on sharks, trolls, and their patent prey—unrealistic...

Research Policy 36 (2007) 134–154

On sharks, trolls, and their patent prey—Unrealistic damageawards and firms’ strategies of “being infringed”

Markus Reitzig a,∗,1, Joachim Henkel b,c,1,Christopher Heath d,1,2

a London Business School, Strategic and International Management, Sussex Place, Regent’s Park,London NW1 4SA, UK

b Dr. Theo Scholler Chair in Technology and Innovation Management, Technische Universitat Munchen,Arcisstr. 21, 80333 Munich/Germany

c Center for Economic Policy Research (CEPR), London, UKd European Patent Office, Board of Appeals, Erhardtstrasse 27, 80331 Munich, Germany

Received 1 May 2006; received in revised form 1 September 2006; accepted 1 October 2006Available online 21 December 2006

Abstract

Patent trolls (or sharks) are patent holding individuals or (often small) firms who trap R&D intensive manufacturers in patentinfringement situations in order to receive damage awards for the illegitimate use of their technology. While of great concern tomanagement, their existence and impact for both corporate decision makers and policy makers remains to be fully analyzed froman academic standpoint. In this paper we show why patent sharks can operate profitably, why they are of growing concern, howmanufacturers can forearm themselves against them, and which issues policy makers need to address. To do so, we map internationalindemnification rules with strategic rationales of small patent-holding firms and large manufacturers within a theoretical model.
Our central finding is that the courts’ unrealistic consideration of the trade-offs faced by inadvertent infringers is a central conditionfor sharks to operate profitably.© 2006 Elsevier B.V. All rights reserved.
JEL classification: M00; M11; M21; K00; K11; K33

K ent

I4

HC

E

r

0

eywords: Patent; Patent shark; Patent troll; Damage award; Infringem

∗ Corresponding author at: London Business School, Strategic andnternational Management, Sussex Place, Regent’s Park, London NW1SA, UK. Tel.: +44 20 7000 8714; fax: +44 20 7000 8701.

E-mail addresses: [email protected] (M. Reitzig),[email protected] (J. Henkel), [email protected],
[email protected] (C. Heath).1 Member of the RIPE Research Network for Intellectual Propertyconomics.2 The opinions expressed by Christopher Heath do not necessarily
epresent the official opinion of the European Patent Office.

048-7333/$ – see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.respol.2006.10.003

1. Introduction

“The operations of patent sharks sometimes compelan inventor to obtain patents for articles which arenever meant to be placed on the market. A fellowoften gets up a machine, and somebody else comes
along, and by getting patents through for certain parts,can give the inventor a great deal of bother and makehim pay well, even if the inventor gets control of it ”(Thomas Edison, 1898).
mailto:[email protected]




dx.doi.org/10.1016/j.respol.2006.10.003

rch Pol

slegwtoAeiitSe1hptlh

hvo

sfsotfcOlac

iait$DctdwttcGtim

M. Reitzig et al. / Resea

“Noblesse oblige”, but property does not; legallypeaking, at least, this is the case when it comes to intel-ectual property. Patent holders are – apart from very rarexceptions2 – not obliged to engage in the production ofoods using their protected technology. They may dohatever they want to with their inventions, and often

hey will consider it most profitable to sell their technol-gy or license it against a royalty fee to a third party (seerora et al., 2001; Arora and Ceccagnoli, 2006). The law

xplicitly supports this form of exploitation, offering var-ous remedies to patent holders whose rights are beingnfringed—no matter whether the patentee uses its pro-ected technology or not (see Lanjouw and Lerner, 2001;chankerman and Scotchmer, 2001). In the ideal worldnvisaged by the forefathers of patent law (see Nordhaus,969), these remedies should “do justice” to the patentolder and restore his/her incentives to invent in the firstlace (see also Macdonald, 2004). They should set incen-ives for patent holders and other parties to enter sales oricensing negotiations from the outset so that no inventoras to fear any infringer at all.3

