier of fact will apply the factors, and for a reviewing court to detect errors in their application. In combination, these problems threaten not only to reduce accuracy and increase costs, but also to make settlement more difficult and to place the more risk-averse party at a disadvantage.
In response to these problems, some recent scholarship and other initiatives advocate restructuring the analysis to focus on a smaller number of economically relevant factors. Most prominent, perhaps, are the Federal Circuit Bar Association’s Model Patent Jury Instructions, which propose that U.S. courts instruct juries to “consider all the facts known and available to the parties at the time the infringement began,” but that “[s]ome of the kinds of factors that you may consider in making your determination are: (1) The value that the claimed invention contributes to the accused product. (2) The value that factors other than the claimed invention contribute to [the accused product]. (3) Comparable license agreements, such as those covering the use of the claimed invention or similar technology.” In a similar vein, Durie and Lemley argue that the Georgia-Pacific factors largely “boil down to three fundamental questions: (1) what is the marginal contribution of the patented invention over the prior art?; (2) how many other inputs were necessary to achieve that contribution, and what is their relative value?; and (3) is there some concrete evidence suggesting that the market has chosen a number different than the calculus that results from (1) and (2)?” Jarosz and Chapman also have advocated a three-step framework, focusing on the incremental value of the invention over alternatives, comparable licenses, and design-around costs.
Following from the above, our principal recommendation is that, when applying a “bottom-up” approach to estimating reasonable royalties, courts should replace the Georgia-Pacific factors (and analogous factors used outside the United States for calculating reasonable royalties) with a smaller list of considerations. More specifically, courts should collapse the Georgia-Pacific factors into the following three steps. (We defend each of the individual parts of this recommendation in detail in Section 3.1 below.)
1. Calculate the incremental value of the invention and divide it appropriately between the parties. A license for the use of a patented technology typically requires the licensee to share with the licensor some portion of the incremental value the licensee derives or expects to derive from the use of that technology. To ensure that a reasonable royalty for the unauthorized use of a patented technology accurately reflects this incremental value, ideally a court would (1) estimate the difference between the value the infringer derived from the use of the patented invention (as distinct from the value contributed by other features of the infringing end product), and the value the infringer would have derived by using the next best available noninfringing alternative instead; (2) divide that differential value between the patent owner and the infringer; and (3) as an aid in carrying out this division, consider any relevant evidence, including possibly the use of a rebuttable presumption that the parties would have agreed, ex ante, to an even (50/50) split.
2. Assess market evidence. In negotiating licenses for the use of patented technologies, parties often consider the rates and other terms disclosed in relevant comparable licenses (or, where applicable, the rates charged by relevant patent pools or disclosed in publications of industry standard rates). Courts also should consider such evidence for purposes of calculating reasonable royalties for the unauthorized use of patented technologies, albeit subject to appropriate adjustments and with due appreciation for the potential limitations of such evidence as discussed in Section 1.3.6.
3. Comparison. When it is feasible and cost justified, courts should carry out both steps described above – each one acting as a “check” on the accuracy of the other – and then attempt to reconcile or adjust the results, as the evidence warrants. That said, one can expect only that courts do the best they can with the evidence available to them. Thus, when the evidence necessary to carry out Step 2 is available but the evidence necessary to carry out Step 1 is not – as will likely often be the case in litigation involving complex products – courts may need to rely exclusively on market evidence. (The converse will be true when the available evidence relates only to Step 1, not 2.) Furthermore, as discussed in greater detail in Chapter 5 on the effect of FRAND commitments on patent royalties, in appropriate cases courts also may consider applying a “top-down” approach either as direct evidence or as a check on the value derived from the use of comparables and other market evidence.
Explanation. As discussed in Section 1.3.1 below, economists generally accept “incremental value” – that is, the difference between the value derived from the patented invention over the next best available noninfringing alternative – as an accurate measure of the value of patented technology. By necessity, such an inquiry also requires the trier of fact to apportion the value attributable to the patented invention as opposed to other features of the infringer’s product, assuming that the noninfringing alternative end product sold by the infringer would have retained those other features. The first part of Step 1 above therefore combines Georgia-Pacific factors 8, 9, 10, and 13 into one overarching concept. Step 1 presumably will be easier to accomplish, however, when the infringing product embodies relatively few patented features. We defend our recommendation regarding the division of incremental value in Section 1.3.3 below.
Step 2 recommends that courts also make appropriate use of comparables and other market evidence of how actors in the real world value the technology in suit. To be sure, courts and commentators have identified numerous potential pitfalls in the use of comparables, which we discuss in greater detail in Section 1.3.6 below. These theoretical problems notwithstanding, however, we do not advocate forgoing the use of comparables (nor do we see that as a likely development, in any event), but rather emphasize the need for careful judgment in applying them. Moreover, at least in some cases a patent pool rate or other comparable may have a very high probative value, though that rate may need to be adjusted (for example, to account for the reasons why the patentee did not join the pool).
Step 3 recommends that, where feasible, courts apply both Step 1 and Step 2, and then compare the results. To the extent the numbers generated by each step diverge, the court will then have to decide how best to reconcile them based on all of the relevant facts and circumstances. For example, a court may be more confident in the result generated by Step 1 when the end product embodies only a small number of patents or when there are few if any licenses that are closely comparable. By contrast, Step 2 may seem more probative when the product’s complexity makes it difficult to distinguish the value contributed by a single patent over the next best alternative. (On the other hand, even in complex products cases it sometimes may be possible to estimate the value of a specific patented feature relative to other features, through the use of conjoint or discrete choice analysis, testimony from technical experts, or application of some form of “top-down” approach as discussed in the FRAND chapter.) Further, in cases (1) involving relatively small stakes, or (2) arising in countries that impose substantial limits on pretrial discovery or the use of expert witnesses, or (3) in which the parties’ evidence on damages is inadmissible or incompetent, the best practice may be to consider comparables, industry standard rates, or other such market evidence, despite its potential drawbacks, rather than to award zero damages or rely on other, even more speculative, evidence of the value of the technology over alternatives.
1.3 Incremental Value and Other Issues
In this Section, we present the analysis underlying our principal recommendation as described in Section 1.2. We also present our recommendations relating to various issues that may arise either in the application of our principal recommendation or in the event courts continue to employ a multifactor, Georgia-Pacific-like approach to reasonable royalties.
