Jonathan S. Masur† As the number of nine- and ten-figure verdicts

THE USE AND MISUSE OF PATENT LICENSES

Jonathan S. Masur

†

ABSTRACT

As the number of nine- and ten-figure verdicts continues to increase

it is impossible not to take notice: patents are becoming an ever bigger

business with more and more wealth at stake. At the center of that business

lie the damages that courts award at trial, and the ways in which courts go

about calculating those damages. Yet the legal standards meant to govern

patent damages are notoriously ambiguous and unhelpful. In the face of

these difficulties, courts have sought a market mechanism that would aid

them in calculating patent damages. The solution they have seized upon is

to use existing licenses, typically granted by the plaintiff to third parties, as

evidence of the proper measure of damages. But the use of existing licenses

to measure reasonable royalty damages creates three significant and distinct

problems: first, it relies upon private information available only to the

parties to the pre-existing licensing agreement; second, it is ineluctably

circular; and third, it creates incentives for the patent holder to distort the

value of the licenses it negotiates in order to mislead the court. The Article

describes and analyzes these three problems, and then turns to potential

solutions. It analyzes a variety of possible reforms, including selection of

particular licenses for comparison or the application of a multiplier to the

value of existing licenses. Though several of these solutions show promise,

none come close to being a complete answer. It may well be that courts

have no choice but to largely ignore existing licenses when calculating

patent damages, leaving them more at sea than ever.

INTRODUCTION

In 2012, the three largest jury verdicts handed down anywhere in the

United States came in patent cases. Carnegie Mellon won a $1.17 billion

verdict against Marvell Technology Group for infringing integrated circuit

†

John P. Wilson Professor of Law and David and Celia Hilliard Research Scholar,

University of Chicago Law School. I thank Michael Abramowicz, Tacy Flint, William H.J.

Hubbard, William R. Hubbard, Mark Lemley, Saul Levmore, Lisa Ouellette, and Eric

Posner for helpful comments and conversations, and John Yang and Jia Zhu for excellent

research assistance. This work was supported by the David and Celia Hilliard Fund.

2 MISUSE OF PATENT LICENSES 24-Feb-15

patents;

Apple won a $1.05 billion judgment against Samsung for patents

related to its smartphones and tablets;

and Monsanto was awarded $1

billion in a lawsuit against DuPont over patents on genetically modified

soybeans.

Five other firms won damages verdicts in excess of $100

million.

The following year, Pfizer avoided shattering the record for the

largest patent verdict in history by settling claims against two generic drug

manufacturers for $2.15 billion just as jury deliberations were about to

begin.

In between these headline-making numbers, dozens of plaintiffs

won verdicts in the millions or tens of millions of dollars, and many more

patent owners negotiated royalty agreements outside of litigation for

millions more.

After enough nine- and ten-figure verdicts it is impossible

not to take notice: patents are becoming an ever larger business with more

and more wealth at stake. At the center of that business lie the damages that

Carnegie Mellon Univ. v. Marvell Tech. Grp., Ltd., 986 F. Supp. 2d 574, 582 (W.D. Pa.

2013); Verdict and Settlement Summary, Carnegie Mellon University vs. Marvell

Technology Group Ltd. & Marvell Semiconductor Inc., 2012 WL 8262739 (W.D. Pa. Dec.

26, 2012) (No. 09-CV-00290).

Verdict and Settlement Summary, Apple Inc. v. Samsung Elecs. Co., 2012 WL 4078433

(Aug. 24, 2012) (No. 11-CV-01846).

Monsanto Awarded $1 Billion in Patent Infringement Case Against Dupont, 19 No. 8

WJINTPROP 4 (Aug. 9, 2012); Verdict returned, Monsanto Co. v. E.I. du Pont de

Nemours & Co., (E.D. Mo. Aug. 1, 2012) (No. 09-CV-00686).

Verdict and Settlement Summary, VirnetX v. Cisco Sys., 2012 WL 7638169 (E.D. Tex.

Nov. 6, 2012) (No. 10-CV-00417) ($368M); Jury Verdicts, General Electric Company v.

Mitsubishi Heavy Industries, Ltd. and Mitsubishi Power Systems Americas, Inc., 2012 Jury

Verdicts LEXIS 2875 (N.D. Tex. Mar. 8, 2012) (No. 10-CV-00276) ($170M); Verdict and

Settlement Summary, Mformation Technologies Inc. v. Research In Motion Limited, 2012

WL 10864224 (N.D. Cal. Aug. 8, 2012) (No. 08-CV-04990) ($147M); Jury Verdicts,

Brocade Communications Systems, Inc. v. A10 Networks, Inc., 2012 Jury Verdicts LEXIS

18042 (N.D. Cal. Aug. 6, 2012) (No. 10-CV-03428) ($112M); Jury Verdicts, WesternGeco

LLC v. Ion Geophysical Corp., 2012 Jury Verdicts LEXIS 14042 (S.D. Tex. Aug. 16,

2012) (No. 09-CV-1827) ($106M).

Teva, Sun Pharma to Pay $2.15 Billion to Settle Pfizer Patent Suit, 20 No. 5

WJINTPROP 8 (Jun. 26, 2013)

Verdict and Settlement Summary, Shelbyzyme LLC v. Genzyme Corp., 2012 WL

5266437 (D. Del. Jul. 17, 2012) (No. 09-CV-00768) ($50M); Verdict and Settlement

Summary, Enzo Biochem v. Applera Corp., 2012 WL 6951353 (D. Conn. Nov. 1, 2012)

(No. 04-CV-00929) ($48.6M); Verdict and Settlement Summary, I/P Engine v. AOL Inc.,

2012 WL 10033892 (E.D. Va. Nov. 6, 2012) (No. 11-CV-00512) ($30.5M); Verdict and

Settlement Summary, Pacific Bioscience Labs. v. Nutra Luxe, 2012 WL 8628035 (W.D.

Wa. Sept. 12, 2012) (No. 10-CV-00230) ($11.6M); Gene Quinn, Patent Business: Deals,

Acquisitions & Licenses July 2012 (Jul. 20, 2012),

http://www.ipwatchdog.com/2012/07/20/patent-business-deals-acquisitions-licenses-july-

2012/id=26073/; Gene Quinn & Angel Krippner, Patent Deals, Licenses and Settlements –

December 2012 (Dec. 30, 2012), http://www.ipwatchdog.com/2012/12/30/patent-deals-

licenses-and-settlements-december-2012/id=30879/.

17-June-15 MISUSE OF PATENT LICENSES 3

courts award at trial and the ways in which courts go about calculating those

damages.

Part of the explanation for the heightened focus on monetary

damages in patent law is the Supreme Court’s 2006 decision in eBay v.

MercExchange.

That decision made it more difficult for prevailing patent

plaintiffs to obtain injunctions.

One of the primary benefits of granting an

injunction was that it often freed the judge from having to calculate

damages independently, leaving that determination to the parties.

With

injunctions becoming less common in cases where a patent holder prevails,

the calculation of damages for patent liability has moved closer to center

stage. The result is renewed attention to damages calculations, from both

the courts

and the scholarly literature.

Nonetheless, courts’ efforts to calculate patent damages continue to

be plagued by a number of well-understood problems. In some instances,

the plaintiff and defendant are both practicing entities that produce

competing products.

In these cases, the judge or jury has the

547 U.S. 388 (2006).

Id. Prior to eBay, there was a presumption in favor of granting an injunction to a

prevailing plaintiff in a patent case. eBay held that the usual rules of equity applied and that

courts might only grant an injunction when a four-factor test favored doing so—which was

not always the case.

Guido Calabresi & Douglas Melamed, Property Rules, Liability Rules, and Inalienability:

One View of the Cathedral, 85 H

ARV. L. REV. 1089 (1972) (noting that an injunction

clarifies the parties’ rights, allowing them to return to the bargaining table and negotiate

privately).

See, e.g., Apple v. Samsung, 2014 WL 549324 (denying Apple’s request for additur

following a second trial on damages); Apple v. Samsung, 2014 WL 6687122 (setting

ongoing royalty payments); Lucent v. Gateway, 580 F.3d 1301 (Fed. Cir. 2009); In re

Innovatio IP Ventures, LLC Patent Litigation, 2013 WL 5593609; Apple, Inc. v. Motorola,

Inc., 869 F. Supp. 2d 901 (N.D. Ill. 2012) (Posner, J.); Apple, Inc. v. Motorola, Inc., 757

F.3d 1286 (Fed. Cir. 2014).

See, e.g., John C. Jarosz, Michael J. Chapman, The Hypothetical Negotiation and

Reasonable Royalty Damages: The Tail Wagging the Dog, 16 S

TAN. TECH. L. REV. 769

(2013); Merritt J. Hasbrouck, Protecting the Gates of Reasonable Royalty: A Damages

Framework for Patent Infringement Cases, 11 J.

MARSHALL REV. INTELL. PROP. L. 192,

215 (2011); David O. Taylor, Using Reasonable Royalties to Value Patented Technology,

49 G

A. L. REV. 79 (2014); Stephen J. Conroy et. al., The Case for Admitting Settlement

License Agreements in A Reasonable Royalty Analysis, 46 L

ES NOUVELLES 291 (2011);

Jaimeson Fedell, A Step in the Right Direction: Patent Damages and the Elimination of the

Entire Market Value Rule, 98 M

INN. L. REV. 1143 (2014); Layne S. Keele, Res"Q"Ing

Patent Infringement Damages After Resqnet: The Dangers of Litigation Licenses As

Evidence of A Reasonable Royalty, 20 T

EX. INTELL. PROP. L.J. 181, 205 (2012).

See, e.g., Lam, Inc. v. Johns-Manville Corp., 718 F.2d 1056 (Fed. Cir. 1983) (“In patent

infringement action in two-supplier market, lost profits damages may be in form of

diverted sales, eroded prices or increased expenses”); Siemens Med. Solutions USA, Inc. v.

Saint-Gobain Ceramics & Plastics, Inc., 637 F.3d 1269 (Fed. Cir. 2011) (substantial

evidence existed for award of lost profits based on existence of two-supplier, high-end

4 MISUSE OF PATENT LICENSES 24-Feb-15

comparatively easier task of determining damages by estimating lost profits:

how many additional sales would the plaintiff have made had the defendant

not infringed its patent?

Of course, one uses the word “easier” advisedly;

this task is far from simple. It can be tremendously difficult to determine

how many sales a patent holder lost because of the infringement, or if it lost

any at all. The patent may cover a feature that is largely irrelevant to

consumer decisions, or it may duplicate an equally attractive non-infringing

technology that the defendant could have employed instead. Thus,

determining lost profits requires a court to answer a complex counter-

factual—how many units would the plaintiff have sold absent the

infringement—without reliable access to much of the relevant information.

Not surprisingly, the legal guidance provided by the courts of appeal—most

notably the Panduit factors, after the case by the same name

—is

notoriously ambiguous and unhelpful.

However, in an increasing number of cases, the patent holder has no

lost profits, cannot prove lost profits, or simply does not wish to attempt to

do so.

In those cases, the patent plaintiff will instead seek a “reasonable

market.); DSU Med. Corp. v. JMS Co., 471 F.3d 1293 (Fed. Cir. 2006) (“[E]ven though

competitor attempted to suggest that the market included numerous other noninfringing

alternatives to its device, patent owner provided evidence that no other non-infringing

alternatives were acceptable during the necessary time periods.”); Mark A. Lemley,

Distinguishing Lost Profits from Reasonable Royalties, 51 W

M. & MARY L. REV. 655

(2009) ("Courts interpreting this provision have divided patent damages into two groups-

lost profits, available to patent owners who would have made sales in the absence of

infringement, and reasonable royalties, a fallback remedy for everyone else.").

See, e.g., King Instruments Corp. v. Perego, 65 F.3d 941, 953 (Fed. Cir. 1995) (“In

determining the amount of damages to which King was entitled, the district court

considered: (1) the number of lost sales; (2) the gross receipts King would have obtained

from the lost sales had there been no infringement by Tapematic; (3) the cost of sales to be

deducted from gross receipts; and (4) King's profit on the lost sales.”); Versata Software,

Inc. v. SAP Am., Inc., 717 F.3d 1255, 1260 (Fed. Cir. 2013) (“Versata claimed this

consisted of 93 lost sales, and it put forward evidence regarding demand, the absence of

noninfringing alternatives, and the capacity to sell Pricer in this market.”); Micro Chem.,

Inc. v. Lextron, Inc., 318 F.3d 1119 (Fed. Cir. 2003) (“To recover lost profits damages,

patentee must show that “but for” infringement it reasonably would have made additional

profits enjoyed by infringer.”).

Panduit Corp. v. Stahlin Bros. Fibre Works, Inc., 575 F.2d 1152 (6th Cir. 1978). The

four Panduit factors are “(1) demand for the patented product, (2) absence of acceptable

noninfringing substitutes, (3) his manufacturing and marketing capability to exploit the

demand, and (4) the amount of the profit he would have made.” Id. at 1156.

In some but hardly all cases, this will occur because the patent owner is a non-practicing

entity that does not produce a product and thus has no profits to lose. See Christopher A.

Cotropia, Jay P. Kesan & David L. Schwartz, Unpacking Patent Assertion Entities (PAEs),

MINN. L. REV. (forthcoming 2015); David L. Schwartz, The Rise of Contingent Fee

Representation in Patent Litigation, 64 A

LA. L. REV. 335 (2012). On other occasions, the

patent owner might simply believe that it is advantageous to pursue a reasonable royalty

17-June-15 MISUSE OF PATENT LICENSES 5

royalty,” which the court must determine.