The above assumes a “credulous” patent holder who
as to fear a deliberate infringer. But what if, con-ersely, R&D-intensive firms start to fear the existencef patent-holding individuals who have ulterior motives
2 These very rare exceptions come in the form of so-called “compul-ory licences”. According to Article 5 A. (2) of the Paris Conventionor the Protection of Intellectual Property, “Each country of the Unionhall have the right to take legislative measures providing for the grantf compulsory licenses to prevent the abuses which might result fromhe exercise of the exclusive rights conferred by the patent, for example,ailure to work”. If at all, such legislative measures are almost only dis-ussed in the context of military or major pharmaceutical inventions.ne recent example includes the discussion of granting compulsory

icenses for HIV-related drugs to generic manufacturers in South Africand Brazil in order to provide affordable medication to these poorerountries’ people.3 It is on this background that one has to understand why even small

ndividual inventors are at times entitled to significant damage awards,nd why this can be desired from a social standpoint (this is less clearn the example illustrated in Footnote 4). An illustrative example ishe case of Gaus. versus Conair (see “Jury blows away Conair with28.5 M infringement award”, Litigation Week, 11 February 2002)r. Gaus received a 28.5 million US$ compensation from the Conair

orporation for the infringement of his patents on circuits used to pro-ect users of hand-held hair dryers from being electrocuted when theryers are immersed in water. Conair knew of Dr. Gaus’ rights andillfully infringed them (Note: the actual sum awarded to Dr. Gaus in

his case is likely a higher price than Conair would have been willingo pay for the technology had they acted legitimately. As will becomelearer in Section 2 of this paper, Conair’s willingness to pay for Dr.aus’ patent before infringement should have amounted to at least a

hird of the awarded damages, however. The example only serves tollustrate why seemingly high remedies awarded to individual inventors

ay still be economically suitable at a second glance).

icy 36 (2007) 134–154 135

of free-riding on a product’s core invention? As theintroductory quotation shows, the great Thomas Edisonrecognized already a hundred years ago that a manu-facturer who does not hold the rights to each and everyinvention embodied in its product may face harassmentfrom such individuals.

What nobody could have foreseen then, however, andwhat has consequently not been discussed comprehen-sively in the literature on “systematic patent strategy”(see Blind et al., 2006; Macdonald, 2004),4 are the evengreater concerns of today’s leading R&D multinationalsof potentially overlooking these (often small) inven-tors’ patents and being caught in the trap of inadvertentinfringement. Today’s patent “sharks” or “trolls”5 seemto place their bets on precisely this corporate “negli-gence” or monitoring deficiency. Sharks or trolls haveno intention of engaging in the production of the tech-nology underlying their patents, but instead make moneyfrom royalty payments they obtain directly from theirlicensees or indirectly in terms of damage awards. Dif-ferent from “classic” licensors, however, trolls intend totake their victims by “surprise” to facilitate their attempts
to force manufacturers into unexpected licensing fees.The relevance this topic has assumed over time is dra-matic, being reflected in a series of recent disputes6 and
4 See Blind et al. (2006) and Macdonald (2004) for two carefullycompiled and recent surveys on the strategic use of patents. As botharticles show, the “strategic use” of patents (the two most importanttypes being blocking and cross-licensing with patent ‘thickets’ playinga major role for the latter), has classically been discussed from the per-spective of those patent holders who either engage in the productionof their own technological goods or consider themselves professionalintellectual property suppliers who repeatedly interact with manufac-turers. The strategy of “being infringed” as pursued by trolls is adifferent phenomenon. While not entirely new (see Section 4.1, quotingLemelson versus Mattel), the broad systematic use of troll strategiesappears to be very recent, however.

5 The term ‘patent troll’ was originally used by an Intel executive inthe early 1980s to denote patent-holding individuals who use patents“that they are not practicing and have no intention of practicing andin most cases never practiced” (Peter Detkin, former assistant generalcounsel at Intel, quoted in Trolling for Dollars, The Recorder, 30 July2001). While today there is a consensus that trolls make it a businessto press other firms for licensing fees, the question of how to definea “patent troll” properly is not a trivial one, and it was only recentlyaddressed systematically by lawmakers (see “Patent Trolls: Fact orFicition”, Hearing before the Subcommittee on Courts, the Internet,and Intellectual Property of the Committee of the Judiciary, House ofRepresentatives One Hundred Ninth Congress, 15 June 2006). For thepurpose of this paper, we therefore present a more elaborate workingdefinition of the term in Section 2.

6 For example, Luxembourg-based InPro Licensing SARL hassued RIM, the maker of the blackberry handheld device, forinfringement. In the suits brought against RIM in 2003 in Ger-many and the UK, rulings by the German Federal Patent Court

rch Pol
136 M. Reitzig et al. / Resea
summarized most impressively in the following manage-ment statement by Peter Halkjaer, Senior IP Manager,Mobile Phones at Nokia:

“From an IP management perspective, patent sharkscurrently pose one of the great challenges to our firm”.