1.3.1 Incremental Value
We perceive a widespread consensus among innovation economists and lawyers that the social value of a technology is its incremental value over the next best alternative, and that the economic value of a patented technology to an implementer is the (actual or expected) profit or cost saving the implementer derives from the use of the patented technology over the next best available noninfringing alternative. We therefore recommend that policymakers adopt, subject to the systemic considerations noted in the Introduction, the guiding principle that the royalties awarded in litigation should be commensurate with the value of the patented technology as so defined. We also recognize, however, that there are substantial difficulties, both practical and conceptual, in assessing that value – particularly in the case of complex products, where the patented technology contributes only a small part to the overall value of the product. In those contexts, a patented feature might be the deciding factor for a few purchasers, and it might increase the value to others, but for most purchasers it is likely to be one of a host of factors that shift buying preferences as a whole. We discuss the conceptual difficulties below.
The first conceptual difficulty involves complementarity between the infringing technology and other patented technologies that are also implemented in the same product. The problem is illustrated most clearly in a case in which two versions of a complex product are sold, with no difference between the two except that one version embodies the patented technology while the other does not. It may seem that this is a case in which it is easy to determine the incremental value of the patented technology; it would seem to be simply the difference between the two prices. However, this is not correct if, as is commonly the case, the patented technology depends on other patented technology. For example, suppose the patented invention provides for 20 percent longer battery life in a smartphone, and a smartphone with the longer battery life sells for $50 more than the phone would with the shorter battery life it would otherwise have. The incremental value of the patented invention would appear to be $50. But the price consumers are willing to pay for the phone depends on its patented wireless technology, and without that wireless technology the phone would be worthless, no matter how long the battery life. In that case, the $50 price difference is only partially attributable to battery technology, because it is also partially attributable to the wireless technology. Put another way, the patentee holding the wireless technology might reasonably demand a higher royalty for the phone with the battery-extending technology than for the base phone, leaving only some part of the $50 to be split between the battery patentee and the phone vendor. Whether the wireless patentee actually demands a higher royalty in such a case is a different question – though it is not unlikely that it would do so. It is common for patentees, particularly those with basic technology patents, to charge an ad valorem royalty on the product price, with the result that the wireless royalty would be higher for the more expensive phone.
3 Patented Alternatives
A second conceptual difficulty arises from the proposition that the value of the invention is its value over the best noninfringing alternative. This proposition is uncontroversial so long as the alternative is unpatented, but its application is not so clear if the alternative is patented. It is not at all uncommon that the best substitutes for a patented technology are also patented, as several inventors devise different solutions to the same problem. The problem is illustrated most clearly when the inventions are near perfect substitutes, and it is particularly salient in the context of standard-essential patents (SEPs), where it is often the case that multiple alternative patented technologies competed for inclusion in the standard.
One possibility is that in such a case the value of the patented invention is zero, on the view that the infringing user in the hypothetical negotiation should be imagined to play one patentee off against another until the patentee is haggled down to its minimum willingness to accept. More generally, on this view the value of the invention is its incremental value over the patented alternative, ignoring the royalties that would have to be paid to use that alternative, on the rationale that those royalties do not reflect the value of the alternative technology but merely the value of the patent right. By the same token, if the infringed technology were not quite as good as the patented alternative, the value of the infringed technology would be zero. We recommend rejecting this approach, on the ground that although it makes sense from a static welfare perspective, it provides a facially inadequate incentive to invent (zero compensation) and therefore appears inconsistent with the conventionally understood purpose of the patent system.
Another possibility would be to assume that a patented alternative that is on the market is available for its established market price, which is normally above marginal cost. Put another way, “[t]he proper comparison is between the cost and value of the patentee’s component and the cost and value of the alternative, including patent royalties that would have to be paid on the alternative where appropriate.” This approach has some support in the case law, though it cannot be considered established law. Nevertheless, although this approach might seem appealing when both technologies are mature and both have an established price, it might be difficult to apply if both technologies are new to the market and neither has an established price. This suggestion is therefore likely to be unhelpful in the SEP context, where alternative technologies competed for inclusion in the standard ex ante, and the alternative that was not selected may not have a market presence at all ex post, or will have a value that is much lower than if it had been selected for inclusion in the standard. Another problem arises when the alternative technology is mature and has an established price, and the infringed technology is new. If the technologies are close substitutes, we would expect the new technology to drive down the price of the established technology, even in the absence of infringement. Thus, if the established price of the alternative is used for comparison purposes, the patentee will be overcompensated in comparison with the royalties it would have received but for the infringement. Perhaps, then, the proper approach in principle would be to assess how the price of the patented alternative would have evolved in response to the introduction of the infringed technology, in the absence of infringement. On the other hand, simply using the established price has clear advantages in terms of ease of proof.
We are not aware of any literature providing a thorough theoretical analysis of this problem. We therefore propose further research on this issue.
1.3.2 Hypothetical Bargain
In the United States, the most common approach to assessing a reasonable royalty is usually referred to as the “hypothetical negotiation” approach:
The hypothetical negotiation tries, as best as possible, to recreate the ex ante licensing negotiation scenario and to describe the resulting agreement. In other words, if infringement had not occurred, willing parties would have executed a license agreement specifying a certain royalty payment scheme.
While this approach is now deeply entrenched, the leading cases emphasize that the goal of the hypothetical negotiation framework is not to replicate the bargain that actual willing parties would have arrived at; that would be “inaccurate, and even absurd,” given that “[t]here is, of course, no actual willingness on either side, and no license to do anything, the infringer being normally enjoined … from further manufacture, use, or sale of the patented product.” The hypothetical negotiation is a “legal fiction,” “employed by the court as a means of arriving at reasonable compensation,” and it is to be “flexibly applied as a ‘device in the aid of justice.’”
We recommend that courts embrace this view of the hypothetical bargain framework as a tool – a proxy for the issues of how to split the surplus from the invention – rather than as a goal in and of itself. For example, it is well established in U.S. law that the parties to the hypothetical negotiation are assumed to have known that the patent was valid and infringed, even though actual parties would not. This rule is required to achieve just compensation, because the opposite view – that the parties should be assumed to discount the royalty to allow for the probabilistic nature of the patent (as would presumably be done by parties to an actual negotiation) – would result in so-called double discounting; not only would the court-approved royalty derived from the hypothetical negotiation include a discount for the risk of non-liability, but then pre-litigation negotiations in which royalties were based on the expectation of such a court award occurring with a less than 100 percent probability would include a further discount for risk of non-liability. For that reason, we agree that this well-established principle of U.S. law is sound. Moreover, based on similar reasoning, the hypothetical negotiation should include an assumption of liability, not just validity and infringement, as well as entitlement to relief and enforceability. And more generally, departures from a strict attempt to reconstruct what real parties would have done had they actually bargained are justified whenever such a departure would be a better means of arriving at reasonable compensation – in particular, compensation that reflects the value of the patented technology over its best noninfringing alternative. Indeed, if sound principles of reasonable compensation require an unwieldy number of departures from a hypothetical negotiation framework, the proper course would be to abandon the framework rather than the sound principles.