Calculating lost profits is by no

means trivial, but it is substantially more determinate and straightforward

than estimating a reasonable royalty. To accomplish this latter task, a court

must attempt to reconstruct a hypothetical negotiation between patent

plaintiff and defendant—which likely never took place—and determine the

amount of money for which the two parties would have agreed to settle.

Worse still, the legal guidance provided to courts and juries is

almost comically counter-productive. When determining reasonable royalty

damages, courts are instructed to consider the fifteen influential Georgia-

Pacific factors,

a laundry list of considerations that shed little light on the

appropriate dollar figures. The Federal Circuit has (correctly) struck down

more determinate types of guidance, such as the “25% rule of thumb,”

according to which 25% of the defendant’s profits should be paid to a patent

defendant as a reasonable royalty.

Similarly, it has (correctly) limited the

number of situations in which a court is permitted to calculate damages

using the “entire market value rule,” whereby a court would base a

reasonable royalty on the full price of the product being sold, rather than the

value of the particular patented component.

But the result has been to

leave courts almost entirely at sea and at the mercy of the parties’ experts

when attempting to assess damages.

In combination, the elimination of the

25% rule and the limits placed on the entire market value rule have left

courts guessing about both (1) the proper royalty rate, and (2) the sales

figure (the “royalty base”) to which it should apply that royalty rate. It is not

surprising, then, that courts have largely floundered when attempting to

estimate reasonable royalties.

instead. Christopher B. Seaman, Reconsidering the Georgia-Pacific Standard for

Reasonable Royalty Patent Damages, 2010 B.Y.U. L. Rev. 1661, 1676–77 (2010); see,

e.g., Monsanto Co. v. McFarling, 488 F.3d 973 (Fed. Cir. 2007).

The Patent Act allows plaintiffs to select among the two options. 35 U.S.C. § 284; Omri

Ben-Shahar, Damages for Unlicensed Use, 78 U. C

HI. L. REV. 7 (2011).

See infra Part I.

These factors derive their name from Georgia-Pacific v. United States Plywood Corp.,

318 F. Supp. 1116 (S.D.N.Y. 1970). Although Georgia-Pacific is merely a district court

case, the Federal Circuit has adopted it as the touchstone for computing reasonable royalty

damages.

Uniloc USA, Inc. v. Microsoft Corp, 632 F.3d 1292 (Fed. Cir. 2011).

VirnetX, Inc. v. Cisco Sys., Inc., 767 F.3d 1308 (Fed Cir. 2014). Courts may use the

entire market value rule only when “it can be shown that the patented feature drives the

demand for an entire multi-component product . . . .” LaserDynamics, 694 F.3d at 67.

For instance, it is much easier to price an entire product—which is being sold in the

market—than a particular component of that product, which might never be sold on its own

for market value. See Brian D. Coggio, Damage Control-What an Adjudged Infringer Can

Do to Minimize the Resulting Damage, 15 AIPLA Q.J. 250, 296 (1987) (“However, the

relative difficulty of establishing apportionment will operate in favor of the application of

the entire market value rule.”).

6 MISUSE OF PATENT LICENSES 24-Feb-15

In the face of these difficulties, it is natural for courts to seek an

alternative means of estimating reasonable royalties. Again, one of the

principal disadvantages of the turn away from injunctive relief as a patent

remedy is that it robs courts of the opportunity to have the parties decide the

value of a patent through arms-length bargaining. An alternative market

mechanism that allows courts to use private valuations to accurately gauge

patent damages would be of obvious value. The solution that courts have

arrived at is to use existing licenses, typically granted by the plaintiff to

third parties, as evidence of the proper measure of damages. A patent

license offers the elusive holy grail: an arms-length transaction between two

private parties that places a monetary value on the patent. Indeed, the

measure of value provided by an existing license is the very first factor

listed by Georgia-Pacific for use in calculating reasonable royalty

damages.

Courts have relied upon existing licenses in calculating damages

for decades,

and the practice has grown even more prominent in recent

years.

At first blush this approach makes sense; if the courts must

reconstruct a hypothetical royalty negotiation, actual pre-existing royalty

agreements might well constitute the best available evidence of the contours

of such a negotiation. Not surprisingly, scholars, commentators, and courts

nearly unanimously bless the use of existing licenses to calculate patent

damages.

But the use of existing licenses to measure reasonable royalty

damages creates three significant and distinct problems. The first is that it

relies upon private information, available only to the parties to the first

licensing agreement, about the plaintiff’s probability of success in litigation.

Every agreement to license a patent is necessarily made in the shadow of

the threat of litigation.

If a patent holder had no means to enforce its

Georgia-Pacific Corp., 318 F. Supp. at 1120 (“1. The royalties received by the patentee

for the licensing of the patent in suit, proving or tending to prove an established royalty.”).

Nickson Indus., Inc. v. Rol Mfg. Co., 847 F.2d 795 (Fed. Cir. 1988).

See, e.g., VirnetX, Inc., 767 F.3d at 1330–31; Apple Inc. v. Motorola, Inc., 757 F.3d 1286

(Fed. Cir. 2014); Versata Software, Inc. v. SAP Am., Inc., 717 F.3d 1255 (Fed. Cir. 2013);

LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51 (Fed Cir. 2012) (validating the

use of some licenses in determining damages and rejecting others); Finjan, Inc. v. Secure

Computing Corp., 626 F.3d 1197, 1212 (Fed.Cir.2010); Ericsson, Inc. v. D-Link Sys., Inc.,

773 F.3d 1201; Waterton Polymer Products USA, LLC v. EdiZONE, LLC, 2014 WL

5782710 (D. Utah 2014); Ultratec, Inc. v. Sorenson Commc'ns, Inc., No. 13-CV-346-BBC,

2014 WL 5080411 (W.D. Wis. Oct. 9, 2014); Gen-Probe Inc. v. Becton Dickinson & Co.,

899 F. Supp. 2d 971 (S.D. Cal. 2012); ReedHycalog, UK, Ltd. v. Diamond Innovations

Inc., 727 F. Supp. 2d 543 (E.D. Tex. 2010); Datatreasury Corp. v. Wells Fargo & Co., No.

CIV.A. 2:06-CV-72 DF, 2010 WL 903259, at *1 (E.D. Tex. Mar. 4, 2010); Spreadsheet

Automation Corp. v. Microsoft Corp., 587 F. Supp. 2d 794 (E.D. Tex. 2007).

See sources cited in supra note 11.

See infra Part II.A.

17-June-15 MISUSE OF PATENT LICENSES 7

patent, no licensee would pay royalties for a license. Accordingly, any

negotiation over royalties will necessarily be based upon the outcome the

parties expected should the case go to trial.

The parties will have to

account for the possibility that courts will find the patent invalid or not

infringed—the possibility, that is, that the licensee will not be forced to pay

anything.

This highlights the second, closely related problem with using

existing licenses to calculate reasonable royalty damages: the procedure is

ineluctably circular.

Licensing agreements are based upon expected

damages awards at trial. But if damages awarded at trial are in turn based

upon licensing agreements, it creates an unconquerable chicken-and-egg

problem. Judicial error with regard to the appropriate measure of damages

will produce smaller royalty amounts outside of litigation, which will in

turn lead to lower judicial calculations of damages, which will then beget

even smaller royalty payments outside of litigation, and so forth.

The third problem stemming from the use of existing licenses is

common to any setting in which a contract between two parties, A and B,

will affect the rights of a third party, C. A and B will always have an

incentive to distort their contract so as to maximize their joint gains against

For instance, suppose that a patent owner P understands that the

licensing agreement it reaches with a licensee L

will affect the damages it

obtains in litigation against a second subsequent licensee, L

, or the royalty

it can negotiate with a third subsequent licensee, L

. P has a strong incentive

to inflate the value of its agreement with L

, perhaps in exchange for

providing L

with something else of value. Courts have to be vigilant in

policing licensing agreements for extraneous considerations before using

those agreements to estimate damages. At the same time, if courts are too

particular in requiring that an agreement include nothing but a license to use

a patented technology, they may find few existing licenses that can serve as

sound bases for calculating damages.

In light of these problems, the question is what role existing licenses

should play in judicial damages calculations. It would seem counter-

productive to throw away information about patent valuations when such

information is so scarce to begin with. But without some means of

accounting for the context within which parties negotiate licenses, there is

doubt as to whether existing licenses can provide reliable evidence of

See Mark A. Lemley & Carl Shapiro, Patent Holdup and Royalty Stacking, 85 TEX. L.

REV. 1992, 1993 (2007); William M. Landes, An Economic Analysis of the Courts, 14 J.L.

& ECON. 61 (1971); Richard A. Posner, An Economic Approach to Legal Procedure and

Judicial Administration, 2 J. L

EGAL STUD. 399 (1973); see also John P. Gould, The

Economics of Legal Conflicts, 2 J. L

EGAL STUD. 279 (1973).

See infra Part II.B.

See infra Part II.C.

8 MISUSE OF PATENT LICENSES 24-Feb-15

reasonable royalty damages. The final Part considers a variety of potential

solutions to these problems. Though several of these solutions show

promise, none comes close to being a complete answer. It may well be that

courts have no choice but to treat existing licenses as providing only the

most limited of guidance when calculating damages at trial.

This Article proceeds in three Parts. Part I describes the courts’

practice of using existing licenses to gauge reasonable royalty patent

damages and outlines the rules that courts have developed to guide that

practice. Part II analyzes the three problems inherent to using existing

licenses to calculate reasonable royalties: courts’ inability to access the

private information that will unlock the license; the circularity involved in

relying upon existing licenses in damages calculations; and the incentives to

distort contracts when a contract between two parties is used to value

property or legal rights in a manner that affects a third party. Part III

considers a variety of potential mechanisms for solving these problems but

concludes that the difficulties they create cannot be entirely ameliorated.

The Article closes with an inquiry into whether there remains any viable

role for existing licenses in the setting of reasonable royalty damages at

trial.

I. LICENSING AGREEMENTS AND REASONABLE ROYALTY DAMAGES

When a patent plaintiff prevails at trial but cannot prove lost profits

or damages—or does not wish to try—the adjudicating court must instead

determine the reasonable royalty that an infringer should have paid the

patent holder to license the patent.

The court must imagine a hypothetical

negotiation between the plaintiff and the defendant, conducted at the

moment before the defendant’s infringement began,

and determine the

royalty the two parties would have settled upon after bargaining at arm’s

length. The court must further assume, for purposes of this hypothetical

negotiation, that both parties know the patent to be valid and infringed—as

the court has just found it to be.

This inquiry is both hypothetical and well

35 U.S.C. § 284 (“Upon finding for the claimant the court shall award the claimant

damages adequate to compensate for the infringement, but in no event less than a

reasonable royalty . . . .”).

See, e.g., Applied Med. Res. Corp. v. U.S. Surgical Corp., 435 F.3d 1356, 1363–64

(Fed.Cir.2006) ( “[T]he hypothetical negotiation relates to the date of first infringement.”);

State Indus., Inc. v. Mor–Flo Indus., Inc., 883 F.2d 1573, 1580 (Fed.Cir.1989) (“The

determination of a reasonable royalty . . . [is based] on what a willing licensor and licensee

would bargain for at hypothetical negotiations on the date infringement started.”).

Lucent, 580 F.3d at 1301; see also Paul M. Janicke, Contemporary Issues in Patent

Damages, 42 A

M. U. L. REV. 691, 722–23 (1993).

17-June-15 MISUSE OF PATENT LICENSES 9

after the fact, and it requires the court to forget many facts that it has

learned and imagine many others that it cannot know.

It is naturally

fraught with both error and complication.

The Federal Circuit

has

compounded the situation by providing only amorphous guidance to the

lower courts. Courts awarding a reasonable royalty are instructed to apply

the fifteen-part test enumerated in Georgia Pacific.

Almost needless to

say, lower courts have found this to be anything but a straightforward

task.

It is thus not surprising that courts have grasped for sources of

market information, believing them superior to the court’s own speculation

and hypothesizing. Georgia-Pacific itself encourages this; the very first

Georgia-Pacific factor is “[t]he royalties received by the patentee for the

licensing of the patent in suit,”

and the second is “the rates paid by the

licensee for the use of other patents comparable to the patent in suit.”

Existing licenses—whether granted by the plaintiff for access to the patent

technology, or purchased by the defendant for similar technologies—are

thus front and center in reasonable royalty calculations.

But this does not mean that all licenses are treated equally. Federal

courts have evolved a set of rules to determine whether an existing license

will be admitted into evidence and, if admitted, what weight it will be

afforded.

First and foremost is the sensible rule that the existing license in

The difficulties that courts encounter in trying to reconstruct this hypothetical negotiation

are further detailed in Parts II and III.

See ResQNet.com, Inc. v. Lansa, Inc., 594 F.3d 860, 869 (Fed. Cir. 2010) (“Determining

a fair and reasonable royalty is often . . . a difficult judicial chore, seeming often to involve

more the talents of a conjurer than those of a judge.”); see also Merritt J. Hasbrouck,

Protecting the Gates of Reasonable Royalty: A Damages Framework for Patent

Infringement Cases, 11 J.

MARSHALL REV. INTELL. PROP. L. 192, 193 (2011) (“Although

the courts have made some progress in other areas of patent law, one troublesome area

remains: the appropriate standard for determining a reasonable royalty damages amount.”).

The Federal Circuit has exclusive appellate jurisdiction from final decisions of district

courts in the area of patent law. 28 U.S.C § 1295(a).

Georgia-Pacific v. United States Plywood Corp., 318 F. Supp. 1116, 1120 (S.D.N.Y.

1970).

See, e.g., Apple, Inc. v. Motorola, Inc., 869 F. Supp. 2d 901, 911 (N.D. Ill. 2012) (“This

is a formidable list . . . . And could a judge or a jury really balance 15 or more factors and

come up with anything resembling an objective assessment?”); see also Bo Zeng, Lucent v.