Without doubt, sharks create uncertainty for innova-tors and their activities may lead to damage awards whichare a multiple of what the shark’s victim, as legitimatelicensee, would have been willing to pay ex ante. Hence,as we will show in more detail in this article, the sharkbusiness entails various economic inefficiencies. This isever more puzzling because, while most of the publiclyknown shark cases took or take place in the US, the phe-nomenon is of great and growing concern also in otherparts of the world.7

Thus, why do trolls exist at all? How can manu-facturers forearm themselves against them? And whatare the policy consequences? Despite the richness ofprior contributions on economics of patent infringement(for example by Schankerman and Scotchmer, 2001;Crampes and Langinier, 2002; Bessen and Meurer, inpress) and managerial aspects of patent licensing (e.g.
Agrawal and Garlappi, 2002; Arora et al., 2001; Aroraand Ceccagnoli, 2006) the literature does not seem toprovide conclusive answers to these pressing questions.
and the English High Court, early 2006, declared the dis-puted patent invalid (http://www.theregister.co.uk, 30 January 2006,http://www.reghardware.co.uk, 02 February 2006). While the deci-sion in Germany is not yet final, the rulings confirm the generalobservation that troll patents are often of low quality. More promi-nent still is the legal battle between RIM and NTP, which wasat times even threatening to shut down RIM’s operations in theU.S. and issued a settlement for US$ 612.5 in March 2006(news.com.com/2061-10801 3-6045871.html). Yet another example isForgent Networks Inc., which has been suing various large companiesfor the alleged infringement of a patent (US patent No. 4,689,672)that, Forgent claims, covers parts of the JPEG image compressionstandard (http://www.eweek.com, 22 April 2004). By April 2005,Forgent had received more than US$ 100 million in licensing feesfrom 35 companies, and is suing 44 companies for infringement(http://www.theregister.co.uk. 25 April 2005). The patent had beengranted in 1987 and had not been used for years, until Forgent startedasserting it in 2002.

7 The court cases brought by InPro against RIM (mentioned in thepreceding footnote) took place in Germany and Britain, respectively.Another prominent example of a troll-like firm is Germany-based TelesAG. Since 2002, Teles has sued at least 10 firms in German courts forinfringement of patents related to communication technology, amongthem Cisco, Nokia, and Deutsche Telekom (http://www.teles.de). Inaddition, interviews conducted by the authors suggest that the Euro-pean troll cases reported in the press may only be the tip of the iceberg,since firms attacked by trolls often prefer to settle the dispute in adiscreet manner.

icy 36 (2007) 134–154

As we will show in this paper, the simple yet unsat-isfactory answer to the core question of “why patentsharks exist” (from which the other aforementionedquestions derive) is: because their activity is not onlyprofitable but also perfectly legal. As we will elabo-rate upon, the threat modern patent sharks can poseto innovative manufacturing firms strongly depends –among other factors, notably surprise – on the applica-ble law governing infringement as well as its practice bythe courts. Surprisingly from an economic standpoint,damage awards may not only be calculated followingdifferent rationales within one jurisdiction, but it lies(to a large extent) within the discretion of the patentholder (and not the court!) to pick the type of remedyhe/she prefers (namely “lost profits,” “infringer’s profits”(unjust enrichment), and “reasonable royalty rates”). Thereal problem occurs, however, as the courts’ interpreta-tion of these damage awards regulations in some casesrenders “being infringed” a more profitable option thanex ante negotiation between the patent holder and thepotential infringer—eventually opening the floodgatesfor the “troll business”.

In more detail, in this paper we pick up on the “patentshark” phenomenon and examine it from a theoreticalperspective, encompassing legal, managerial, and eco-nomic aspects. To understand this phenomenon, we takea game-theoretical approach to modeling the behaviorof trolls and their victims. Here, we primarily sketch thedecisions sharks may expect both their prey and courtsto make within a complex technological and legal world,and focus less on the activities of the sharks themselves.This paper is dedicated to analyzing:

(a) which (managerial) incentives exist for firms to beinfringed, and how legal rules governing damageaward calculations affect these incentives,

(b) why the economic importance of the phenomenon ofbeing infringed (i.e. acting as a shark) has probablyincreased over time,

(c) which actions appear appropriate to be taken byfirms that are (potentially) threatened by sharks, aswell as

(d) which discussions are required from a policy per-spective in the light of the current business practice.