1.3.3 Dividing Incremental Value
The second and third parts of Step 1 involve identifying the appropriate division of the incremental value. Both sides often can make a substantial claim to at least a portion of the incremental value – the patentee because this value results from use of the claimed invention, and the adjudged infringer because it made complementary or supplementary investments that resulted in a commercial embodiment of that invention. How then should the value be divided?
In theory, an invention can give rise to pure economic rents, reflecting the value of the invention over the best noninfringing alternative. If two parties, such as a patent owner and a licensee, must cooperate to realize those rents, there is no simple theoretical answer as to how the parties will split the rents between them, since even a very lopsided split, in either direction, would leave both parties better off as compared with using the noninfringing alternative. The most prominent solution to the problem is the Nash Bargaining Solution (NBS), which implies a 50/50 split. However, the NBS requires unrealistically restrictive assumptions about the parties, such as that they are identical in every way. Economic theory is relatively underdeveloped in terms of fleshing out how pure rents would be split when the parties are modeled more realistically. The Nash Bargaining Solution is sometimes used, not because it is a particularly accurate model, but for lack of anything better.
Moreover, the division of the incremental profit due to the invention is unlikely to be a split of pure rents. Turning a patented invention into a commercialized innovation that actually commands a premium in the marketplace requires some or all of manufacturing, distribution, marketing, process refinement, technical support to the licensee by the patentee, end-user support, and so on, all of which involve risk and investment by one party or the other. The royalty paid by the licensee to the patentee does not reflect a split of pure rents, but also, or even instead, compensation to the party who made the investments and shouldered the risks relating to these ancillary services. Georgia-Pacific factor 13 recognizes this possibility, as have cases such as Tights, Inc. v. Kayser-Roth Corp.:
The Court finds, in the context of this case, that the patentee would have been reasonably entitled to receive from 25% to 50% of the cost saving as reasonable royalties. This Court finds that 25% of the cost saving is a reasonable entitlement where the parties anticipate that the licensee will have to make substantial contributions to practical commercialization. This Court finds that 50% of the cost saving is a reasonable entitlement where the parties anticipate that the licensee will have to make only routine creative contributions toward commercialization.
We therefore recommend that, to the extent possible, the split of the incremental profit should reflect the value of any such ancillary services or risks that either the patent owner or the infringer, in fact, undertook. In our view this is consistent with the hypothetical bargaining construct because it reflects the agreement the parties themselves would have arrived at in similar circumstances. Recall that the principal justification for the hypothetical bargain is that it preserves the patent incentive by restoring the patent owner to the position it would have occupied absent the infringement. That position would depend in part on how the parties would have agreed, ex ante, to divide the value to be derived from the use of the patented invention, in comparison with alternatives. However, we emphasize that we recommend taking such services into account to the extent they are actually incurred. Even if an actual licensee would have provided marketing for the invention, and an actual royalty would have reflected that value, the reasonable royalty award should only reflect that if in fact the infringer undertook the marketing.
More broadly, we propose further research to unpack and refine the nature of “bargaining power” as it relates to the division of the incremental value of the invention. We suspect that the division in any given case is determined in part by compensation for ancillary services and in part by industry norms (which may themselves reflect reflect standard practices about provision of ancillary services). To some extent this unpacking is a matter of obtaining better evidence as to what factors actually drive the division of the incremental profit in practice. In addition, there are some conceptual or normative issues to be resolved. In particular, one intuitive understanding of “bargaining power” is that a party with deeper pockets has greater bargaining power, and so would be able to extract a greater share of the incremental value in an actual licensing negotiation. For example, if the patent owner was a small cash-strapped start-up, and the potential licensee was a large company, the licensee might in practice be able to extract very favorable terms. It can be argued that it would be appropriate to replicate that unequal division in a reasonable royalty assessment, on the view that the patentee should not be made better off than it would have been had the parties actually licensed. On the other hand, the favorable terms might be considered to be an illegitimate holdout by the licensee, which should not be replicated in a reasonable royalty, on the view that it does not reflect the incremental value of the invention, just as the courts should not give the patentee a higher royalty if it would have been able to engage in holdup in an actual negotiation. These questions deserve further exploration.
Having decided which factors should be relevant to the division of the incremental profit, the second question is what evidence should be used to establish that division. A few possibilities come to mind. First, comparables may shed light, either explicitly or implicitly, on how the parties would have agreed to divide the surplus. As discussed above, evidence also could reflect any ancillary services or risks that either the patent owner or the infringer, in fact, incurred, so as to adjust the royalty derived from the comparable license. Second, there may be evidence of what the parties would have agreed to based on their own prior negotiations, the patentee’s course of dealing with other parties, or the custom of the industry. To illustrate, in United States Frumentum Co. v. Lauhoff, the U.S. Court of Appeals for the Sixth Circuit held that evidence was admissible as to what share of the profits or of the selling price “it may be customary in that or similar business to allow for the use of such an invention.” (Of course, questions may arise as to just how similar a “similar business” must be.) When there is no such evidence of how the parties would have agreed to split the incremental value, however, what then? On the one hand, it would seem wrong to award the patent owner nothing – and in any event U.S. law normally would preclude such a result because section 284 of the U.S. Patent Act requires courts to award “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.” Indeed, one of the reasons for the gradual adoption of the reasonable royalty remedy in the United States in the early to mid-twentieth century was precisely to avoid situations in which courts could award only nominal damages, due to difficulties in quantifying the owner’s loss or the infringer’s gain with sufficient certainty. Rather, as Judge Learned Hand expressed it back in 1933, “[t]he whole notion of a reasonable royalty is a device in aid of justice, by which that which is really incalculable shall be approximated, rather than that the patentee, who has suffered an indubitable wrong, shall be dismissed with empty hands.” By the same token, it would seem equally wrong to award the patentee 100 percent of the profit the infringer earned from the use of the claimed invention simply because the infringer couldn’t prove the appropriate division (unless the patentee was seeking, and was entitled to under the relevant substantive law, an award of the infringer’s profits).