Gateway: Putting the "Reasonable" Back into Reasonable Royalties, 26 B

ERKELEY TECH.

L.J. 329, 333 (2011) (“In essence, Georgia-Pacific's hypothetical, individually-negotiated

approach complicated reasonable royalty determinations . . . .”).

Georgia-Pacific Corp., 318 F. Supp. at 1120.

Id.

See Zelin Yang, Damaging Royalties: An Overview of Reasonable Royalty Damages,

29 B

ERKELEY TECH. L.J. 647, 668–69 (2014) (“Although reasonable arguments could be

made for each of these factors to be the starting point in determining a royalty rate, the

10 MISUSE OF PATENT LICENSES 24-Feb-15

question must involve a patent very similar (if not identical) to the patent in

suit.

A comparison is not valuable if it is not apples-to-apples, or close to

it. Thus, existing licenses that bundle together multiple patents, or a patent

and something else of value (such as a trademark or trade secret), are

typically not accepted as valid comparisons.

Similarly, courts disfavor

licenses for unrelated technology, on the theory that they may be more or

less valuable than the technology in the patent at suit.

In addition, courts

generally frown upon the use of one type of license—for instance, a lump

sum license—to calculate a different type of license, such as a running

royalty in which the defendant pays per unit sold over time.

In practice,

this means that licenses granted by the patent owner for the same patent are

more commonly used in calculating damages than licenses taken by the

defendant on similar technologies.

Federal Circuit has increased the level of scrutiny in assessing whether licenses are truly

‘comparable.’”).

See Lucent Technologies, Inc. v. Gateway, Inc., 580 F.3d 1301, 1329 (Fed. Cir. 2009)

(“damages award cannot stand solely on evidence which amounts to little more than a

recitation of royalty numbers . . . particularly when it is doubtful that the technology of

those license agreements is in any way similar to the technology being litigated here.”); see

also LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51, 79 (Fed. Cir. 2012)

(“When relying on licenses to prove a reasonable royalty, alleging a loose or vague

comparability between different technologies or licenses does not suffice . . . . We insisted

that the ‘licenses relied upon by the patentee in proving damages [be] sufficiently

comparable to the hypothetical license at issue in suit.’”) (quoting Lucent Technologies,

Inc., 580 F.3d at 1325).

See ResQNet.com, Inc. v. Lansa, Inc., 594 F.3d 860, 873 (Fed. Cir. 2010) (“In sum, the

district court erred by considering ResQNet's re-bundling licenses to significantly adjust

upward the reasonable royalty without any factual findings that accounted for the

technological and economic differences between those licenses and the ′075 patent.”); see

also John Elmore, The Technological Comparability of Patent License Agreements, 46 L

NOUVELLES 115, 116 (2011) (“[C]ase law cautions that patent license agreements

providing substantial non-patent benefits or multiple patents may not be comparable to a

“straight” patent license.”).

See ResQNet.com, 594 F.3d at 869 (“This court has long required district courts

performing reasonable royalty calculations to exercise vigilance when considering past

licenses to technologies other than the patent in suit . . . . Any evidence unrelated to the

claimed invention does not support compensation for infringement but punishes beyond the

reach of the statute.”); see also Zeng, supra note 37, at 356 (“The Federal Circuit has

eliminated unrelated past licenses from consideration in patent damage analyses and should

do so because every licensing agreement is unique.”).

See, e.g., ePlus, Inc. v. Lawson Software, Inc., 764 F. Supp. 2d 807 (E.D. Va. 2011).

See Thomas F. Cotter, Four Principles for Calculating Reasonable Royalties in Patent

Infringement Litigation, 27 S

ANTA CLARA COMPUTER & HIGH TECH. L.J. 725, 748 (2011)

(“Strictly speaking, then, for a license to be economically comparable it should relate to the

same patent or patents at issue”); Roy Weinstein et. al., Taming Complex Intellectual

Property Compensation Problems, 22 F

ED. CIRCUIT B.J. 547, 553 (2013) (“In view of

ResQNet and Lucent, comparable licenses can only include licenses to the patent-in-suit

17-June-15 MISUSE OF PATENT LICENSES 11

Second, and more important for present purposes, courts

and

commentators

generally disfavor licenses that parties negotiated as

settlements to ongoing litigation. Courts have reasoned that litigation

distorts the licensing prices that defendants are willing to pay, skewing

prices upward.

According to these courts, the primary cause of this

distortion is the cost of litigating: in order to avoid litigation costs, patent

defendants might be willing to pay more than they otherwise would to settle

a dispute and license a patent.

For many years, courts flatly refused to

consider any settlement under threat of litigation as reliable evidence of a

patent’s value,

or at minimum the courts greatly discounted the probative

value of such a license.

Those types of settlements were often barred from

itself, essentially removing from consideration licenses contemplated under Georgia-

Pacific Factors 2 and 12.”).

See, e.g., Tights, Inc. v. Kayser-Roth Corp., 442 F. Supp. 159 (M.D.N.C. 1977).

See, e.g., Keele, supra note 11, at 216 (arguing that licenses negotiated during litigation

settlement are highly prejudicial and rarely probative); Tejas N. Narechania & Jackson

Taylor Kirklin, An Unsettling Development: The Use of Settlement-Related Evidence for

Damages Determinations in Patent Litigation, 2012 J.

LAW, TECH & POL’Y 1 (2012)

(arguing that courts should bar all evidence related to settlement of litigation disputes). A

few scholars have argued that licenses negotiated as settlements to litigation should be

allowed into evidence when assessing damages, though this remains a minority viewpoint.

See Conroy et al., supra note 11; Michael Chapman, Using Settlement Licenses in

Reasonable Royalty Determinations, 49 IDEA: The Intellectual Property Law Review 313

(2009).

Matter of Mahurkar Double Lumen Hemodialysis Catheter Patent Litig., 831 F. Supp.

1354, 1379 (N.D. Ill. 1993) (“[P]eople may settle patent litigation to reduce the costs of the

legal process. The terms of a settlement reflect these costs as well as the parties' estimates

about the probable outcome on the merits if the case proceeds . . . .”); Keele, supra note 11,

at 205–06 (noting the effect of litigation costs on settlement value).

Lumen View Tech., LLC v. Findthebest.com, Inc., 24 F. Supp.3d 329, 336 (S.D.N.Y.

May 30, 2014) (“Lumen's motivation in this litigation was to extract a nuisance settlement

from FTB on the theory that FTB would rather pay an unjustified license fee than bear the

costs of the threatened expensive litigation.”); Steering Committee Report, 1989 D

UKE L.J.

811, 823 (1989) (“[T]he threat of unreimbursable litigation costs can give weak claims a

nuisance settlement value they do not deserve.”); J.P. Mello, Technology Licensing and

Patent Trolls, 12 B.U.

J. SCI. & TECH. L. 388, 397 (2006) (“Patent trolls typically demand

licenses that are significantly less than the expected cost that each target company will

incur in litigation. Thus, many target companies opt for the economically efficient path and

pay a license fee to the patent troll rather than incur litigation costs.”).

See, e.g., FED. R. EVID. 408; Rude v. Westcott, 130 U.S. 152, 164 (1889) (“[A] payment

of any sum in settlement of a claim for an alleged infringement cannot be taken as a

standard to measure the value of the improvements patented, in determining the damages

sustained by the owners of the patent in other cases of infringement.”); P

ATENT LAW

DAMAGES & PRACTICE § 3.17 (2014).

See, e.g., Deere & Co. v. Int'l Harvester Co., 710 F.2d 1551, 1557 (Fed. Cir. 1983)

(“[A]s the White license was negotiated against a backdrop of continuing litigation and [the

defendant's] infringement of the Schreiner patent, the district court could properly discount

the probative value of the White license with regard to a reasonable royalty.”).

12 MISUSE OF PATENT LICENSES 24-Feb-15

evidence. However, in the 2010 case ResQNet v. Lansa,

the Federal

Circuit appeared to invite consideration of licenses negotiated as settlements

to litigation, though they would be awarded less evidentiary weight.

Since

2010, courts have occasionally considered licenses negotiated during

litigation as indicators of patent value.

By and large, however, courts have

continued to express a strong preference for licenses negotiated “in the

wild”—outside of litigation.

These licenses form the bulk of sources to

which courts have turned for guidance in setting reasonable royalty

damages.

LaserDynamics v. Quanta Computer provides an illustrative

example.

There, the parties introduced into evidence twenty-nine prior

licenses of the patent in suit.

On appeal, the Federal Circuit noted that the

“vast majority” of these twenty-nine licenses had been negotiated outside of

litigation, and that those licenses provided the most reliable evidence of the

true value of the patent.

The court then singled out for disapprobation a

license negotiated by a firm named BenQ that had arisen under unusual

circumstances:

This settlement agreement was executed within two weeks of the

anticipated trial against BenQ. . . . By the time of the settlement,

BenQ had been repeatedly sanctioned by the district court for

discovery misconduct and misrepresentation. The district court had

ResQNet.com, Inc. v. Lansa, Inc., 594 F.3d 860, 870 (Fed. Cir. 2010).

Id. at 870–71.

E.g., Volumetrics Med. Imaging, LLC v. Toshiba Am. Med. Sys., Inc., No. 1:05CV955,

2011 WL 2470460, at *14 (M.D.N.C. June 20, 2011); Small v. Nobel Biocare USA, LLC,

808 F. Supp. 2d 584, 592 (S.D.N.Y. 2011); Automated Merch. Sys. Inc. v. Crane Co., 279

F.R.D. 366, 373 (N.D. W. Va. 2011).

E.g., Ravo v. Covidien LP, No. CIV.A. 11-1637, 2014 WL 5439787, at *7 (W.D. Pa.

Oct. 24, 2014) (“Settlement agreements are generally not relevant ‘because in the usual

course they do not provide an accurate reflection of what a willing licensor would do in an

arm's length transaction.’”) (citing LaserDynamics v. Quanta Computer, 694 F.3d 51, 77–

78 (Fed. Cir. 2012)); Atlas IP, LLC v. Medtronic, Inc., No. 13-CIV-23309, 2014 WL

5741870, at *6 (S.D. Fla. Oct. 6, 2014) (refusing to consider a license negotiated during

litigation); Sentius Int'l, LLC v. Microsoft Corp., No. 5:13-CV-00825-PSG, 2015 WL

451950, at *12 (N.D. Cal. Jan. 27, 2015) (same). Commentators, even those who have

advocated giving non-zero weight to litigation settlements, have expressed similar

preferences for non-litigation licenses. See Conroy et al., supra note

11, at 302 (suggesting

that licenses negotiated as settlements to litigation should be admitted but arguing that they

are less probative than licenses negotiated outside of litigation); Taylor, supra note 11, at

114 (same); Keele, supra note 11, at 207 (same).

Yang, supra note 40, at 669.

694 F.3d 51 (2012).

Id. at 78.

Id.

17-June-15 MISUSE OF PATENT LICENSES 13

allotted BenQ one-third less time than [the patent owner] for voir

dire, opening statement, and closing argument, had awarded

attorneys' fees to [the patent owner] for bringing the sanctions

motion, had stricken one of BenQ's pleaded defenses, and had

sanctioned BenQ $500,000.00 as an additional punitive and

deterrent measure.

Perhaps not surprisingly, BenQ agreed to license the patent for a great deal

more money than any of the other licensees. BenQ paid $6 million, while

none of the other licenses exceeded $1 million.

The Federal Circuit held

that the district court had abused its discretion in admitting the BenQ

license into evidence.

It declared that the “unique coercive circumstances”

surrounding this license made it a particularly unreliable gauge of the

patent’s value.

There can be little doubt that BenQ’s setbacks in its litigation

against LaserDynamics (the patent owner) contributed to its larger licensing

figure. Had BenQ not been embroiled in litigation against LaserDynamics,

and had it not been faring so poorly, it would very likely have settled for

much less—probably $1 million or less, in line with the other licensees.

Nonetheless, I wish to suggest that the Federal Circuit had it exactly

backwards: the BenQ license, and not the other twenty-eight licenses, was

the most accurate indication of the “true” value of the patent. The court

should have admitted the BenQ license and excluded the others, or at least

afforded them little weight. More generally, courts’ approach to using

existing licenses to determine patent damages at trial is both incoherent and

backwards—a rare combination. Courts misunderstand patent licenses at

their foundation and in so doing have devised doctrines that are not merely

misguided, but counterproductive. The next Part explains how courts have

gone astray, and the final Part investigates what might be done to correct

them.

II. THREE BARRIERS TO USING EXISTING LICENSES

A. Existing Licenses and Private Information

No one would ever license a patent absent the threat of litigation. If

a patent holder could not threaten to enforce its patent against a putative

Id. at 58.

Id.

Id. at 78.

14 MISUSE OF PATENT LICENSES 24-Feb-15

licensee in court, the licensee would have no reason to negotiate a license in

the first place.

Patent licenses are best understood as civil settlements in

anticipation of possible litigation.

Accordingly, the licensing fee for a given patent will depend upon

the parties’ expected outcomes at trial.

That is not to say that licensing

amounts are driven only by expected trial outcomes. The cost of going to

trial will factor into the parties’ calculations as well, and thus in some cases

patent holders may be able to obtain licensing fees greater than the expected

outcome at trial.

But expected trial outcomes will necessarily play a

significant role. To win at trial, the plaintiff has to show both that the patent

is valid and that the defendant infringed the patent. Thus the set of potential

trial outcomes includes the possibility that the patent will be found invalid,

or that the court will find that the defendant did not infringe the patent, and

thus that there will be no award of damages.