In line with the aforementioned questions, the paper’sfour-fold thrust is as follows:

(a) small firms, and in particular non-producing firms,have incentives to be infringed as they may beawarded remedies by the courts that are system-atically higher than they could have obtained in

http://www.theregister.co.uk/

http://www.reghardware.co.uk/

http://www.eweek.com/

http://www.theregister.co.uk/

http://www.teles.de/

rch Pol

(

(

Siialrtpainrtpaib5

2f

ltSooi(i

(typically) means that it orders the infringer to refrainfrom producing and/or selling the infringing goods (see,e.g., Lanjouw and Lerner, 2001, on the managerialimplications).10 As for damages, most jurisdictions pro-

8 To the best of our knowledge this definition should, for the mostpart, be in concordance with the attempts to define “patent trolls” duringthe House of Representative Subcommittee Hearings “Patent Trolls:Fact or Fiction” on 15 June 2006 (see above). See for example Mr.Edward Reines’ definition on page 8 of the Hearing’s report (serialnos. 109–104).

9 In order to show the universal dimension of the problem, we willsummarize the relevant jurisdictions of five of the major industrial


licensing agreements with large patent holdersbefore infringement;

b) increasing complexity of some technology fields,the increasing number of patents worldwide, andthe resulting difficulty in monitoring the existingstate-of-the-art technology, as well as the increasingfirm sizes of large patent holding and manufacturingcorporations should – in accordance with observa-tions from the real world – lead to an increase in theimportance of the “shark business”;

(c) large patent-holding and manufacturing firms arewell-advised to spend extensive resources on ensur-ing access rights to technological substitutes of theircore inventions as well as complementary techno-logical assets, to allocate more money to technologymonitoring, and to lobby for legislative changes;

d) courts need to reflect upon their interpretation ofexisting legal regulations and work towards a trulywelfare-maximizing patent indemnification rule.

The remainder of the paper is structured as follows:ection 2 provides the basis for understanding why

nnovators can create a profitable business from hav-ng their patents systematically infringed. We presentsynopsis of international patent indemnification regu-

ations and show how they should appear in the strategicationales for different types of innovating firms. Sec-ion 3 picks up the issue in a formalized fashion andresents a formal theoretical model of technology choicend patent infringement. Here, we juxtapose outcomesn patent litigation cases according to existing indem-ification regulations with alternative outcomes fromealistic ex ante technology sales or licensing negotia-ions (avoiding infringement). We can show that hidingatent-protected technology to be infringed emerges asdominant strategy for low-tech capacity-constrained

nnovators (“trolls”). Section 4 discusses the results fromoth a managerial and a policy perspective, and Sectionconcludes and provides an outlook on future research.

. Theoretical considerations—institutionalrame and managerial rationales

In this section we will elaborate on the theoreticalegal and managerial considerations required to explainhe existence of the “troll business” through our model inection 3. To start with, we define the term “patent shark”r “patent troll” in more detail. We denote patent sharks
r trolls as individuals or firms that seek to generate prof-ts mainly or exclusively from licensing or selling theiroften simplistic) patented technology to a manufactur-ng firm that, at the point in time when fees are claimed,
icy 36 (2007) 134–154 137

already infringes on the shark’s patent and is thereforeunder particular pressure to reach an agreement with theshark.8

In order to illustrate the prerequisites for beinginfringed becoming a profitable and legitimate strategy(see next section), this section is split into two parts.The first part contains a summary of the necessary legalinformation regarding international9 patent indemnifica-tion rules in order to understand the damage award threatpoints that the law sets down for potential infringers.Eventually, as we will show, these threat points affectthe choice to become a shark or not. We provide botha picture of the current legal regulations (Section 2.1.1)and a brief sketch of the most important changes that areunderway (Section 2.1.2).

In the second part, we will show how these indemni-fication rules come to bear on strategic decision making.Here, we will distinguish between two distinctly differ-ent types of firms, using the capacity to manufacture thetechnological goods under consideration as the differen-tiating trait. Namely, these are (a) large firms with thecapacity to supply the entire market with their techno-logical goods, and (b) small firms lacking productioncapacity altogether. The stage set in this section willform the basis for the formal analysis in Section 3. Here,we will map the legal regulations and the managerialrationales, showing in particular how the indemnificationrules create incentives for being infringed.

2.1. Patent infringement indemnification

2.1.1. A brief sketch of existing regulationsThere are two types of complementary remedies

against patent infringement: injunctive relief and dam-ages. When a court grants injunctive relief, this

nations worldwide. Moreover, we include the Netherlands in our sur-vey, since Dutch courts have traditionally played an important role inthe cross-border patent litigation jurisdiction.10 Often injunctive relief is considered to be the more important rem-

edy from the perspective of the infringer as it creates immediate and

rch Pol
vide for at least two and sometimes up to three “standardmethods” for indemnification assessment. Namely, theserefer to the calculation of (a) lost profits, (b) an ordinarylicensing fee (reasonable royalty), and (c) infringers’profits (unjust enrichment). In the following, we brieflypresent the biggest “generic” international denominatorfor each of these damage award rules (for a more detailedelaboration on the international differences see Heath etal., 2005).