Arguably then, the best practice would be to permit the parties to introduce whatever competent evidence they have on the division of profits, including comparables, while also permitting the fact finder to take note of, for example, findings from behavioral psychology and economics (e.g., the ultimatum game) suggesting that people in Western societies generally view a 50/50 split of benefits as fair. (Similarly, the Nash Bargaining Solution, application of which often may result in a 50/50 split, is a widely used construct in game theory – albeit with economists often employing the 50/50 split as a plausible assumption, rather than substantiating it as an empirical fact of how two actual parties would have bargained.) For example, in Summit 6, LLC v. Samsung Electronics Co., the Federal Circuit recently affirmed a damages judgment based on an expert witness’s purported isolation of the incremental profit Samsung had derived from the use of the patented invention, and his subsequent division of that profit between the parties based on analysis of Samsung’s bargaining power and application of the Nash Bargaining Solution.
We therefore recommend that, when faced with the question of how to divide the incremental value derived from the use of the invention over the next best alternative, courts permit the parties to introduce any competent evidence on this issue – including, where necessary to estimate a royalty “in aid of justice,” empirical findings that people in Western societies generally view a 50/50 split of benefits as fair, and that economists often use the Nash Bargaining Solution in modeling bargaining behavior. Further to this point, policymakers may wish to consider adopting a rebuttable presumption that the parties would have agreed to a 50/50 split – which presumption, however, should come into play only after there has been an initial determination of the incremental profit derived from the use of the invention, and should not be difficult for the parties to rebut by means of more specific evidence (comparables, industry practice, risk allocation, etc.).
By contrast, we would not recommend use of a stronger presumption (e.g., one that can be rebutted only by clear and convincing evidence) out of concern that, inter alia, the trier of fact (particularly a lay jury) might accord such a presumption too much weight. Further, a weak presumption of this sort should be sharply distinguished from the 25 percent rule of thumb previously used by U.S. courts and rejected by the Federal Circuit in Uniloc USA, Inc. v. Microsoft Corp. Unlike that rule of thumb, under which a damages expert was permitted to presume that “the licensee pay a royalty rate equivalent to 25 per cent of its expected profits for the product that incorporates the IP at issue,” the presumptive value split suggested here would apply only after the incremental profit properly attributable to use of the claimed invention has been isolated from all other portions of overall revenue and profit. In complex product cases, there are likely to be multiple innovations besides the claimed invention that have contributed to overall revenue and profit. Consequently, in such cases, 50 percent of the incremental profit attributable to the claimed invention can be expected to often be only a small percentage or even only a small fraction of a percent of the overall profit from the complex product.
1.3.4 Timing of Hypothetical Negotiation
The standard view in U.S. case law is that the hypothetical bargain occurs just prior to the date on which the infringement began. This timing has been controversial in two main respects. First, it is the basis for the mainstream view in U.S. law that the hypothetical bargain should be based only on information that is available to the parties ex ante, and that ex post information is relevant only as indirect evidence of what the parties would have expected ex ante (the “book of wisdom” approach). The standard in Germany, by contrast, states that the court should consider the bargain the parties would have reached ex ante had they foreseen all relevant ex post information; and a few commentators argue for the expanded use of ex post information in U.S. law as well, on the view that this allows for a more accurate valuation of the patented technology. Second, there is substantial scholarly commentary, particularly in the SEP context, suggesting that the timing of the hypothetical negotiation should be earlier, just prior to the time when sunk costs were incurred; this commentary reflects the view that the user would inevitably have incurred sunk costs by the time of the first infringement, so that a license negotiated at that time would allow the patentee to hold up the user for part of those sunk costs, leading to a royalty in excess of the value of the invention.
With these critiques in mind, we recommend that, to the extent courts continue to employ a hypothetical bargaining construct at all, they should apply a flexible approach that takes into account the hypothetical bargain’s status as a legal fiction employed as an aid to arriving at reasonable compensation, rather than as a foundational principle in its own right to be applied strictly and literally. With regard to timing in particular, in many cases the precise date of the hypothetical negotiation does not have any impact on the reasonable royalty, and the early U.S. decisions invoking the hypothetical negotiation approach did not usually specify the time when the negotiation took place. On the other hand, in the cases establishing the time of the first infringement as the appropriate date, courts have chosen the infringement date not because it reflects the time of a negotiation between truly willing parties, but because that timing does justice on the facts of the particular case. As discussed in Sections 1.2 and 1.3.5, our preferred approach is not to employ a hypothetical negotiation as such, but rather to identify the surplus that the parties are negotiating over, and to divide that surplus in an appropriate manner. But whichever construct is used, the ultimate goal is to ensure that the division does not reflect lock-in but that it does reflect any ancillary services or risks that either party has shouldered.
To illustrate our recommendation, consider a case in which the court deems the hypothetical negotiation to have taken place at the time of first infringement, but after the infringer has incurred sunk costs. In such a case, the patentee might be able to extract some of the value associated with those sunk costs even if it has no substantial relation to the value of the patented technology, contrary to the consensus view that the patentee normally should not be able to extract such unrelated value. Fortunately, we are not aware of any cases in which the courts have approved of allowing the patentee to extract value associated with such sunk costs specifically on the basis that this would have happened had the parties bargained on the infringement date. In our view, the key point is to ensure that the evidence used to establish the reasonable royalty avoids problems associated with sunk costs. If, for example, the bargain is constructed using comparable licenses, a strict adherence to the principle that the bargain takes place prior to infringement would bar the use of any comparable licenses entered into after that date. But if the party to the relevant comparable had negotiated its license prior to incurring any sunk costs, then neither that license nor a royalty based on it would reflect sunk-costs holdup, and there would be no reason to reject the use of that comparable based on its date of execution.
Further, while the view that a reasonable royalty should not reflect the infringer’s sunk costs is generally sound, it doesn’t necessarily require that the royalty be based on evidence that predates those sunk costs. The previous example highlights one such scenario. As another example, often it may be easier to determine the date on which infringement began than the date on which the infringer began incurring sunk costs, in which case – as long as the sunk costs are not too large – the marginal increase in accuracy resulting from moving up the date of the hypothetical negotiation may not be justifiable in view of the additional administrative expense. Alternatively, consider the facts of Tights, Inc. v. Kayser-Roth Corp., in which the court noted that a licensee would pay a lower royalty if it would be required to make substantial contributions to practical commercialization, and a higher royalty if it made less contributions toward commercialization. The timing of the hypothetical negotiation was important because the product market was relatively mature by the time of the first infringement, and so the reasonable royalty was higher than it would have been had the infringer entered a nascent market. If the bargain date were moved back to avoid sunk-costs holdup, this would imply that the reasonable royalty in Tights would have to be reduced correspondingly. In our view, Tights was correctly decided on its facts, and a lower royalty to notionally avoid sunk-costs holdup – which was not in issue – would be inappropriate.