In formal terms, the

plaintiff’s expected outcome at trial is p × d – c, where p is the probability

that the patent will be held valid and enforced, d is the likely amount of

damages the court will assess, and c is the cost of litigation. Conversely, the

defendant’s expected payout is p × d + c. The plaintiff will be willing to

grant a license—that is, settle—for any amount greater than p × d – c, while

the defendant will be willing to purchase a license for any amount less than

p × d + c. Accordingly, if the two parties are able to agree upon a license, it

Keele, supra note 11, at 205 (“Royalties are usually paid to avoid litigation--most people

who thought that they could infringe a patent with impunity would likely do so.”); Dov

Greenbaum, Academia to Industry Technology Transfer: An Alternative to the Bayh-Dole

System for Both Developed and Developing Nations, 19 F

ORDHAM INTELL. PROP. MEDIA &

ENT. L.J. 311, 388 (2009) (“…with no potential enforcement by the owner of the IP,

potential licensees may see no incentive to ever license the patent; infringing at will.”);

Brian Fung, The Man Who Invented Priceline.com Wants to Shake Up America’s Approach

to Patents, W

ASHINGTON POST, Jan. 2, 2015, available at

http://www.washingtonpost.com/blogs/the-switch/wp/2015/01/02/the-man-who-invented-

priceline-com-wants-to-shake-up-americas-patents/ (“I learned very quickly that nobody

wanted to license my solutions unless I threatened to sue them—and in most cases, when I

actually sued them . . . .”).

See infra Part II.B (explaining that there is no reason to license a patent other than to

avoid litigation); Mark A. Lemley & Carl Shapiro, Patent Holdup and Royalty Stacking, 85

EX. L. REV. 1992, 1993 (2007) (analyzing licenses as litigation settlements).

Keele, supra note 11, at 205–06.

Lemley & Shapiro, supra note 65, at 2000–09 (analyzing mechanisms that can drive

licensing prices upward).

Id. (“Like any other settlement, the amount a party is willing to pay or accept for a

litigation license . . . generally consists of three core components: the likelihood of liability

. . . ; the expectation value of the damages . . . ; and the party's expected litigation costs . . .

.”).

17-June-15 MISUSE OF PATENT LICENSES 15

will be for a royalty R such that p × d – c < R < p × d + c.

The midpoint of

that range is p × d, and thus the parties should be expected to agree upon a

royalty in the vicinity of p × d: R ≈ p × d.

By way of example, imagine that a patent holder (P) and a putative

licensee (L

) are negotiating a license over a particular piece of patented

technology. Suppose the parties recognize that a court is only 25% likely to

hold the patent valid and infringed by L

. Suppose further that the parties

agree that if the patent is found valid and infringed, the court is likely to

assess $20 million in reasonable royalty damages. Finally, suppose that the

patent litigation will cost each party $2 million. The patent holder’s

expected outcome from trial is $20 million × 25% – $2 million = $3 million.

The putative licensee’s expected trial outcome is –$20 million × 25% – $2

million = –$7 million. P would be willing to grant a license for any amount

greater than $3 million; L

would be willing to pay anything less than $7

million. The parties should be expected to negotiate a reasonable royalty

near the midpoint of those two figures: approximately $5 million.

The

license that the parties eventually negotiate will be driven by the expected

damages at trial as well as the likelihood that the patent will be found valid

and infringed in the first instance.

Now imagine that P sues a second putative infringer (or licensee) L

for infringing the same patent. The case goes to trial, and P wins: the court

holds that the patent is valid and has been infringed by L

. In order to

determine damages, the court is expected to assess the outcome of a

hypothetical negotiation between P and L

in which the parties agree upon a

reasonable royalty. Under governing Federal Circuit law, the court must

assume that this hypothetical negotiation took place immediately before the

moment of infringement—before L

began its infringing activities.

Critically, however, the court must also assume that the patent is valid and

infringed and that both parties know it to be valid and infringed.

In that

sense, the trial has officially resolved any ambiguity or uncertainty that

Landes, supra note 27, at 67 (analyzing a model of settlement in which litigation costs

create bargaining space within which parties can settle); Posner, supra note 27, at 414

(same).

Posner, supra note 27, at 414 (analyzing this dynamic).

See Taylor, supra note 11, at 115 (“Negotiated royalties thus include discounts based on

risk borne by the patent owner associated with proving liability, relief, and enforceability . .

. .”).

Lucent Technologies, Inc. v. Gateway, Inc., 580 F.3d 1301, 1324 (Fed. Cir. 2009)

(“[T]he hypothetical negotiation or the ‘willing licensor-willing licensee’ approach,

attempts to ascertain the royalty upon which the parties would have agreed had they

successfully negotiated an agreement just before infringement began.”).

Id. at 1325 (“The hypothetical negotiation also assumes that the asserted patent claims

are valid and infringed.”); see also LaserDynamics v. Quanta Computer, 694 F.3d 51, 77

(2012) (“. . . validity and infringement of the patent not being disputed.”).

16 MISUSE OF PATENT LICENSES 24-Feb-15

previously surrounded the patent. As a matter of law, P has prevailed over

, with all of the attendant consequences.

The court is thus faced with the task of reconstructing the price that

would have paid to license the patent from P if both parties had agreed

that the patent was valid and infringed. It is natural for a court to look to the

existing license between P and L

for guidance. But when that license was

negotiated, the parties could not have been certain that the patent was valid

and infringed—or, at minimum, there is no way for a court to know whether

P and L

believed that the patent was valid and infringed and no reason to

believe that they viewed it as a certainty. That is private information,

inaccessible to the court. P and L

might have believed that it was 100%

likely or nearly 100% likely that the patent was valid and infringed; or they

might have believed it to be 50% likely, or 25% likely, as in this

hypothetical.

Accordingly, the court should not simply use the licensing figure

from the agreement between P and L

—$5 million—when calculating

damages in the P v. L

litigation. The parties negotiated the $5 million

royalty with the understanding that there was some (likely non-zero)

probability that the patent would not be found valid and infringed. Now that

P has prevailed in its suit against L

, that probabilistic inquiry has been

resolved in favor of P, and the court is expected to assess damages as if the

parties were negotiating under the belief that the patent was valid and

infringed. If the court is to use the license between P and L

as evidence of

damages in the litigation against L

, it must determine the fee P and L

would have agreed upon had they believed that the patent was 100% likely

to be valid and infringed.

But the court cannot determine this hypothetical licensing fee

without knowing what P and L

believed were their probabilities of success

at trial. That is, the court is attempting to determine d—the parties’ view of

Most scholars and practitioners of patent law understand that patents are probabilistic

entities through and through. Mark A. Lemley & Carl Shapiro, Probabilistic Patents, 19 J.

ECON. PERSP. 75 (2005) (summarizing and analyzing this mode of thinking about patents).

Patents that are valid with 100% certainty or infringed by a given technology with 100%

certainty are like the Loch Ness Monster: various people claim to have seen them, but most

informed parties realize that they cannot possibly exist. Most patent scholars would say the

same thing about even patents that have been adjudicated. Just because one court (or jury)

has found a patent valid and infringed is not a necessary guarantee that a different court or

jury would have reached the same result. See David L. Schwartz, Practice Makes Perfect?

An Empirical Study of Claim Construction Reversal Rates in Patent Cases, 107 M

ICH. L.

REV. 223 (2008) (demonstrating that even very experienced trial courts have their patent

decisions overruled at a high rate). Nonetheless, it is inherent to a court’s self-image and

the very nature of a trial that the law would view the decision of a court as final and

determinative—at least with respect to the parties and issues involved in that case—and

would treat it as eliminating any ambiguity accompanying the legal questions presented.

17-June-15 MISUSE OF PATENT LICENSES 17

what damages a court would assess if the patent were found valid and

infringed—by observing R, the actual royalty that the parties negotiated.

But d is not the same as R; the negotiated royalty R would normally be

approximately p × d, where p is the probability that the patent will be found

valid and infringed. By simple rearrangement, d ≈ R / p. That is to say, a

court cannot use an existing royalty (R) to determine what damages it

should assess (d) without knowing p, the probability of success that the

parties assigned to the patent.

And there is almost no way for the court to

reliably determine p, absent unusual circumstances (to be discussed later).

The expected probability of success on the merits, p, is the parties’ private

information, unknowable to the court. There is no way for the court to

determine what probabilities P and L

assigned to the patent litigation

merely from scrutinizing the licensing agreement. The court can hardly

force representatives of the two sides to testify to their internal perceptions

of the patent at the time of the settlement.

And without that information

the court cannot use the existing license to reliably estimate the value of the

patent. All it can know is that P and L

valued the patent—if valid and

infringed—at no less than $5 million. The existing license thus provides a

floor for valuing the patent, not a reasonable estimate.

It is easy to see that the court would err if it simply assessed

damages against L

equal to the amount of the prior license—here, $5

million.

Consider, for example, a subsequent putative infringer, L

, who is

accused of infringement by P. L

would understand that if it chose to litigate

against P and lost, it would likely face damages of $5 million—the amount

of the licensing agreement between P and L

. Outside of litigation, it would

be willing to pay P a royalty equal to $5 million, discounted by the

probability that P will win at trial. Imagine that P’s probability of success in

Of course, it is entirely possible that the two parties involved in a license—P and L

—did

not actually agree on the probability that the patent was valid and infringed. P might have

thought that the patent was 50% likely to be valid and infringed and believed that a court

would award it $10 million if it prevailed at trial, while L

might have believed that the

patent was only 25% likely to be valid and infringed but anticipated a $20 million verdict if

P prevailed. The two parties would have arrived at the same licensing figure through

different routes. (If they arrive at different licensing figures, with P’s substantially higher,

then they will not be able to negotiate a license and will end up in litigation.)

If this is the case, then the existing license is an even less valuable guide to the

damages calculation. Instead of it being difficult or impossible for the court to determine

the parties’ joint belief about the proper amount of damages, there is no joint belief. The

court might as well ignore the license. In this respect, the analysis that follows stacks the

deck in favor of using existing licenses to assess patent damages. If this analysis

nonetheless compels the conclusion that existing licenses are not reliable guides to patent

damages, it will not be for lack of having granted those licenses the benefit of the doubt.

This point is developed further in infra Part III.B.

Contra Cotter, supra note 45, at 752-53 (2011) (arguing precisely the opposite).

18 MISUSE OF PATENT LICENSES 24-Feb-15

a suit against L

is still only 25%. (The litigation between P and L

might

have established that the patent is almost certainly valid, but L

’s product

might not infringe.

) In this case, L

’s expected outcome, should it go to

trial, is only $5 million × 25% = $1.25 million. P and L

will likely settle for

approximately that amount. This stands in stark contrast to the $5 million

license that P negotiated with L

. The only thing that has changed to drive

down the licensing price of the patent is the court’s misinterpretation of the

licensing agreement between P and L

Stated more formally, P and L

negotiated a royalty R where R ≈ d ×

p. Then, the court in P v. L

litigation erred by awarding damages in the

amount of R, rather than attempting to determine d. Now, L

recognizes that

if it loses at trial, it will only be forced to pay d × p. Accordingly, it is only

willing to settle for (d × p) × p, or d × p

. The court’s failure to understand

that prior licenses are discounted by the probability of success at trial, and

its use of such licenses as guidelines for subsequent damages awards,

artificially reduces the value of the patent and the royalties that patent

holders will receive.

B. Circularity

The preceding analysis should make clear that the use of past

licenses to determine patent damages is plagued by a fundamental problem

of circularity. Licenses are necessarily negotiated in the shadow of trial: the

royalty depends upon the parties’ expected outcomes at trial.

When courts

use existing licenses to determine damages at trial, the tiger is chasing its

The fact that a patent has been judged valid in one trial does not necessarily mean that it

must or will be judged valid in another. Under the doctrine of non-mutual collateral

estoppel, a patent plaintiff’s judgment against one party is not binding against a different

party who was not involved in the initial case. Shelcore, Inc. v. Durham Indus., Inc., 745

F.2d 621, 627 (Fed. Cir. 1984) (Holding that an earlier determination of patent validity had

no stare decisis effect); Timothy Denny Greene, ‘All Substantial Rights’: Towards Sensible

Patent Licensee Standing, 22 F

ED. CIR. B.J. 1, 14-19 (2012). However, the initial validity

judgment is still persuasive precedent, and so as a practical matter a patent that has once

been found valid is likely to be found valid again. Gillette Co. v. S.C. Johnson & Son, Inc.,

919 F.2d 720, 723 (Fed. Cir. 1990) (“The fact that the validity of those claims has

previously been upheld in an earlier litigation is also to be given weight, though not stare

decisis effect.”).

See Cotter, supra note 45, at 732.

See, e.g., VirnetX v. Cisco Sys, Inc., 767 F.3d 1308, 1330-31 (Fed. Cir. 2014) (making

this error); Versata Software, Inc. v. SAP Am., Inc., 717 F.3d 1255, 1255(Fed. Cir. 2013)

(same); LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51, 66 (Fed. Cir. 2012)

(same); Finjan, Inc., v. Secure Computing Corp., 626 F.3d 1197, 1212 (Fed. Cir. 2010)

(same); Ericsson, Inc. v. D-Link Sys. Inc., 773 F.3d 1201 (Fed. Cir. 2014) (same).

Id. at 2021 (discussing the effects of bargaining in the shadow of trial on licensing

behavior).

17-June-15 MISUSE OF PATENT LICENSES 19

own tail. Trial outcomes cannot depend on licenses if licenses depend on

trial outcomes.

Treating an existing licensing agreement as if it represents a true

valuation of a valid and infringed patent will force the patent into an

artificial downward spiral in value. A license will drive expected trial

outcomes lower, which will in turn drive future licenses lower, which will

in turn drive future expected trial outcomes even lower, and so forth. This

type of positive feedback loop is unsustainable and will lead to ever greater

distortions.