2.1.1.1. Lost profits. Here, the patentee shall be rein-stated in a position where he/she would have beenbut for the infringement, with the restriction that onlylosses from the patentee’s own production are takeninto account, not, e.g., from licensing. Note that thisrestriction marks an important discrepancy between theeconomic and the dogmatic (legal) notion of lost profits(see below for more details). The calculation method isaccepted by all major jurisdictions (US: 35 USC Section284; Japan: Section 102(1) Patent Act; Germany: Sec-tion 139 Patent Act; UK: Section 59 Patents Act; France:Art. L615-l(2) Intellectual Property Code). The leadingUS case required the patentee to show the following:11

(1) demand for the patented product (as indicated bypast sales);

(2) absence of competing and non-infringing products(see below);

(3) ability of the patent owner to actually marketthe quantity of goods12 for which lost profits are

13
claimed ;(4) the amount of profit that would have been made in
the absence of infringement.14

significant ‘losses’ for them. Hence, the literature in the field focused oninjunctions for a long time. The U.S. Supreme Court’s recent decisionin eBay Inc. v. MercExchange, L.L.C., however, has tilted the balancetowards damages. The Court unanimously determined that the findingof patent infringement should not automatically imply the issuanceof an injunction (http://www.supremecourtus.gov/opinions/05pdf/05-130.pdf). This case is in line with the envisaged changes to U.S. patentlaw. We thank one of our referees for sharing this thought with us.11 Panduit Corp. v. Stahlin Brothers Fiberworks. April 25, 1978, 575

F.2d 1152 (6th Circuit 1978).12 The existing law generally accepts that in the absence of marketing

capacity, the patentee cannot claim lost profits due to a lack of causality.13 A requirement that is also specifically mentioned in Section 102(1)

Japanese Patent Act and that has been applied in the UK decisionCatnic Components v. Hill & Smith [2]. English High Court. 16 March1983 [1983] F.S.R. 512. The German courts also require demand forthe product and actual production capacity: German Federal SupremeCourt, 10 July 1979, GRUR 1979, 869–872.14 Only Japanese (and Korean) patent law differs in this respect: Sec-

tion 102(1) Japanese Patent Act allows the patentee to calculate his

icy 36 (2007) 134–154

Where competing and non-infringing products are onthe market, element (2) above requires a so-called marketshare analysis and an award based on a pro rata per-centage of the infringer’s sales.15 Lost profits cannot beawarded where the infringing products do not qualify asa substitute for the ones of the patentee.16

2.1.1.2. Ordinary licensing fee. The most commonform of claiming damages is the ordinary licensing fee(or “reasonable royalty”) for three reasons. First, it is theform of indemnification where plaintiff and defendantcan bilaterally agree on the size of the reward. Second, incontrast to the case where the plaintiff files for lost prof-its or infringer’s profits, relatively little effort has to beexpended by the rights-owner to prove his case. Finally,many patent owners do not wish to lay open their internalcost structures (which they would have to when filing forlost profits, but not in the case of an ordinary license fee).

It is standard practice to calculate a reasonable roy-alty “on the basis of what royalty a willing licenseewould have been prepared to pay and a willing licensorto accept.”17 Two aspects appear particularly notewor-thy. Despite its theoretical ex ante focus on what thepatentee and the infringer would have agreed upon beforeinfringement, the rule is interpreted with ex post knowl-edge and typically simplified in its application.18 In thecase of an innocent infringer, this means that his hypo-thetical non-infringing options in the case of completeinformation (e.g., inventing around) will not be taken intoconsideration. On the other hand, a deliberate infringerwill not be made worse off than an ordinary licensee
by this type of indemnification; as a matter of fact,sometimes the deliberate infringement might be moreprofitable than ex ante licensing: until about 1998, it wasstandard practice in Japan to use royalty rates calculated
damages by multiplying the number of infringing products sold by theinfringer by the profit the patentee would ordinarily realize when sell-ing his own products. Such a calculation method has been explicitlyrejected by the UK decision Gerber Garment Technology v. LectraSystems. Patents Court, 20 March 1995 [1995] R.P.C. 383, and theGerman decision, Federal Supreme Court, 6 March 1980, GRUR 1980,844—“Tolbudamid”: “Uncertainty that one does not know if the defen-dant would have been able to achieve the same turnover in infringingproducts at higher prices.”15 For example, US decision State Industries Inc. v. More-Flo Indus-

tries Inc., 883 F.2d 1573 (Fed. Cir. 1989); UK decision CatnicComponents v. Hill & Smith [2]. English High Court, 16 March 1983,[1983] F.S.R. 512.16 US decision Bic Leisure Products v. Windsurfing International, 4