This illustrates the importance of addressing the underlying issue rather than focusing solely on the date of the hypothetical negotiation. A negotiation date that is appropriate for some purposes (avoiding sunk-costs holdup) may be inappropriate for others (ensuring that the royalty reflects the infringer’s contribution to commercialization). Moving the negotiation date back to solve one problem might simply create other problems, when all that is really necessary is to ensure that the specific evidence on which the royalty is based does not inappropriately incorporate sunk-costs holdup.
1.3.5 Information Set
We now turn our attention to the issue of changed information. Suppose that at the time of the first infringement, the parties anticipated that the invention would be a great success, and so they would have contracted for a very high royalty, but in fact the invention was a failure. If the royalty is to be based only on the information that was available to the parties at the time of the first infringement, the damages award would be very high; but if it is based on the knowledge that the invention is in fact worthless, then the royalty would be very low. For example, following a jury trial in 2012, a federal district court entered judgment in the amount of $1 billion in favor of Monsanto in a patent infringement dispute against DuPont. This amount reflected the jury’s best estimate of the lump-sum amount that DuPont would have agreed to pay and that Monsanto would have accepted, just before the infringement began, even though DuPont never sold any of the infringing seed at all. Notably, the opposite story may also be told. A technology expected to be worthless may prove to be valuable. The mainstream view in U.S. law nevertheless is that ex post information can be used only to establish what the parties believed at the time of first infringement, and if it can be established that their views turned out to be wrong, then the reasonable royalty will be calculated on the basis of those wrong views, and not on the basis of what actually transpired.
We recommend, however, that contrary to the mainstream U.S. approach, courts should adopt what Siebrasse and Cotter refer to as the “contingent ex ante approach” under which the hypothetical negotiation is generally assumed (subject to the caveats noted in the preceding section) to take place before any sunk costs are incurred, but with the benefit of ex post information. The rationale for this approach is that the bargain must be assumed to take place ex ante, so that the patentee is not entitled to extract any holdup value; but at the same time, using ex post information more accurately reflects the true incremental value of the invention, and so provides a more accurate reward to the patentee. This is not really inconsistent with a hypothetical negotiation framework, because parties often negotiate on a contingent basis. For example, it is routine to negotiate a running royalty, the effect of which is to make the return to the patentee contingent on ex post information. Using ex post information in the hypothetical negotiation posits that the parties would contract on a broadly contingent basis, taking into account all relevant factors, not just the volume sold. This approach would not exclude evidence that the parties actually would have agreed upon a lump sum royalty, but merely presumes that the parties would have preferred a royalty that took into account the risk of lack of success of the patented technology. This approach is also consistent with the established rule that the parties to the hypothetical negotiation are assumed to know that the patent is valid and infringed even though during actual negotiations they would have discounted the royalty for risk of non-liability.
This view also has some support in U.S. case law, most prominently in the statement by Justice Cardozo in Sinclair Refining that:
An imaginary bid by an imaginary buyer, acting upon the information available at the moment of the breach, is not the limit of recovery where the subject of the bargain is an undeveloped patent. Information at such a time might be so scanty and imperfect that the offer would be nominal. The promisee of the patent has less than fair compensation if the criterion of value is the price that he would have received if he had disposed of it at once, irrespective of the value that would have been uncovered if he had kept it as his own.
This is often said to reflect only the principle that ex post information may be used as evidence of what the parties would have believed at the time of the first infringement, but on its face it supports the use of ex post information more generally. Similarly, in Georgia-Pacific, the district court actually did consider post-infringement evidence, and on appeal the Second Circuit held that the district court had not erred in so doing. More recently, it appears that the courts have begun to be more liberal in the use of ex post evidence.
On the other hand, one objection to the use of ex post information is that courts have tended to invoke the “book of wisdom” asymmetrically to benefit patentees but not infringers. One obvious response to this objection is that it is wrong to do so. Presumably clarifying that the use of ex post evidence is generally permissible would help avoid an unprincipled asymmetric approach. Lee and Melamed further argue that using ex post information substantively, rather than merely as evidence of what the parties would have known or believed at the time of the first infringement, leads to two mistakes:
First, the rationale assumes that the actual profits would have been unforeseen entirely at the time of the hypothetical negotiation, when the parties negotiating ex ante would likely have understood that there would be a range of possible outcomes (some leading to higher profit and some leading to little or no profit for the infringer) and would have taken all of them into account in selecting a reasonable royalty ex ante. Second, … a royalty determined on the basis of ex post evidence will generally include a premium based on ex post economic developments that increase the infringer’s reliance on the patent – in particular, lock-in costs – and that are unrelated to the incremental benefit the patent confers.
The first objection, however, misses the point. When the parties’ expectations are accurate ex ante, there is no difference between an approach that uses ex post information and one that does not. The rationale for the use of ex post information is that it allows more accurate determination of the royalty when the parties are mistaken. The second objection is sound so far as it goes, though it actually applies equally to the standard position that the negotiations are assumed to take place at the time of first infringement, by which time the infringer will normally have already incurred lock-in costs. The response is the same whether or not ex post information is to be taken into account; it is to refuse to award royalties that reflect lock-in costs. Put another way, Lee and Melamed implicitly assume that in order to take into account ex post information, it is necessary to assume that the hypothetical negotiation takes place ex post; but under the Siebrasse and Cotter proposal, the hypothetical negotiation is assumed to take place before sunk costs have been incurred, but in light of all ex post information, not just information regarding validity and infringement.
In short, rather than excluding ex post information entirely, the better response is to clearly articulate the rationale, which is not simply to increase the patentee’s reward, and thereby make it clear that ex post information is admissible no matter what effect it has on the reasonable royalty damages. Consequently, we are of the view that the contingent ex ante approach is sound.
1.3.6 Comparable Licenses
If we imagine a reasonable royalty as the product of a hypothetical negotiation between the parties using certain assumptions, the use of comparable licenses – what similarly situated parties “did in fact agree to” – as an aid in making this determination seems quite sensible. Indeed, when a license meets the stringent requirements to qualify as an “established” one, its probative value might seem clear. There are nonetheless significant practical and conceptual problems involved with using comparable licenses – even “established” ones – as evidence of a reasonable royalty. Although we do not suggest that courts should forgo the use of comparable licenses, we recommend that courts should be aware of the problems discussed below, and to the best of their ability take these considerations into account when using comparables.