This spiral will result even if the patent is never litigated. It relies

only on parties correctly understanding how a court will behave and how it

will treat existing licenses. Consider the previous example, in which L

agrees to license P’s patent for $5 million, with both parties calculating that

P is approximately 25% likely to succeed at trial and the court likely to

award $20 million in damages if P prevails. Suppose that P now approaches

and threatens litigation if L

does not agree to license the patent. If both P

and L

understand that the court will use the licensing agreement between P

and L

to set damages in the trial, then the two parties will recognize that L

faces only $5 million in potential liability if it goes to trial. Accordingly, if

P is 25% likely to prevail at trial, L

will agree to license the patent for

approximately $1.25 million.

Now suppose that P approaches L

and again threatens litigation if

does not agree to license the patent. What royalty can they be expected to

negotiate? There are now two existing licenses: the $5 million license

between P and L

and the $1.25 million license between P and L

. Suppose

that L

, like L

and L

, is 25% likely to be held liable for infringement in the

event of a trial. What liability would L

face? If the court were (incorrectly)

treating existing licenses as indicative of a patent’s value, it would likely

assess damages in an amount between $1.25 million (the less expensive

license) and $5 million (the more expensive license). The midpoint of that

range—$3.125 million—is a reasonable estimate. If P and L

understand

this fact, then they would likely negotiate a license for approximately 25%

of $3.125 million, or approximately $780,000.

As P negotiates with L

, and L

, the value of the patent has decreased from $5 million to $1.25

million to $780,000, all without the patent ever seeing the inside of a

courtroom.

The downward spiral is driven entirely by the parties’ belief

$781,250, to be exact.

These numbers are of course merely hypothetical; the precise metes and bounds of this

downward spiral will depend on the values at issue in any given case. The more general

point is that this decay in value will occur in any case in which the patent owner is less

than 100% certain to prevail. In practice, that means that it will occur in every case; no

patent owner can ever be completely assured of victory.

20 MISUSE OF PATENT LICENSES 24-Feb-15

that the court will improperly rely upon prior licenses as evidence of the

patent’s value.

As much as courts would like to rely upon market measures in

estimating damages, there is no reliable route out of this circularity.

The

reason is that patent licensing fees can only ever be grounded in a threat of

suit, and thus in the parties’ best estimate of what a court will force them to

pay.

There is simply no reason to license a patent other than to alleviate

the threat of suit. It is not as if any technology is actually being transferred

when a patent is licensed; the public patent document already discloses the

technology on its face, and a putative licensee can read the patent without

licensing it.

It is of course possible that the patent holder would transfer

technical knowledge along with a license for the patent,

and this latent

knowledge may well be more valuable than any technical information

disclosed by the patent itself.

But this transfer amounts to a provision of

valuable information and services above and beyond a license for the patent

itself. Courts have regularly refused to use licenses that involve a transfer of

Of course, some scholars have theorized that licensing fees are already too high, driven

upward by the patent owner’s ability to obtain an injunction or courts’ own

miscalculations. See, e.g., Lemley & Shapiro, supra note 65, at 2019 (analyzing various

dynamics that can lead to excessive licensing royalties and patent damages). Some might

suggest that the downward spiral caused by misuse of existing licenses is a necessary

corrective to this trend. This is not impossible, but it is highly unlikely. If the two effects

balance one another, it would be through sheer fortuity. No sound long-term legal regime

should rely on courts making two types of legal mistake, and hoping that each mistake

counteracts the other. It is far better to attempt to correct both mistakes. Here, that means

grappling with the problems of private information and circularity endemic to existing

licenses.

Contra Taylor, supra note 11, at 142–43 (suggesting that making certain assumptions

about infringement and validity offer a “partial solution” to this circularity). As this section

and the sections that follow will explain, there is no egress from this circularity. Indeed, it

is the circularity that renders insurmountable the problems created when courts base patent

damages on existing licenses.

See Mark Schankerman & Suzanne Scotchmer, Damages and Injunctions in Protecting

Intellectual Property, 32 RAND J. Econ. 199, 200 (2001).

See Robin Feldman & Mark A. Lemley, Does Patent Licensing Mean Innovation?, at

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2565292 (unpublished manuscript

2015) (surveying parties involved in patent licensing and finding that licenses rarely

involve the transfer of technical information).

See Peter Lee, Transcending the Tacit Dimension: Patents, Relationships, and

Organizational Integration in Technology Transfer, 100 C

AL. L. REV. 1503 (2012)

(describing the transfer of tacit information that often accompanies patents).

See Lisa Larrimore Ouellette, Do Patents Disclose Useful Information?, 25 HARV. J. L.

& TECH. 532 (2012) (surveying scientists on their use of the information disclosed in

patents and finding that patents are less-than-perfect disclosure devices).

17-June-15 MISUSE OF PATENT LICENSES 21

more than just patent rights as a guide to patent damages in litigation.

This

is appropriate; if a patent licensing agreement simultaneously involves the

transfer of what amounts to a valuable trade secret, it cannot serve as a

reliable guide to pricing a patent license that involves no such transfer.

The point is that patent licenses are inherently parasitic on litigation:

without the threat of litigation, there would be no licensing. This is why it is

incoherent for courts to refuse to consider licenses negotiated during

litigation or the threat of litigation. Whether the courts realize it or not, there

is no other context in which licenses might arise. If licenses are parasitic

upon litigation, litigation cannot also rely upon licenses for guidance. At the

heart of judicial practice lies a conceptual impossibility.

What, then, is the “true” value of a patent? The problem, as I have

argued, is that there is no inherent value to the patent—it is worth only what

a court will force a party to pay. One possible way out of this quagmire is

that a patent is worth whatever price the parties would bargain to if the

court found the patent valid and infringed and awarded an injunction to the

patent holder. But this answer is both unhelpful in practice and untrue in

theory. As a theoretical matter, the prices that defendants pay to lift

injunctions often reflect holdup costs, assuming the defendants have already

invested in producing the infringing good.

These holdup costs are an

artifact of the plaintiff having the defendant over a barrel, not a true

measure of what the defendant would have paid before the infringement

began. And in practice, injunctions are sufficiently rare that it is unlikely

that a court will any to use as a model. What are the odds that a particular

patent plaintiff would have previously won a verdict against another

defendant, been granted an injunction, and then licensed the patent to the

ResQNet.com, 594 F.3d at 870 (“Dr. David based his damages on seven ResQNet

licenses, five of which had no relation to the claimed invention. These five re-branding or

re-bundling licenses . . . furnished finished software products and source code, as well as

services such as training, maintenance, marketing, and upgrades, to other software

companies in exchange for ongoing revenue-based royalties.”).

Oracle Am., Inc. v. Google Inc., 847 F. Supp. 2d 1178, 1187 (N.D. Cal. 2012)

(“Damages experts cannot use noncomparable licenses, with little relationship to the

claimed invention or parties-in-suit, as a basis for calculating reasonable royalties.”);

Axcess Int'l, Inc. v. Savi Technologies, Inc., No. 3:10-CV-1033-F, 2013 WL 6839112, at

*8 (N.D. Tex. Jan. 25, 2013) (“With regard to the non-comparable licensing agreements

analyzed by Dr. Hakala, the Court is of the opinion that they provide no assistance to his

analysis. The Federal Circuit has made clear that ‘[a]ny evidence unrelated to the claimed

invention does not support compensation for infringement but punishes beyond the reach of

the statute.’ ResQNet.com, Inc., 594 F.3d 869. Therefore, such analysis fails to ‘carefully

tie proof of damages to the claimed invention's footprint in the market place.’”).

Lemley & Shapiro, supra note 65, at 2019 (analyzing various dynamics that can lead to

excessive licensing royalties and patent damages).

22 MISUSE OF PATENT LICENSES 24-Feb-15

defendant? Given the turn against injunctions—the point with which this

article begins—this circumstances must be very uncommon.

Accordingly, it makes sense to think of a patent’s value as whatever

a court would force a defendant to pay, absent any consideration of existing

licenses. That is to say, it is whatever figure a court would arrive at after

using the Georgia-Pacific factors other than the factors that direct a court to

consider existing license. These factors include “the nature and scope of the

license,” the “established profitability of the products,” the “utility and

advantages of [the patent] over old modes and devices,” “the extent to

which the infringer has made use of the invention and the value of such

use,” and so forth.

If courts rely upon these economic factors, and

licensing fees are based upon these court decisions, there is no circularity.

This is the only coherent and practical way to conceptualize the value of a

patent. The problem is that it is difficult for courts to estimate these

values—hence the desire for market measures in the first place.

Of course, the problem of circularity is not unique to patent law. At

a deep level, the value of goods and legal rights in the marketplace will

always depend to at least some degree on predicted outcomes in court.

Whenever a court uses a market transaction to value a good or a legal right,

the potential for circularity exists. This is most evident in negotiations over

a surplus, where there is no clear right or wrong answer. For instance, a

union and an employer bargaining over wages will sometimes agree to

submit the dispute to arbitration. In reaching a decision, the arbitrator will

look to agreements that similarly situated parties have reached in the past.

Those past agreements, in turn, will depend to at least some degree on what

the parties would have expected an arbitrator to decide. Private contracts

and arbitration decisions are locked in a circle.

(Other legal issues, such

as the standard of care in tort law, can similarly give rise to circularities. If

the standard of care depends on standard industry practices, and standard

industry practices depend on the level of care a court deems necessary, the

same type of circularity arises.)

In most cases, however, the influence of judicial decisions on

market values is very slight. Imagine a situation in which A steals B’s

bicycle and B sues A for compensation. If a court finds for B, it will

presumably look to the market price of the bicycle to determine the

appropriate compensation. At some very deep level, that market price could

Georgia-Pacific, 318 F. Supp. At 1119-20.

This can occur in other contractual contexts as well. See Jim Leitzel, Damage Measures

and Incomplete Contracts, 20 RAND J. Econ. 92 (1989) (suggesting that courts can create

a circularity if they use typical private reliance as a measure of reasonable reliance

damages, which in turn will influence the degree to which parties are willing to rely upon

promises).

17-June-15 MISUSE OF PATENT LICENSES 23

depend on a judicial decision. A putative bicycle purchaser might instead

consider stealing the bicycle and taking his chances in court. In reality,

though, bicycle ownership rights are backed by threat of injunction, or jail

time, or reputational sanctions, or any number of other factors beyond the

price a thief will be forced to pay. There is no real circularity.

Although the problem of circularity in valuation is not unique to

patent law, it is especially stark and critical in that context. Unlike most

other goods, the value of a patent depends entirely on its likely fate in court.

Courts, for their part, have emphasized their desire to rely upon existing

licenses to value patents whenever possible. The circularity problem thus

squarely infects a broad swath of patent cases.

C. Contracts with Third-Party Effects

The use of existing licensing agreements to determine patent

damages raises another problem common to a wide variety of contracts that

affect the rights and duties of third parties: it creates incentives to

manipulate the value of the contract. Judicial use of licenses in valuing

patents rests on the notion that the patent is being negotiated at arm’s length

between two parties who are dividing a joint surplus.

That is, neither party

has any incentive to give the other side any consideration beyond what it is

due in the course of the negotiation.

In other words, for a license to be

reliable evidence, a court must believe that the parties are operating in good

faith and at arm’s length to value the patent. The reliability of the license

depends on the presumed behavior of the parties.

But in many contractual settings, including many patent licenses, the

two parties to the contract are not the only ones whose rights or interests

may be at issue. There is of course a well-known literature in the law of

Transocean Offshore Deepwater Drilling, Inc. v. Maersk Drilling USA, Inc., 699 F.3d

1340, 1357 (Fed. Cir. 2012) (“Damages for patent infringement based on hypothetical

negotiation for reasonable royalty seeks to determine the terms of the license agreement the

parties would have reached had they negotiated at arms length when infringement began.”);

Lucent, 580 F.3d at 1324 (“Two alternative categories of infringement compensation are

the patentee's lost profits and the reasonable royalty he would have received through arms-

length bargaining.”); Mark A. Lemley & Carl Shapiro, A Simple Approach to Setting

Reasonable Royalties for Standard-Essential Patents, 28 B

ERKELEY TECH. L.J. 1135, 1147

(2013) (“Under patent law, a reasonable royalty normally is based on a hypothetical, arms-

length negotiation between a willing buyer and a willing seller that takes place at the time

the infringement begins.”).

This is also true as a general matter. Courts typically use recent sales, negotiated at arm’s

length, as an indication of fair market value, absent some reason to believe otherwise. See,

e.g., Schonfeld v. Hilliard, 218 F.3d 164, 178 (2d Cir. 2000).

24 MISUSE OF PATENT LICENSES 24-Feb-15

contracts on third-party beneficiaries to contracts.

But the issue of third

parties arises with special force when a contract between A and B affects

how a court will value some property or service in future litigation between

A and a third party, C. For instance, imagine a contract between an

automobile owner A and insurer B. A contracts with B to insure A’s

automobile in the amount of $10,000 and pays a premium based upon that

amount. Under normal circumstances, A has no reason to insure the

automobile for more than it is worth (unless A plans to commit fraud).

The

greater A’s insured amount, the higher the premium that A must pay.

has no reason to pay a premium to purchase insurance greater than the

amount of loss that A will actually suffer.

100

Now suppose that A has reason to believe that A is likely to be

involved in an automobile accident in which the other driver is at fault.

(Perhaps A drives a substantial distance each day and has noticed a

significant number of reckless drivers along the route.) If A’s automobile is

wrecked in an accident with a third-party driver C, and C is at fault, C will

be liable to A for the value of the automobile.