August 1993, lF.3d 1214 (Fed. Cir. 1993).17 UK decision Catnic Components v. Hill & Smith (see Footnote

13).18 Casucci (2000, 692/702).

http://www.supremecourtus.gov/opinions/05pdf/05-130.pdf

rch Pol

bbTfA

2jptaiAluwsosinntpcSfieht

motj

JKR

R

DIte2

b((t


y across-the-board industrial averages of royalty ratesetween Japanese companies for domestic patents.19

his changed once the word “ordinary” was deletedrom the wording of Section 102(2) Japanese Patentct.

.1.1.3. Infringers’ profits (unjust enrichment). Someurisdictions allow the patentee to recover the infringer’srofits as one way of calculating damages. In Japan,his remedy is limited to cases where the patentee hasctually used the patent.20 In the UK, the claim for thenfringer’s profits is statute based (Section 60 UK Patentct: “account of profits”), and in Germany based on the

egal fiction that in using another’s patent, the infringerndertook a business on behalf of the rights-owner, whoould thus be entitled to obtain all profits made from

uch business.21 Both jurisdictions allow fairly gener-us deductions where the infringer has used his ownkill, labor and expenses in the marketing of the infring-ng products.22 Granting “infringers’ profits” is formallyot allowed in France and the US. However, whether orot the US term “unjust enrichment” reflects a remedyhat essentially corresponds to the notion of infringers’rofits remains arguable from a dogmatic standpoint. Inoncordance with earlier works (see Schankerman andcotchmer, 2001) and for the purpose of this simpli-ed legal analysis, we subscribe to the view that unjustnrichment is a part of infringers’ profits and we willence treat the two terms synonymously for the rest ofhis article.

Table 1 recalls in which of the countries plaintiffs
ay choose among several calculation methods. More-
ver it summarizes the calculation methods and howhey are currently applied in some of the major patenturisdictions.23 In addition to showing the subtle differ-

19 Such statistical averages were taken from Hatsumei Kyokai (Ed.),isshi ryoritsu (Use and Compensation) (Tokyo, 1980); Hatsumeiyokai (Ed.), Gijutsu torihiki to royalty (Technology Transfer andoyalties) (Tokyo, 1992).

20 Osaka District Court, 27 March 1980.21 For example, German Imperial Supreme Court, 22 October 1930,GZ 130, 108.

22 For the UK, Gerber v. Lectra (see Footnote 4); for Germany,usseldorf District Court, 25 July 1996, 4 O 217/95—“Winkelprofil

II.” However, according to the German Federal Supreme Court,he infringer cannot deduct costs that relate to general managementxpenses: German Federal Supreme Court, 2 November 2000, GRUR001, 329—“Gemeinkostenanteil.”23 More detailed information on the individual country legislation cane found in the following references. Maloney (2000) for the US, Heath2000) for Japan, Marshall (2000) for Germany, Cornish and Llewelyn2000) for the UK, Petit (2000) for France, and Brinkhof (2000) forhe Netherlands.

icy 36 (2007) 134–154 139

ences between the countries, however, it also illustratesthat the treatment of the different norms – namely lostprofits, infringers’ profits, and reasonable royalties –is, whenever applicable, internationally comparable toa large extent. This stresses the global importance of thephenomenon we analyze, which is also partly reflectedin upcoming harmonization of international laws (seeSection 2.2).

2.1.2. (Potential) changes underwayWhile Section 2.1 describes the current legal situation

with respect to patent indemnification regulations in thesix focal jurisdictions, we find it useful to highlight afew legislative changes that are either currently beingdiscussed (in the U.S.) or have been passed at the supra-national (European) level and now await conversion intonational law. We focus only on a very few selectedaspects, as a complete synopsis would be beyond thescope of this paper.