The most obvious hurdle in using comparable licenses is to ensure comparability. It is rare to find actual licenses entered into in exactly the circumstances of the hypothetical negotiation. In theory, a license may be sufficiently comparable to be considered as evidence of a reasonable royalty even though it was not negotiated in circumstances exactly corresponding to the hypothetical negotiation, though adjustments then may have to be made to allow for the differences. And if the license is too dissimilar, it may be properly excluded – particularly in U.S. practice, in which judges play an important gatekeeper role by excluding evidence from consideration by juries.
While licenses involving different patents for related technologies may in principle be useful comparators, there are evident problems in determining whether a different technology is sufficiently comparable. Consequently, courts prefer to rely on licenses granted by the patent owner for the same patent, but even then problems arise. Licenses often bundle many patents together, including the patent of interest, which makes it difficult to separate out the value of the technology protected by the patent in suit. Licenses involving technology transfer, as opposed to a mere promise not to sue, routinely include other forms of supporting IP such as trademarks or trade secrets relevant to the patented technology, as well as other obligations on both sides such as grantback clauses or obligations to provide ongoing technical support. In litigation, the hypothetical negotiation concerns a very different transaction, often one involving a bare license to the patent itself. Nevertheless, it may be possible to make adjustments to compensate for the value attributable to other factors. At least U.S. courts appear generally well attuned to this problem, and commonly exclude licenses including substantial non-patent benefits.
Moreover, even licenses to the same patent with similar ancillary clauses are not necessarily comparable in terms of the royalty, because patentees are likely to price discriminate – that is, to charge different users prices that reflect the variation in value among those users. A few square centimeters of Gore-Tex may save a life when used in a vascular graft, while a square meter of it may be needed for added comfort in a rain jacket. If the patent owner charged the same amount per unit area to the raincoat manufacturer as to the stent manufacturer, it would either forego substantial profits on the license for the stent, or forgo the raincoat license entirely. Price discrimination is consistent with the principle, enunciated at the outset of this chapter, that the patentee should be entitled to a reward commensurate with the value of its technology over the next best alternative. If that value varies between applications, the patentee is likely to charge a different price for those applications. This means that the royalty in a license for the use of the patented technology in a raincoat is probably not a valid comparable in litigation of the use of the technology in a stent, even if the ancillary clauses (and even the licensee) are exactly the same. (Indeed, even licensees that manufacture both stents and raincoats may well pay a different royalty to the patentee for the different uses.) Similarly, a patentee may also price discriminate between different users, even for the same application, if for example one of the users has access to complementary technology while the other does not.
Another possible problem with using comparable licenses is circularity. Because the use of comparables to determine a reasonable royalty is one of the most predictable aspects of a reasonable royalty assessment, one would expect the parties to anticipate the use of comparables if the matter were to proceed to litigation, and to factor this into their bargaining. Thus, if there is any systematic and predictable error in the courts’ assessment of the royalty, this error will then be amplified through the use of comparables. Moreover, circularity can arise even if the parties never litigate, as it depends only on the parties’ expectation of the litigation outcome.
Circularity can come in two distinct forms, which we will refer to as “holdup/holdout circularity” and “probabilistic circularity.” First, if the prior licenses being used as comparables were negotiated in circumstances where the licensee was subject to holdup or the patentee subject to holdout, the comparable will reflect holdup or holdout value, not just the value of the patented technology over the noninfringing alternative. One cure for holdup circularity would be to eliminate the risk of holdup itself by denying injunctive relief, though the question of whether denying injunctions in a broader class of cases is desirable, is a significant issue in and of itself (and one that probably should not be driven by the problem of holdup circularity). Alternatively, courts can avoid holdup circularity even if they grant injunctions by excluding evidence of licenses that were negotiated in circumstances giving rise to holdup. This implies excluding evidence of licenses that were negotiated after the licensee had incurred sunk costs. But this may not be easy, as it requires knowledge not just of the prior license itself, but the circumstances under which it was negotiated. In addition, Lemley and Shapiro argue that a form of holdup arises when the user would have had to keep its product off the market after litigation to allow for redesign, and this form of holdup also can be magnified by circularity. This “redesign holdup circularity” can be avoided by excluding licenses negotiated in those circumstances, but this rule too would seem difficult to implement, since it would require knowledge of what the licensee would have thought its best option was in the counterfactual world in which its licensing negotiations failed. The problem of redesign holdup circularity nevertheless can be mitigated if stays are normally granted to allow redesign, as discussed in Chapter 4 on injunctions.
A different kind of circularity can arise due to the probabilistic nature of patents. As discussed in Section 7.3.1, parties to an actual negotiation would discount the value of the patented technology by the probability of liability, thus potentially giving rise to the double discounting problem if courts use a negotiated royalty as the basis for a reasonable royalty. (The doubly discounted reasonable royalty awarded by the court then would serve as background to the negotiation of the next license, which would then be trebly discounted and so on.) In contrast with the problem of holdup circularity, which potentially inflates negotiated royalties as compared with the benchmark value of the patented technology, this problem of “probabilistic circularity” deflates negotiated royalties as compared with the benchmark. (Note too that it is likely to infect even established royalties, notwithstanding their more elevated status in the hierarchy of comparables as noted above.) Furthermore, unlike holdup circularity, which does not arise if the parties do not anticipate that a permanent injunction will be granted, probabilistic circularity arises whether or not the parties expect a permanent injunction to be granted.
Conceivably, holdup circularity and probabilistic circularity may offset one another in some cases, but given the difficulty in assessing the magnitude of both types of circularity, it will be impossible to determine the degree to which this is so. The most that might be said is that when the prior license involved a license to a patent that was not already known to be valid and infringed, and it was negotiated after the licensee had incurred sunk costs, the negotiated royalty might be too high or too low, depending on which effect dominates.