101

A court might attempt to

assess that value independently, by scrutinizing the make, model, year and

13 WILLISTON ON CONTRACTS § 37 (4th ed. 2010); 9-43 CORBIN ON CONTRACTS § 43

(2014); H

OWARD O. HUNTER, MODERN LAW OF CONTRACTS § 20 (2014); Melvin Aron

Eisenberg, Third-Party Beneficiaries, 92 C

OLUM. L. REV. 1358, 1359 (1992); Robert S.

Adelson, Third Party Beneficiary and Implied Right of Action Analysis: The Fiction of One

Governmental Intent, 94 Y

ALE L.J. 875 (1985);.Anthony Jon Waters, The Property in the

Promise: A Study of the Third Party Beneficiary Rule, 98 H

ARV. L. REV. 1109 (1985).

See, e.g., Richard A. Posner, Efficient Responses to Catastrophic Risk, 6 CHI. J. INT'L L.

511, 523 (2006) (explaining that there is no reason to expend resources beyond the point at

which marginal costs exceed marginal benefits, and thus beyond the point at which there

would be insurance for more than the value of a loss).

Ronen Avraham, The Economics of Insurance Law-A Primer, 19 CONN. INS. L.J. 29, 58

(2012); see also Steven P. Croley & Jon D. Hanson, The Nonpecuniary Costs of Accidents:

Pain-and-Suffering Damages in Tort Law, 108 H

ARV. L. REV. 1785, 1917 n.27 (1995)

(explaining the manner in which premiums are typically calculated).

100

Of course, generally speaking an insurance company will not sell insurance worth more

than the replacement cost of property for fear of moral hazard. The concern is that the

insured will take less care now that she is insured—for instance, driving more recklessly—

particularly if she is insured for more than the value of the property. However, an insurance

company may not know the insured’s subjective valuation of the property. The insurer also

may not have information regarding defects to the property that lower its value. So it is

entirely possible that an insured party could end up with insurance greater than the value of

the property being insured.

101

Arthur Best, Impediments to Reasonable Tort Reform: Lessons from the Adoption of

Comparative Negligence, 40 I

ND. L. REV. 1, 6 (2007); Christopher Curran, The Spread of

the Comparative Negligence Rule in the United States, 12 I

NT'L REV. L. & ECON. 317, 322

(1992) (“While there are exceptions to the rule, in general, under comparative negligence

the courts use the degree of negligence of the parties to an accident to determine the

percentage of the costs of the accidents each party will pay.”).

17-June-15 MISUSE OF PATENT LICENSES 25

prior condition of the automobile, but that task could be complicated

because the automobile is now in pieces. Alternatively, the court might

attempt to value the automobile by looking to the value of the insurance

agreement between A and B.

This creates an incentive for A to insure the automobile for more

than it is worth. If the automobile is worth only $8000 to A, but the court

assigns it a value of $10,000 because A has contracted for insurance in that

amount, A will pocket a profit of $2000 after C totals A’s car. Accordingly,

depending upon the additional premium that A must pay to insure the

automobile for $10,000 rather than $8000, and depending upon A’s

perception of the probability that the automobile will be damaged by a third

party, A may have an entirely rational reason for insuring the automobile for

more than it is worth.

102

In essence, A can use the contract with B (the

insurer) to artificially inflate the perceived value of the automobile,

anticipating that a third party (C) will later be forced to make a payment

based upon that inflated value.

The litigation related to the attacks on the World Trade Center on

September 11, 2001, offers an example of how these incentives might

operate.

103

In July of 2001, a real estate developer (World Trade Center

Properties, or “WTCP”) leased the Trade Center from its owner, the Port

Authority of New York and New Jersey, for approximately $2.8 billion.

104

When the towers were brought down by terrorists using airplanes as

weapons, WTCP sued the airlines for negligence, arguing that the terrorists

would not have been able to take control of the airplanes had the airlines

taken reasonable care in securing them against hijackers.

105

The court held

that the fair measure of the Trade Center’s market value was the lease that

WTCP had just signed with the Port Authority—$2.8 billion.

106

As the

Trade Center’s replacement cost was much higher, this was the full

recovery to which the WTCP would be entitled.

107

Here, as in the

hypothetical example above, the court relied upon the arm’s length bargain

102

For instance, suppose that A must pay an additional $10 per year to insure the

automobile for $10,000 instead of $8000, but A believes there is a 1% chance that she will

be in an accident that is the fault of the other driver. By insuring her automobile for

$10,000 instead of $8000, A must pay an additional $10/year but stands to gain in

expectation $2000 × 0.01 = $20/year. This creates an incentive for A to purchase more

insurance for her car than it is worth.

103

In re September 11th Litigation, 590 F. Supp. 2d 535 (S.D.N.Y. 2008).

104

Id. at 536.

105

Id.

106

Id. at 540-44. New York law, which governed the case, provided that a tort plaintiff

whose property was damaged was entitled to the lesser of (1) the property’s market value,

or (2) its replacement cost. Id. at 541 (citing Hartshorn v. Chaddock, 31 N.E. 997, 998

(1892)).

107

Id. at 541.

26 MISUSE OF PATENT LICENSES 24-Feb-15

between WTCP and the Port Authority to establish the market value of the

Trade Center. As the court explained, “Generally, a recent sale price for the

subject asset, negotiated by the parties at arm’s length, is the best evidence

of its market value.”

108

When it signed the lease with the Port Authority, the WTCP

undoubtedly understood that a court would look to that lease to estimate the

value of the Trade Center if it was damaged by a tort. The WTCP thus had

an incentive to artificially inflate the rental price. The Trade Center had

already once been subjected to a terrorist attack, and the WTCP insured the

buildings for over $3.5 billion against any damage, including damage from

a terrorist act.

109

Of course, it would not have made sense for the WTCP

simply to pay a higher price for additional insurance. There was only some

small probability that the Trade Center would be damaged by a third party

(who could then be made to pay), and so each additional dollar that it paid

to the Port Authority would likely lead to only a few additional cents of

recovery.

But the price is rarely the only term in a contract. The lease may

have committed the Port Authority to provide related services, or financing

for the lessor, or any number of other benefits. The WTCP thus could have

arranged to purchase other positive terms in the lease for a higher lease

price, figuring that it had a chance to recoup the higher price if the WTCP

was damaged by a third party in tort. For instance, there might be some

other contract term that the Port Authority would be willing to provide for

$100 million and that WTCP would value at $98 million. It would be

inefficient for the parties to agree to this term. But if the WTCP believed

that there was a 3% chance that the Trade Center would be damaged or

destroyed in a tortious action, then it would have an incentive to agree to the

term because it could recoup 3% of the cost: ($100 million × 3%) + $98

million = $101 million, which is greater than $100 million. The implication

is that using market agreements between two parties to assess tort damages

upon a third party can lead to inefficient behavior by the two contracting

parties, not just the third-party tortfeasor.

These examples may seem far-fetched in the context of automobile

insurance, where most policies do not offer the car’s owner the opportunity

to specify an insured amount, or the World Trade Center, where an act of

terrorism may seem too unlikely to affect behavior.

110

But its applicability

108

Id. at 546-47 (citing and quoting Schonfeld v. Hilliard, 218 F.3d 164, 178 (2d

Cir.2000)).

109

Id. at 538-39.

110

See Saul Levmore, Self-Assessed Damages for Tort and Other Law, 68 Va. L. Rev. 771,

821-24 (1982) (suggesting that insured parties will not use insurance contracts strategically

where the likelihood of a triggering event is low).

17-June-15 MISUSE OF PATENT LICENSES 27

to patent law is much more straightforward. When a patent holder P agrees

to license a patent to a licensee L

, it must anticipate that this license will be

used to set damages in any future litigation between P and future licensees

or L

. Accordingly, P has an incentive to drive the price it sets with L

high as it possibly can.

How would P go about this? As a first step, it might engage in hard

bargaining, refusing to license the patent for a reasonable amount. The

social cost is that licensing agreements might become much less common if

patent holders refuse to license their IP for reasonable sums that

approximate expected trial outcomes. The result could be a decrease in

licensing and an increase in socially costly trials.

An alternative would be for P to package a patent license with other

valuable inducements in order to obtain a higher price. The typical patent

license includes just two terms: a royalty payment, and a license for the

patent for a period of years.

111

Yet this need not necessarily be the case. Just

as the WTCP might have obtained more favorable loan terms or any

number of other contractual benefits in exchange for a higher lease price, P

might provide subsidiary benefits—in addition to the patent license itself—

in exchange for greater royalties. P could offer to share technical

information with L

, or make available its employees to assist L

in utilizing

the patented technology,

112

or promise L

a discount on future patent

licenses, or package the patent license with a trademark license or other

intellectual property, or any number of other inducements. Even if P is

effectively “selling” the good or service for less than it is worth (if L

will

not take it for full value), the exchange is still worthwhile for P if it will

increase the royalties it might eventually receive from L

or L

P might also negotiate a license in which it absolves L

responsibility for all past and future infringement while appearing to be

selling a license only for a portion of that time period.

113

For instance,

suppose that L

has sold 1 million allegedly infringing units of a product

and intends to sell 1 million more. Imagine that P and L

agree upon a

royalty of $1 per unit. P could offer L

a blanket license for a lump sum

payment of $1.5 million, which represents a discount of $500,000 compared

with what L

might have expected to have paid. P could then structure the

license so that it only references L

’s future conduct, making it appear as if

is actually paying a royalty rate of $1.50 per unit.

111

See Simple Patent Licenses, 3 ECKSTROM'S LICENSING - FORMS § 5:1 (“There is

undoubtedly a large number of naked patent licenses regularly granted which are

unaccompanied by rights to use other forms of intellectual property.”).

112

See Lee, supra note 88, at 1505 (arguing that patent license can facilitate this type of

knowledge transfer).

113

See Keele, supra note 11, at 228 (describing a similar type of arrangement).

28 MISUSE OF PATENT LICENSES 24-Feb-15

These concerns are not hypothetical. For instance, in Ericsson v.

InterDigital,

114

a third party (Nokia) accused InterDigital of artificially

inflating the value of its patents in order to drive up the licensing price that

Nokia would be required to pay. InterDigital had agreed to license patents

to Nokia for a price based in part on what other firms would pay

InterDigital to license the same patents.

115

InterDigital then succeeded in

negotiating a lucrative license with Ericsson and demanded a substantial

payment from Nokia. Nokia, in return, accused InterDigital of artificially

inflating the value of its license with Ericsson. This example is perhaps

more acute than the typical case in which licenses are used to compute

damages because the price of Nokia’s license depended directly on the

agreement between InterDigital and Ericsson. Nonetheless, the same types

of concerns pervade both situations.

Because of these concerns, courts generally do their best to prevent

these types of arrangements from infecting license-based valuations. If a

patent license includes additional benefits—above and beyond a simple

license to the patent—courts typically refuse to treat the license as evidence

of a reasonable royalty.

116

As noted above, this is entirely appropriate in the

context of litigation damages where the only benefit “purchased” by the

defendant is a license to use the patent. P might still attempt to hide the

other terms of the deal, describing the royalty rate in one document and

leaving the other inducements for a separate document or no document at

all. If it were later discovered that a license relied upon by the court

contained other, unstated terms, it might be possible to reopen the damages

judgment based on fraud on the court. Nonetheless, it will be incumbent

upon courts and parties to remain vigilant in policing these types of

behaviors. As courts rely more and more upon licenses for measuring

reasonable royalties, patent owners will have incentives to inflate licensing

prices and then attempt to obscure or conceal that inflation by any means

available to them.

It is worth noting that the effects detailed in this section and in

section II.A push in opposite directions. Because licenses are necessarily

probabilistic calculations of expected trial outcomes—with victory for the

patent holder uncertain—they will tend to depress damages calculations at

trial. At the same time, patentees have an incentive to inflate the price of

licensing agreements, even at the expense of inefficient transactions, which

will tend to increase damages calculations at trial. It may be tempting to

conclude that these effects will balance one another out, or at least come

close enough to doing so that it is safe to ignore them. But this would be

114

418 F.3d 1217 (Fed. Cir. 2005).

115

Id. at 1220.

116

See supra note 40 and accompanying text.

17-June-15 MISUSE OF PATENT LICENSES 29

error. It would be pure fortuity if the two effects negated one another, and

there is no reason to expect that they would. Moreover, the price distortion

from the former effect—the fact that licenses represent settlement of

uncertain patent claims—will likely dwarf any distortion that patentees can

introduce by inflating license prices. If a given patent is only 50% likely to

be valid and infringed—which may if anything be an overestimate of the

patent’s probability of success—then the licensing price for the patent will

be discounted by 50%. It seems unlikely that a patent holder could inflate

the licensing price of a patent through contract to the same degree.

Nonetheless, any type of contractual manipulation is harmful both for the

mispricing it can cause, and for the social waste and rent-seeking it

generates.

IV. A WAY FORWARD?

The question that remains is whether there exists a solution to the

problems described above. Is there a mechanism by which courts can render

patent licenses a useful guide to calculating reasonable royalty damages? It

is on this issue that courts and commentators have floundered. This Part

takes up the challenge of finding such a solution.

A. The Selection of Licenses

As an initial step, courts should attempt to select those licenses that

provide the most accurate estimate of damages. The dollar value of a license

(roughly) represents the underlying value of the patent discounted by the

probability that the patent will be found invalid or not infringed at trial. If

the parties believe that there is only a 10% chance that a court will find the

patent valid and infringed, the license value will be 10% of the patent’s

underlying value—which is what the court is attempting to discover.