In the context of the US patent reform, the follow-ing amendments are currently under consideration. First,particularly with respect to the surge of patent trolls, aclaim for treble damages in the case of willful infringe-ment shall be limited to cases where the infringer hasbeen furnished with a direct allegation of infringement.24

Second, in order to allow competitors to switch to non-infringing alternatives in a timely fashion, it has beensuggested that publication of all US patent applica-tions after 18 months, no matter whether the filing ofrelated applications abroad is intended or not, should bea requirement.25

In Europe, the “Directive 2004/48/EC of the Euro-pean Parliament and of the Council on the enforcementof intellectual property rights” was passed on 29 April2004. Among others, it foresees the harmonization ofpatent indemnification laws in the member countries ofthe European Union in that the three generic damageaward calculations (see above) shall be applicable in allmember states. In theory, the directive will also allowcourts to apply the different damage award calculations(e.g. lost profits and ordinary licensing fees) simultane-ously in one case in order to assess “realistic” damages.While passed in 2004 and required to be implemented
by 31 May 2006 into national law, the directive is still tounfold its relevance in the future, as most member stateshave not yet implemented it.
24 See “Patent Trolls: Fact or Fiction”, Hearing before the Sub-committee on Courts, the Internet, and Intellectual Property of theCommittee of the Judiciary, House of Representatives One HundredNinth Congress, 15 June 2006, serial no. 109–105, p. 9.25 Ibid, pp. 10–11.

140 M. Reitzig et al. / Research Policy 36 (2007) 134–154

Table 1Indemnification regulations within and across countries—an international comparison

Lost profits Licensing fee Infringer’s profits Choice for plaintiff

U.S. 35 USC Section 284.Requirements: (1)demand; (2) marketingcapacity; (3) absence ofcompetition,non-infringing substitutes

Fall-back provision wherelost profits cannot be orare not claimed

No Yes

Japan Section 102(1) PatentAct: multiplication ofinfringer’s turnover withprofits the patentee wouldhave made for such anumber of products.Marketing capacity ofpatentee must be proven

Section 102(3) PatentAct: fall-back provision;estimate of royalty rate

Section 102(2) PatentAct. Not applicable wherepatent was not used bypatentee

Yes

Germany Section 249 Civil Code:restitution of the statusquo ante. Limitation byproduction capacity andproof that infringingproduct could act as asubstitute

Most common form ofcalculation, normallyagreed upon in courtsettlement. No“infringer’s surcharge”can be claimed except forcopyright matter (doubleroyalty)

Based on the legal fictionthat infringer undertakes abusiness allocated to thepatentee. Deduction ofinfringer’s expenses.Infringer’s marketingefforts taken into account

Yes: claim for inspectionof infringer’s accountsallowed prior to choiceof calculation base

UK Yes, likelihood of havingmade the infringer’s sales,deduction of infringer’sefforts to commercialize

Yes, a notional royalty asthe minimum of lostprofits

Yes, but rarely requested Yes, after review of thedefendant’s commercialdocuments

France Only if patent is used;calculated by amount ofcounterfeit products, lossof turnover (determinedinter alia by the quality ofthe patent) and amount oflost profits. Market shareof patentee considered

Where the invention is notused. Infringer’s turnovermultiplied by anappropriate royalty rate

No, clarified in Patent Act1968

If patent is actuallyused: Yes

The Netherlands Same as Germany.Section 42(2) Patent Act1910, Section 70(3)

Regarded as the minimumthat can be claimed as lostprofits

Section 43(3) Patent Act1910; Section 70(4)Patent Act 1995: the

Yes, after inspection ofdocuments

Patent Act 1995

2.2. The managerial perspective—innovationexploitation strategies as a function of productioncapacity and other factors

In order to convey the central thought of our paperclearly we need to describe the situation that we analyzein more detail. In doing so, we are setting the stage for theformal model (Section 3), which will pick up the centralcharacteristics of our scenario.

One central assumption for the rest of the paper is
that the technological goods are complex (see Mergesand Nelson, 1990; Kash and Kingston, 2001); in otherwords there are various patentable inventions in oneproduct. This assumption is fulfilled rather well (Cohen
infringer should not beallowed to keep his profits

et al., 2000; Reitzig, 2004) in industries such as software,telecommunications, and consumer electronics—forwhich our analysis probably shows more relevance thanfor other industries (such as chemicals or textiles). Inthis situation, the different patentable inventions enter-ing the product are technological complements. For thesake of clarity we assume that one or a few inventionsof a product can be considered “core” inventions, whichresulted from technologically sophisticated research andare difficult to substitute. The design of a product is
centered on them and not on their complements (whichmay be acquired or created in-house). In line with theaforementioned thoughts we assume that the difficultyof (legally) inventing around a patent decreases with

rch Pol

d(aiwed

iipce

2i

sfiutdcliwficm(edbc

p(rltpehttm

wcpu

role for IP vendors and high-tech start-ups since nei-


ecreasing technological sophistication of the patentGallini, 1992). Additionally, we assume that all firmsre somewhat constrained (see also below) in their mon-toring resources for existing technology. Moreover,e presume that decision makers act rationally to the

xtent that they have the necessary information at theirisposal.