In principle, the problem of probabilistic circularity can be avoided by suitably enhancing the actual royalties to compensate for discounting. There are two problems with this response, however. The first is that, in practice, it seems that such an enhancement is rarely made. The second, and more fundamental, problem is the difficulty of making an appropriate adjustment. The ideal multiplier would turn on the belief of the parties to the comparable license as to the probability of liability at the time they negotiated the license. But this will be very hard to prove, as it turns “upon private information, available only to the parties to the first licensing agreement, about the plaintiff’s probability of success in litigation.” The information may not exist at all outside the minds of the negotiators, and because the prior licensee is not a party to the current litigation, any internal memoranda shedding light on the licensee’s view of the probability of liability probably would not be discoverable. (In some cases, the patentee’s internal memoranda might shed light on the issue, but even using this information would be problematic, as it would normally represent only the patentee’s view.) And in any event, this inquiry would require time-consuming and expensive satellite litigation. An alternative would be for the court to try to estimate the discount based on objective factors relative to the particular prior license, such as the testimony of experts as to the probability of liability. But this would be a difficult inquiry on a new issue that would not otherwise have to be litigated, and that does not seem especially susceptible to the production of reliable results. Thus, in many situations, courts might be better off without adjusting for the implicit discount, and instead simply being mindful that the comparable license provides “a floor for valuing the patent, not [necessarily] a reasonable estimate.”
Finally, one could imagine using a standard multiplier. For example, if “any given patent owner has a 25% chance ex ante of prevailing against any given alleged infringer, then the appropriate multiplier is four.” But a standard multiplier not calibrated to evidence of discounting in a particular case merely recasts the circularity problem. This is because a standard multiplier will overcompensate patentees with strong patents. Anticipating this, parties bargaining in the shadow of the expected trial outcome will negotiate a royalty based on the inflated damages value, and that inflated royalty will feed back into future awards, and so on. This would result, in effect, in a new source of holdup that would allow a patentee with a strong patent to extract more than the value of its invention. The same spiral would happen in the other direction with patents that are weaker than average.
It therefore would appear very difficult in most cases to reliably enhance the actual royalty arrived at in prior comparable licenses, even though the licenses are themselves otherwise very similar to that at issue in litigation. An alternative approach would be to try to select licenses in which the royalty was not discounted, because they were negotiated in circumstances in which the probability of liability is high. One example would be licenses negotiated after a patent had been held to be valid in other litigation. But even then, the previous judgment of validity would not be binding in litigation involving a different infringer, so it is likely there would still be some discount for the probability of invalidity. And unless the implementation was exactly the same (as might be the case in the SEP context), there might be substantial discounting as to infringement as well, let alone discounting due to risks of invalidity or unenforceability. Further, this approach would severely restrict the cases in which comparables could be used.
In the same vein, some authors have suggested that prior settlements, which courts in the United States normally (though not always) exclude from evidence, actually should be preferred, particularly if the settlement was entered into when the patent owner appeared to be winning the underlying litigation. This proposal is again only helpful in a relatively narrow range of cases, as prior settlements are not always available. Moreover, it must be clear that the patentee was winning on the basis of objective factors, such as preliminary motions favoring the patentee, or the discounting problem will not be addressed. Further, if the patentee in the prior litigation would have expected to obtain an injunction if successful, the settlement may reflect holdup value – thus solving the problem of probabilistic circularity at the expense of inviting the problem of holdup circularity. Another concern with settlements is that they may reflect the value of avoiding litigation costs rather than the value of the patented technology, though this would be a significant problem only when litigation costs are at least comparable to the value of the patented technology.
3 Dynamic Considerations
Some of these problems are likely to get worse in contexts where patentees can predict that a reasonable royalty will be the primary remedy, because we would expect them to adjust their licensing practices to reflect this expectation. These adjustments might have two kinds of unwanted effects. First, they may make determining accurate damages in the particular case even more difficult. A patentee worried about probabilistic discounting depressing its recovery in future litigation may insist on artificially bundling unnecessary trade secrets or other sham terms into a license solely to ensure that it cannot subsequently be used as a comparable. This is wasteful in itself, though if the parties are careful to include terms that they know are in fact of no value, it will not otherwise distort the transaction. Another possibility is that a patentee would include self-indulging statements in license agreements about large discounts in light of significant risks of non-recovery. Conversely, the patentee may try to game the system by negotiating licenses with artificially high rates, in hopes that these will be used as comparables. This tactic is also wasteful in terms of increased transaction costs, but again it will not affect the licensing terms more generally if courts can detect and exclude such licenses from being used as comparables (which, however, is debatable).
Second, such adjustments may distort the general licensing behavior of the patentee in ways that will have more general effects. As discussed above, price discrimination means that a patentee will rationally charge a high royalty to a high-value user and a low royalty to a low-value user. But if the patentee anticipates that its license to a low-value user will be used as a comparable in subsequent litigation against a high-value user, it may prefer not to license the low-value user at all. This hurts both parties, and society as a whole. The cure for this, in principle, would be for courts to exclude licenses negotiated with a low-value user as comparables in subsequent litigation with the high-value user, but it is far from clear that courts could reliably and predictably differentiate the two cases. And of course, the first step would be for courts to acknowledge the need to do so. Otherwise, the use of comparable licenses to assess reasonable royalties may actually result in restricted licensing of the technology. This would be highly undesirable if it is now, or is likely to become, a problem in practice.
All of this is not to say that comparables are not probative at all, or that the above problems can never be mitigated or avoided. For example, Judge Robart’s use of the MPEG LA H.264 patent pool rate in Microsoft v. Motorola probably did not trigger a serious probabilistic discounting problem, because even if some individual patents in the pool might have been invalid or not infringed, parties to the pool could be highly confident that it was necessary to take a pool license to practice the technology in question. The price discrimination problem also did not appear to arise in that case, because the pool did not price discriminate other than on the basis of volume, and a pool license would have been available to the infringer. Sunk-costs circularity also probably did not arise because, at least as it appears, the pool rates were set to attract licensees who had not yet incurred sunk costs. Moreover, it may be the case that the circularity problems noted above are more theoretical than practical. Although the annual patent litigation studies produced by PwC and Lex Machina, discussed above in Section 1.1.1, reveal some variations from year to year, there does not appear to be any trend toward consistently higher (or lower) median damages awards in the United States over the past decade. Theoretical difficulties aside, therefore, it may be that courts already are adequately counteracting the potential spiraling effects of circularity.
Overall, then, we recommend that courts should apply comparables and other market evidence with caution. Such evidence often may be the best that is available, and even when there is other evidence of the value of the technology over alternatives, it may still be useful to consider market evidence by way of comparison. Nonetheless, courts probably could make more accurate determinations if more license terms were publicly accessible. We therefore recommend (and propose further research devoted to) ongoing efforts to encourage such disclosure.