117

the parties believe that there is a 50% chance that a court will find the patent

valid and infringed, then the parties will agree to license the patent for 50%

of the patent’s underlying value. Accordingly, the most accurate gauge of a

patent’s value will be provided by licenses negotiated by parties who agreed

that a patent was 100% likely to be found valid and infringed. If the parties

had no doubt as to the expected outcome at trial, then they would likely

have negotiated a licensing amount approximately equivalent to d—the

expected damages at trial, and the value that the court is seeking to

discover. More generally, the greater the probability that the patent owner

would prevail at trial (per the beliefs of the parties to a licensing

117

See supra notes 57-65 and accompanying text.

30 MISUSE OF PATENT LICENSES 24-Feb-15

negotiation), the closer the value of the license to the “true” value of the

patent, and the greater the weight that license should be afforded by a court

when assessing damages.

In many cases, this means licenses negotiated as litigation

settlements will be more accurate and more useful gauges of patent value

than licenses negotiated outside of litigation. In particular, the most reliable

indicator of value will be a license negotiated in the course of a trial that the

patent owner was winning, or (better yet) winning handily.

118

The closer the

plaintiff is to being 100% certain of prevailing, the more accurate the value

of the license.

119

Courts should thus look for licenses that were negotiated

under circumstances that were highly unfavorable to the defendant. If the

defendant has received an unfavorable claim construction ruling,

120

or had

its invalidity defenses thrown out on summary judgment,

121

and elects to

settle, it is safe to assume that the plaintiff and defendant believe it is highly

probable that the patent will be found valid and infringed—and surely more

probable than they did before the trial started. A defendant who is losing at

trial will often see the writing on the wall and settle the case for close to the

patent’s full value.

122

To be clear, the point is comparative: if courts must

use existing licenses, they are better off with licenses negotiated when the

defendant was losing at trial.

Recall the issue in LaserDynamics, described in Part I: the parties

introduced twenty-nine licenses into evidence, twenty-eight of which were

for amounts of $1 million or less, and one of which—the BenQ

settlement—was for $6 million.

123

The BenQ settlement was negotiated as

an end to a trial in which the defendant had already been repeatedly

sanctioned, faced a stark disadvantage at trial, and was very likely to

lose.

124

The BenQ license was thus an extreme outlier—and of all the

licenses in evidence, it was the one that most accurately captured the value

118

For the purposes of using a license to indicate the value of a patent, it does not matter

why the patent owner is winning the case. All that matters is that the parties believe it is

very likely that the patent will be held valid and infringed. The one exception is if the

defendant is at risk of being forced to pay treble damages for willful infringement. See

Power Lift, Inc. v. Lang Tools, Inc., 227 U.S.P.Q. 435, 438 (Fed. Cir. 1985). The threat of

treble damages would distort the licensing price.

119

See id.

120

Claim construction is the process by which a court interprets or construes the claim

terms in a patent. See Markman v. Westview, 517 U.S. 370 (1996) (outlining the procedure

for courts to construe claims).

121

See 35 U.S.C. §§ 101, 102, 103, 112 (describing the various bases upon which a

defendant might argue that a patent is invalid).

122

See Posner, supra note 27, at 406 (analyzing bargaining dynamics within trial).

123

LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51, 57–58 (Fed Cir. 2012)

124

Id. at 58 (“The district court had allotted BenQ one-third less time than Mr. Kamatani

for voir dire, opening statement, and closing argument . . . .”).

17-June-15 MISUSE OF PATENT LICENSES 31

of the patent. This is precisely because BenQ was so likely to lose at trial.

Six million dollars is the amount that a defendant was willing to pay

LaserDynamics when it seemed certain that it would lose at trial and be

made to pay one way or another; $1 million (or less) was the amount that

licensees were willing to pay when there was some substantial likelihood

that they would prevail if it came to a trial. The LaserDynamics court

should have adopted exactly the opposite posture: it should have treated the

BenQ settlement as its guiding star and relegated the other twenty-eight

licenses to secondary status. It is in this respect that the courts’ approach to

using licenses to assess patent damages is not merely incoherent, but

backwards.

Of course, licenses negotiated as settlements to trial will not always

provide more accurate guides than licenses negotiated outside of trial. The

probability of prevailing at trial, even against the same patent owner, can

differ widely from licensee to licensee. This is primarily because they may

be selling different products with different probabilities of infringing the

patent.

125

There may be cases in which a license negotiated outside of

litigation provides the most accurate guide to patent value because that

licensee happens to believe it has the lowest probability of success at trial.

But in the aggregate, the licenses that provide the most accurate indications

of value will be those negotiated in the midst of trials that were going well

for the plaintiffs and poorly for the defendants.

This is why courts’ and commentators’ hostility toward litigation

settlements as a gauge of patent damages is not just misguided but

backward.

126

In refusing to consider licenses negotiated during litigation,

courts have ignored not merely a useful source of information, but in many

cases the most useful source of information. Of course, that is not to say that

licenses negotiated in litigation will necessarily be terribly useful,

particularly when the licensee was not faring poorly. These licenses still

represent only floors to a patent’s value, not accurate point estimates. But in

many cases the courts will have no better options.

B. An Estimated Multiplier

Another solution would be for the court assessing damages to apply

a multiplier to an existing license. If a court concludes that the parties to an

existing license believed there was a 25% probability that the patent was

valid and infringed, it could simply multiply the license value fourfold and

125

If the patent is invalid, it is invalid with respect to all potential infringers. Greene, supra

note 78, at 5-7. But some putative infringers might make products that almost certainly

infringe, while others might manufacture products that are highly unlikely to do so.

126

See sources cited in supra notes 46-47.

32 MISUSE OF PATENT LICENSES 24-Feb-15

assess damages in that amount.

127

This solution seems promising (and

simple) at first glance, but it runs immediately into the problem of private

information that permeates this issue. The court needs to know how the

parties to the license perceived the strength of the patent—what

probabilities did they assign to validity and infringement?

128

The only truly

reliable information on this question is in the possession of the parties to the

license themselves, and it is very unlikely that the court could ever discover

this information. In most cases what information that exists will be

protected by attorney-client privilege.

129

If a licensing party formed a belief

about the probability of invalidity or infringement, it likely did so in the

context of a communication with its lawyers. Even where the relevant

information is discoverable, there may be no written record; the court would

need to rely upon the testimony of the parties. And of course the patent

owner has no reason to testify honestly and nothing to prevent it from

artificially inflating the patent’s value.

In the alternative, a court might seek objective indications of the

probabilities that underlay an existing license. When the parties present

expert evidence on damages, their experts might include estimates of the ex

ante probability that a patent would be found valid and infringed—in other

words, the expert’s best guess as to the parties’ beliefs, at the time the

license was negotiated, of the probability that the patent owner would

prevail at trial.

130

(Or, for that matter, the court might hire its own expert or

special master to provide an independent evaluation of the same

question.

131

) In essence, the court would be constructing a miniature trial on

the merits of the prior license, attempting to reconstruct the terms of the

bargain that the parties intended. Indeed, it appears that some patent

127

See Daralyn J. Durie & Mark A. Lemley, A Structured Approach to Calculating

Reasonable Royalties, 14 L

EWIS & CLARK L. REV. 627, 643 (2010) (proposing a structured

method by which courts might calculate patent royalty damages); Taylor, supra note 11, at

131 (suggesting that “adjustments” be made to license values, without specifying what

those adjustments might be); Jarosz & Chapman, supra note 11, at 797. None of these

scholars recognizes or analyzes the advantages and problems involved in such an approach

as detailed in this section.

128

See supra Part II.A.

129

See Burlington Industries v. Exxon Corp., 65 F.R.D. 26 (D. Md. 1974) (defining and

describing the scope of the attorney-client privilege); In re Revastigmine Patent Litigation,

237 F.R.D. 69 (S.D.N.Y. 2006) (same); see also John Dragseth, Note, Coerced Waiver of

the Attorney-Client Privilege for Opinions of Counsel in Patent Litigation, 80 M

INN. L.

REV. 167 (1995) (analyzing situations in which parties can be forced to disclose otherwise

privileged information).

130

Cf. Apple, 869 F. Supp. 2d at 905 (discussing the role that damages expert play in

establishing reasonable royalty damages at trial).

131

This practice is relatively rare, but it does occur on occasion. One instance was the

Apple v. Motorola litigation before Judge Posner. See id. (discussing the court’s use of an

independent damages expert); see also F

ED. R. EVID. 706.

17-June-15 MISUSE OF PATENT LICENSES 33

damages experts have begun suggesting multipliers in their expert

reports.

132

In theory, courts could draw upon experts’ recommendations and

attempt to calculate multipliers to license values.

Yet there are (at least) four significant problems with this approach.

The first is that it involves using objective information to answer a

fundamentally subjective question. When an expert attempts to assess the

likelihood that a patent would have been held valid and infringed in a prior

litigation, the expert must endeavor to determine the parties’ perceptions of

the patent’s strength at the time the license was negotiated. But there is no

reason to believe that this expert’s guess will hit anywhere close to the

mark. The expert might discover important prior art that the licensee could

not find; or the expert might miss important prior art that the licensee

possessed.

133

Similarly, the expert might have at her disposal a set of

arguments that the prior licensee did not, or lack some legal theory that the

prior licensee viewed as critical. Using an expert to estimate a prior

licensee’s view of its prospects at trial rests on a grand assumption: the

expert will have access to the same evidence and the same legal arguments

as the prior licensee. It is of course possible that this assumption will hold in

one case or another, but there is little reason to believe that it will be

consistently true. After all, the damages expert is operating at the end of a

full trial on the merits, during which the parties have presumably produced

every significant piece of evidence and argument available.

134

The prior

license might have been negotiated well before any trial, after much less

investigation and study.

The second problem is that the prior licensing negotiation involved

private information that an expert in a later case cannot access.. The

infringement issues in the instant case might differ dramatically from the

infringement question that confronted the parties to the earlier license. The

prior licensee (L

) might have been producing a very different product from

132

However, reported cases in which experts are even permitted to testify in favor of

multipliers are few and far between. Compare Mondis Tech., Ltd. v. LG Electronics, Inc.,

No. 2:07-CV-565-TJW-CE, 2011 WL 2417367, at *7 (E.D. Tex. June 14, 2011) (allowing

an expert to testify that an existing royalty should be tripled) with Avocent Redmond Corp.

v. Rose Electronics, 2013 WL 1890007 (W.D. Wash.) (“This Court correctly prohibited Dr.

Kerr from arbitrarily trebling his proffered royalty based on a generic 33% litigation

success rate.”).

133

Prior art is any information in the public domain that predates the patented invention. A

patent can be invalidated as obvious or not novel on the basis of prior art. Robert Patrick

Merges & John Fitzgerald Duffy, Patent Law and Policy: Cases and Materials 42-49 (6

ed. 2013).

134

See Timothy Flynn, Jr., On "Borrowed Wits": A Proposed Rule for Attorney

Depositions, 93 C

OLUM. L. REV. 1956, 1962 (1993) (“The second fundamental justification

offered for the adversary model is that its incentive structure makes it the best system for

eliciting truth.”).

34 MISUSE OF PATENT LICENSES 24-Feb-15

the product that was at issue in the current litigation between the patent

owner (P) and the new putative infringer (L

). Moreover, with respect to the

prior license, L

might be in possession of critical information related to

infringement. After all, it was L

’s product that was alleged to infringe the

patent. L

may have difficult accessing this private information; L

is not a

party to the lawsuit and can be served only with certain types of

discovery.

135

And without that information, L

’s expert can only guess at

the probability that L

would have been found to infringe.

The third problem with attempting to estimate a multiplier for a

prior license is that it forces the parties to make arguments during the

damages phase of the trial that directly contradict the arguments they made

during the liability phase. The defendant will argue that the court should

apply a low multiple—perhaps a multiple of 1—to the prior license when

calculating damages. In other words, the defendant will argue that when P

licensed the patent to L

, it was a near-certainty that the patent was valid

and infringed. This follows a trial in which the defendant argued precisely

the contrary (particularly with respect to invalidity). The patent owner (P),

for its part, would argue for a high multiple, claiming that when it

negotiated with L

it was doubtful that a court would find the patent valid

and infringed. The patent owner has of course just spent the entire trial

arguing the opposite: that the patent is obviously valid and infringed.

Accordingly, both parties would find themselves trying to proffer

arguments that they are likely estopped from raising due to positions they

had taken earlier in the litigation.

136

The result would be an awkward mess

for the court. To be sure, these conflicts between the liability and damages

phases might limit the sorts of outlandish claims the parties might try to

make. But while that might be a good mechanism for reaching a moderate

result, it will not necessarily lead to an accurate one.

137

Finally, even if the parties’ experts manage to produce insightful and

accurate estimates of the licensing multiple, the court will likely misuse or

even ignore them. Consider the point at trial at which this issue will arise.

The judge and jury have just completed a trial in which the patent was

found to be valid and infringed. They are now being asked to find—

135

Compare FED. R. CIV. P. 30 (allowing depositions of nonparties) with FED. R. CIV. P.

33, 35, 36 (barring interrogatories and other types of discovery directed at nonparties) and

ED. R. CIV. P. 45 (permitting document requests directed at nonparties only with leave of

the court).

136

Teledyne, Inc. v. NLRB, 911 F.2d 1214 (6th Cir. 1990) (analyzing and describing

judicial, equitable, and collateral estoppel).

137

This particular problem could be avoided if the court simply appointed its own expert.

But then the court would lose the benefit of the adversarial process and the high-powered

incentives it creates. There is no guarantee that the court’s expert will obtain all of the most

relevant evidence or raise the most important arguments on either side.

17-June-15 MISUSE OF PATENT LICENSES 35

contrary to what they have just decided—that there was a significant ex ante

probability that they would have reached the opposite decision. This is an

implausible mental task for nearly anyone, including judges.