1.3.7 Entire Market Value Rule and Smallest Saleable Unit
Another practical concern that often arises when applying a multifactor approach to reasonable royalties is that the parties may make strategic choices with respect to the royalty base and royalty rate that they present to fact finders. To reach a specific reasonable royalty award, a patentee could argue (or a fact finder could determine) that a relatively small rate should be applied to a relatively large base, or conversely that a relatively large rate should apply to a relatively small base. For example, a 1 percent royalty rate applied to a $10,000 base and a 25 percent rate applied to a $400 base both lead to a $100 reasonable royalty award.
In theory, it should be irrelevant which method a litigant elects when presenting a damages case in court, and a fact finder should be able to determine an appropriate royalty employing either method. In line with this observation, in many jurisdictions courts routinely use the value of the end product as the royalty base.
In recent years, however, U.S. law has placed limits on patentees’ ability to introduce evidence of the profit or revenue derived from sales of the entire accused product. These restrictions have been motivated at least in part by the long-recognized need to ensure that damages are properly apportioned to the patented features of the accused device, and not to other elements. Concerns over large bases resulting in overcompensation thus have led the Federal Circuit to articulate a general rule that the royalty base should be the “smallest saleable patent-practicing unit” (SSPPU) in the accused product, and that use of the “entire market value” of the end product as the base is permissible only when the patent drives the demand for the end product. In yet more recent cases, however, the Federal Circuit has permitted use of the entire market value when the parties themselves negotiated ex ante on the basis of the entire accused product, or comparable licenses were negotiated on the basis of entire products.
In addition to concerns over apportionment, the Federal Circuit justifies its preference that the royalty base be the smallest saleable unit on the grounds that the value of the entire accused product will tend to have an undue influence on jurors in cases where the asserted patent covers just one of many components or features that comprise the entire product, and in such cases may lead to damages awards that are overcompensatory. The concern may stem from a cognitive bias known as “anchoring,” i.e., the human tendency to give undue weight to the first data point one encounters, even if that data point is arbitrary or irrelevant. In the context of U.S. litigation, anchoring tends to reinforce the importance of the plaintiff’s damages case, which is virtually always presented first and in some cases is not countered at all by the infringer. Experimental studies using fact patterns involving personal injury cases and punitive damages awards have found evidence of an anchoring effect and suggest that, all else equal, a plaintiff that requests more damages will tend to receive a larger award. Thus, there is a risk that reasonable royalty awards based on the entire value of the accused multicomponent products will systematically overvalue patent rights that cover just a fraction of the products’ components or features.
A related problem is that, according to one study based on royalties awarded from 1982 to 2005, U.S. juries tend to award royalty rates that are within the general vicinity of 10 percent, regardless of the size of the base that the rate is applied to. Combined with anchoring, this finding (if it is still valid) suggests that a patentee who is permitted to present large revenue figures to a jury or judge might receive a larger damages award as a result, even if the revenue figures themselves bear little relation to the value of the patented technology. On the other hand, we are not aware of any more recent studies on the issue, and it is possible that the effect has diminished over time (due, perhaps, to the abolition of the 25 percent rule of thumb). There is also concern that juries prefer whole-number rates even when the evidence suggests that the appropriate rate is less than 1 percent. We therefore propose further research on the question of whether juries are susceptible to awarding inappropriately high damages given concerns with apportionment, anchoring, and preferences for particular royalty rates.
In addition, there may be a risk that use of the entire market value as the royalty base will skew litigation outcomes by encouraging patentees to sue downstream parties that are ill suited to defend patent cases. Imagine for example, an allegedly infringing component that is produced by manufacturer M, incorporated into a consumer electronics product produced by company C, shipped to retailer R, and sold to user U. Because infringement can occur by making, selling, or using patented technology, M, C, R, and U are all potential targets for suit. However, in all likelihood it is M that is best positioned to defend a patent suit. R and U, in particular, may well know nothing about how the component operates, not to mention the intricacies of patent law. Nonetheless, the effect of anchoring will tend to inflate the amount of damages a patentee can expect to recover from C, R, or U. While M may sell the chip to C for pennies or a few dollars, C may earn dozens or hundreds of dollars per unit in sales to R, and R may sell the final product to users for several hundred dollars more per unit. In addition, U may use the product as part of a business that generates many thousands of dollars a year. Given the option to choose, a patentee will find it advantageous (for reasons that have little to do with the value of the patented technology) to seek damages from component purchasers, retailers, or even users, all of whom have suboptimal incentives to test the patent’s validity and the patentee’s infringement contentions. On the other hand, if patentees are suing retailers and users who are ill-positioned to defend themselves in an effort to obtain inappropriately high royalties, the first best solution may be to rein in the ability to maintain patent infringement lawsuits down the chain of distribution, rather than to alter damages law.
Finally, limiting damages to the smallest saleable unit may have certain practical benefits. For example, defaulting to a smaller royalty base will tend to reduce the effect of error in royalty rate selection. It will also tend to narrow the range of possible trial outcomes, which benefits risk-averse parties and increases the likelihood of pretrial settlement.
At the same time, there are several economic arguments in favor of using the entire market value as the royalty base. First, limiting damages calculations to the component level may undervalue patented technology by failing to share with the patentee a portion of the spillover value created by its invention. A new high-resolution computer screen, for example, may be undervalued by U.S. patent law because, though demand for computers is not primarily driven by their screens, better screens enable or improve other computer functionality, such as video gaming and movie watching. While in many circumstances we would expect spillover value to be reflected in the sales price of the patented component, it may not be in some instances. Thus, as Petit has argued, “general purpose” technologies with many relatively low-value uses may be undervalued in patent suits against parties that use the technology for less common applications that produce especially large cost savings or profits.
Relatedly, to the extent price discrimination is economically efficient, it makes sense to allow patent owners to extract a higher royalty from implementers who market comparatively expensive end products for which the patent confers substantial value. In addition, as noted above, in real-world licensing transactions parties often, though not invariably, use the entire market value as the base. To the extent reasonable royalty awards should mimic real-world licenses, use of the entire market value often would seem unexceptional.
Given the wide variety of arguments for and against the entire market value/SSPPU rules as employed in the United States, we first propose further research, both with regard to the economic issues highlighted above and into the psychology of judges and juries (e.g., can anchoring and other biases be overcome in other ways?). Given the likelihood that anchoring does play a role in jury deliberations, however, we further recommend that, for now at least, the Federal Circuit retain rules substantially restricting the use of the entire market value. By the same token, given the likelihood that professional judges are less affected (though perhaps not unaffected) by anchoring, for now we do not recommend that other countries (which do not employ juries to decide patent cases) alter their more liberal approach to the use of the entire market value.