138

People too

often fall prey to motivated reasoning—the tendency to believe selectively

those facts and arguments that support their prior conclusions and

dispositions.

139

Any judge or juror would be very hard-pressed to give a fair

hearing to an expert who claims that their decision was less than 100%

foreordained. Inevitably, then, courts and juries will consistently

underestimate the appropriate multiplier to apply. In many cases they will

ignore the need for a multiplier entirely. It is for this reason that no court

has ever applied or even discussed the use of a multiplier in a published

opinion, even though many have discussed the use of licenses to set

reasonable royalty damages.

140

C. A Standard Multiplier

Instead of attempting to calculate a multiplier for any given license

or case, courts could instead apply a standard, constant multiplier to all

licenses across all cases. For instance, patent plaintiffs prevail in

approximately 25% of all patent cases.

141

If we assume, as a very rough cut,

that any given patent owner has a 25% chance ex ante of prevailing against

any given alleged infringer,

142

then the appropriate multiplier is four.

143

Courts have thus far appeared resistant to using a standard multiplier, but on

138

Stephen A. Siegel, The Constitution on Trial: Article III's Jury Trial Provision,

Originalism, and the Problem of Motivated Reasoning, 52 S

ANTA CLARA L. REV. 373, 455

(2012); Jon Hanson & Douglas Kysar, Taking Behavioralism Seriously: The Problem of

Market Manipulation, 74 N.Y.U.

L. REV. 633, 653 (1999); Ziva Kunda, The Case for

Motivated Reasoning, 108 P

SYCHOL. BULL. 480 (1990).

139

DANIEL KAHNEMAN, THINKING FAST AND SLOW (2014); see also Eric A. Posner & Cass

R. Sunstein, Institutional Flip-Flops (unpublished manuscript 2014),

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2553285 (discussing motivated

reasoning in the context of institutional questions).

140

See supra Part I.

141

Mark A. Lemley, The Fractioning of Patent Law, in INTELLECTUAL PROPERTY AND THE

COMMON LAW, (Shyamkrishna Balganesh, ed.) (2012).

142

This is of course an entirely heroic assumption. It is highly possible that the probability

of prevailing against a party that eventually agreed to license a patent diverges substantially

from the actual probability of prevailing at trial, due to selection effects. The point is not to

arrive at the perfect number, but just to find a rough and ready estimate that can be

deployed across cases. As the discussion will demonstrate, there are substantial problems

that accompany even this use of a rough number that go beyond any question of accuracy

of what that number should precisely be.

143

1 / 0.25 = 4.

36 MISUSE OF PATENT LICENSES 24-Feb-15

the grounds that it lacks a connection to the case at bar.

144

However, that is

precisely the advantage of such an approach. The use of a standard

multiplier would free courts and experts from the informational problems

described above. It would also eliminate the concern that judges and juries

would underestimate the appropriate multiplier due to motivated

reasoning.

145

The use of a standard multiplier would create problems, though.

Even if the standard multiplier is correct in the aggregate, in the sense that

the average license involves a patent that was 25% likely to be found valid

and infringed, it will still overcompensate and undercompensate most patent

holders. If the proper multiplier for a license is greater than four, the

standard multiplier will undercompensate the owner of that patent if it

prevails at trial. If the proper multiplier is less than four, the standard

multiplier will overcompensate the patent owner. The patent owner will be

properly compensated only in the rare case where the particular licensed

patent was exactly 25% likely to be found valid and infringed at trial. Under

normal circumstances, this type of systematic overcompensation and

undercompensation would be a problem.

146

After all, it is not the aggregate

outcome that matters. If certain types of inventors are being systematically

undercompensated, they may reduce their investments in research and

development. And if other inventors are being systematically

overcompensated, they might engage in socially wasteful expenditures in

order to acquire more patents. Wastefully high levels of resources will flow

to the types of inventions that are being overcompensated, leaving other

types of innovation underfunded.

However, in this context, the overcompensation and

undercompensation may turn out to be a feature, rather than a bug. The

reason is that the patent owners who will be undercompensated are those

who succeeded in licensing patents with a low probability of winning at

trial. For instance, if a patent owner (P) and licensee (L

) agree that there is

only a 10% chance that the patent will be found valid and infringed, they

will discount the licensing price by a factor of 10. If P eventually prevails

against a subsequent licensee (L

) at trial, the proper multiplier would be 10.

A standard multiplier of 4 would undercompensate this patent holder. By

contrast, the patent owners who will be overcompensated are those who

licensed patents with a high probability of winning at trial. If P and L

had

144

See, e.g., Avocent, 2013 WL 1890007 (prohibiting an expert from testifying in favor of

a standard multiplier).

145

See supra Part III.B.

146

Cf. Anthony Niblett, Case-by-Case Adjudication and the Path of the Law, 42 J. LEG.

STUD. 303, 304-06 (2013) (explaining, in the context of judging, that two extreme judges

will not cancel one another out but will instead likely produce extreme law).

17-June-15 MISUSE OF PATENT LICENSES 37

agreed that the patent was 75% likely to be found valid and infringed, the

appropriate multiplier (in the P v. L

trial) would be 4/3 or approximately

1.33. A multiplier of 4 would overcompensate P.

This means that parties with strong patents who assert good claims

will be overcompensated, while parties with dubious patents who assert

weak claims will be undercompensated. From a social perspective, this is

desirable. A standard multiplier will dissuade patent owners from

demanding licenses where they have only a weak claim to validity and

infringement, potentially curbing some of the worst abuses perpetrated by

so-called patent “trolls.”

147

At the same time, a standard multiplier will

reward parties with strong patents who pursue only obviously infringing

parties. This may be unnecessary, as those types of patent owners are likely

to be rewarded regardless, but it is probably not especially harmful.

148

Accordingly, at least at first glance, this approach has much to recommend

it.

Yet there is a significant problem with using a standard multiplier.

The problem lies with the circularity of using licenses to calculate trial

damages and vice versa, and the positive feedback loop that it creates.

Suppose that P is the owner of a strong patent that has never been litigated

or licensed. P demands that L

license the patent, and the parties agree that

the patent is 75% likely to be found valid and infringed. The parties further

agree that a court would likely award $10 million in damages if P prevailed.

They agree that L

will license the patent for $7.5 million.

149

Now P

approaches L

—who is selling a product similar to L

’s—and demands that

license the patent. The parties agree that the patent is 75% likely to be

found valid and infringed if the case were to go to trial. But in light of the

license between P and L

, the parties realize that if P were to prevail at trial,

the court might well award $30 million in damages—the $7.5 million

license between P and L

, adjusted upward by a multiple of 4.

150

Facing a

75% chance of incurring a damages verdict of $30 million, L

will be forced

to pay $22.5 million for a license, vastly more than L

. And then, if P were

147

See, e.g., T. Christian Landreth, The Fight Against “Patent Trolls:” Will State Law

Come to the Rescue?, 15 N.C.

J.L. & TECH. ON. 100 (2014) (describing the practice of

sending demand letters to entities that are likely not infringing in the hope of negotiating

small settlements).

148

See Anup Malani & Jonathan S. Masur, Raising the Stakes in Patent Cases, 101 GEO.

L.J. 637 (2013) (arguing that it makes economic sense to over-reward winning patent

litigants in order to compensate them for the possibility that a court might have errantly

invalidated their patents).

149

$10 million × 0.75 = $7.5 million.

150

More conservatively, the court might award damages in the amount of $20 million,

which is halfway between the $10 million figure that the court might calculate with the

help of experts, and $30 million, which the license would dictate. The effect on subsequent

licenses would be the same; the magnitude would just be slightly smaller.

38 MISUSE OF PATENT LICENSES 24-Feb-15

to demand a license from a third putative infringer L

, the upward spiral in

value will continue. The result will be vast overcompensation of P. While

some modest degree of overcompensation might not be problematic, an

uncontrolled upward spiral in the value of the patent would almost surely

lead to wasteful diversion of resources.

151

The same sort of spiral will occur, though in the downward

direction, with respect to weaker patents. Suppose P demands that L

license a patent that both parties agree is 10% likely to be found valid and

infringed at trial. The parties further agree that a court would likely award

$10 million in damages if P prevailed. L

thus agrees to license the patent

for $1 million. P then approaches a similarly situated L

, and the parties

agree that there is a 10% probability that the patent would be found valid

and infringed at trial. Now, however, L

is facing potential liability of only

$4 million if it litigates and loses at trial, because of the license between P

and L

152

Accordingly, L

will be willing to license the patent for only

$400,000. Just as the value of a strong patent will spiral upward, the value

of a weaker patent will spiral downward.

Put more formally, any time the standard multiplier M is greater

than 1 / p (the inverse of the plaintiff’s probability of success at trial), it will

create an upward spiral in value. Any time M < 1 / p, the standard multiplier

will create a downward spiral in value. Only when M happens to be chosen

perfectly—that is, M = 1 / p—will this spiral not develop.

Again, in and of itself this is not necessarily a decisive problem; it

might be appropriate to undercompensate parties who assert weaker

patents.

153

But it could create harmful incentives and lead to other types of

wasteful behavior. For instance, if P has a weak case against L

but knows

that any settlement it reaches will harm it in future cases, it might choose to

litigate rather than settle.

154

This could lead to wasteful litigation

expenditures and social costs. Alternatively, P and L

might engage in other

types of socially wasteful behavior in an attempt to obscure the value of the

license or render it inapplicable to future cases. For instance, P might

bundle the patent license with other goods that L

neither wants nor needs,

151

See Malani & Masur, supra note 148, at 652 (explaining the social harm that can be

caused by dramatic overcompensation of patent owners, even when their patents are valid

and infringed).

152

Again, the court might settle on a value somewhere between $4 million and $10 million.

Regardless, the effect will be the same; only the magnitude of the effect will differ.

153

See Malani & Masur, supra note 148, at 657 (analyzing the effects of under-

compensation on patent owners).

154

On the other hand, P faces the risk that its patent would be invalidated at trial. But if P

has a valid patent, and the weakness in its case is that L

may not be infringing, then it has

strong reasons to proceed to trial rather than allowing L

to negotiate a license.

17-June-15 MISUSE OF PATENT LICENSES 39

such as trademarks or tacit knowledge.

155

The result would be to eliminate

the license as a useful measure of patent value, using an inefficient and

socially wasteful transaction.

The upshot is that while the static overcompensation and

undercompensation caused by using a standard license multiple might be

harmless or even desirable, the dynamic overcompensation and

undercompensation that results from feedback between licensing and trial

will be harmful. This speaks to the insuperable nature of the difficulties

generated from the licensing-litigation circularity.

156

Without some

mechanism for breaking this circularity, dynamic under and

overcompensation will frustrate any attempt to use licenses as a reliable

measure of patent damages over time.

157

The prospects for finding a true

market measure of patent value do not seem promising.

158

CONCLUSION

Courts inevitably struggle to assess reasonable royalty damages, and

it is only natural that they would turn to market-based measures such as

existing licenses. However, courts’ attempts to use these licenses to

determine patent damages at trial are frustrated by three problems.

Calculating the underlying value of the patent from an existing license

requires private information that the court cannot access; doing so involves

155

See Lee, supra note 88, at 1507 (describing such transfers).

156

See supra Part II.B.

157

It is extremely common for a patent holder to approach multiple parties sequentially and

demand licenses. See, e.g., LaserDynamics, 694 F.3d at 58 (noting that there were 29

licenses on record). The dynamic problems caused by licensing-litigation circularity are

likely to exist across a broad swath of the relevant patents.

158

There are even more imaginative possibilities. One, which is suggested by recent work

by Sarah and Michael Abramowicz, would be to force parties to negotiate each element of

a settlement separately and sequentially, without comingling the separate issues. Sarah

Abramowicz & Michael Abramowicz, Severing Settlements (unpublished manuscript

2015), on file with author. In the patent context, Congress could require patent owners and

licensees to negotiate damages and probability of infringement separately. That is, instead

of the two parties simply settling on a royalty, they would first negotiate the damages

money they expect the court to assess against the defendant in the event the patent was

found valid and infringed. They would then negotiate the probability that the patent would

in fact be found valid and infringed at trial. The result would be a license that actually

contained information regarding the parties’ view of the economic value of the patent. The

problem with this arrangement is the immense incentive to cheat; both parties would

benefit from agreeing to an inflated damages figure and a reduced probability.

Accordingly, this proposal may not be easily imported to patent law. Nonetheless, similar

types of revelation mechanisms could hold promise as means of placing market values on

patents, and they are a fruitful subject for future study.

40 MISUSE OF PATENT LICENSES 24-Feb-15

a circularity that is difficult to evade; and parties to a license have

incentives to distort the value of that license in order to affect future

proceedings. There are various correctives that a court can employ,

including selecting the most information-rich licenses and applying a

multiplier to license values where appropriate. But even these measures

have limited efficacy.

It is in the nature of legal scholarship to write comedies, rather than

tragedies.

159

Each legal problem should be accompanied by a clever (and

preferably plausible) solution. But it does not seem that this story is meant

to end well. Finding an accurate measure of patent damages is critical, and

never more so than right now, as patents assume an ever more important

role in the legal landscape and damages take center stage. But existing

licenses cannot provide a useful guide to the value of a patent, only the bare

minimum of a valuation floor. Courts have no choice but to muddle through

technical analysis and expert reports per the remaining Georgia-Pacific

factors. There is no reliable substitute, and no other way to make sense of

the “true” value of a patent without creating a circularity. It is unfortunate

that courts will not be able to draw upon market indications of value, but

sometimes no guidance is better than guidance that can only lead astray.

159

To my knowledge, Tom Miles was the first to offer this observation.