One of the major determinants in the choice of annnovation exploitation strategy is production capac-ty. To illustrate the importance of this determinant, weick out the following two stylized extremes along theapacity spectrum and show the different innovationxploitation rationales for these firms in our setting.

.2.1. Production capacity-unconstrainednnovative firms

At one end of this continuum we consider R&D inten-ive and production capacity-unconstrained firms, i.e.rms that enjoy the (theoretical) possibility to fully sat-rate the market under consideration with goods fromheir own production. Unless such a firm’s costs of pro-uction clearly exceed those of a competitor that isapacity-unconstrained as well,26 our focal firm will beikely to pursue a strategy in which it seeks to exploit itsnnovations by selling self-produced goods (consistentith the fundamentals of the resource based view of therm). To do so, the firm needs to ensure access to theore technologies of the product as well as the comple-entary ones required to produce it. Following Teece’s

1986) logic, to pursue this type of exploitation strat-gy the firm needs to dedicate extensive resources to theevelopment of the “core” components (that are proba-ly not for sale and for which technological alternativesannot easily be developed).

The rational management of such a firm will incor-orate considerations regarding both passive and activethough possibly inadvertent) patent infringement in itsationale. Depending on the set of applicable legal regu-ations in a particular case of passive infringement (i.e.he focal firm’s patent is illegitimately used by a thirdarty), the firm’s management may pick the type of rem-dy that suits the firm best—that is, the one yielding theighest payoff. Moreover, the firm will try to incorporate
he chance of actively infringing a third party. Assessinghe importance of this eventuality, however, will be far
ore difficult for the focal firm, at least if the infringe-

26 Note: strictly speaking it does not have to be only one competitorho is capacity-unconstrained; it could also be a group of individually

apacity-constrained competitors operating at (overall) lower costs ofroduction. Transaction costs and economies of scale render this verynlikely, however.

icy 36 (2007) 134–154 141

ment is inadvertent (i.e. the firm has no intent to infringe).This is for two reasons. First, the payable amount to the(infringed) patent holder will depend on this third patentholder’s own innovation exploitation strategy, which isin turn affected by the options the law provides. Second,however, the pure chance for the focal firm to activelyinfringe a third party is also determined by the innova-tion exploitation strategy pursued by the third party aswill become clearer in the following.

2.2.2. Production capacity-constrained patentholders

At the other end of the capacity continuum weconsider patent holders (small firms or individual inven-tors) who do not possess any capacity of their ownto produce technological goods, and may also lackother complementary assets (in particular, complement-ing technologies). These firms will differ from each otherwith respect to the technological sophistication of theirinventions.

Among production capacity-constrained innovatorsthat are truly innovative (i.e., ones that generate poten-tial core inventions), the two most important types arededicated R&D firms and high-tech (university) start-ups (see Lowe and Ziedonis, 2006; Mowery et al., 2001).Dedicated R&D firms that engage in repeated sales andlicensing negotiations with manufacturers will likelyinvent high quality components (both core and com-plementary components) that are of real value to thepurchasing or in-licensing firms.27 Next to these spe-cialized R&D/IP vendors, high-technology (university)start-ups will generate sophisticated R&D of a kind thatmay become a “core” component of a product, eventhough they are unlikely to engage in repeated interac-tion with a manufacturer. Both of the aforementionedfirm types seek to sell their intellectual property to firmsthat do have access to production capacity,28 and bothwill consider active and passive patent infringement intheir innovation rationales. However, in contrast to thecapacity-unconstrained innovators, passive infringementis likely to be a greater managerial concern than activeinfringement. Active infringement plays an insignificant

ther intends to produce a product by itself. On the otherhand, the firms’ profit maximizing rationales require

27 An example is the UK-based IP vendor ARM, which sells designsfor semiconductors.28 The biotechnology sector provides a wide series of examples. The

business plans of so-called dedicated biotechnology firms (DBFs) buildon the intent to be taken over by a large manufacturing pharmaceuticalproducer.

rch Pol

a short-cut in order to concentrate on the core of thispaper’s contribution (the systematic effects that render


safeguarding their inventions against manufacturers whocould – if there were no legal remedies – oust their“innovative suppliers”. Again depending on the set ofapplicable legal regulations, in a case of passive infringe-ment the specialized IP vendors and the high-techstart-ups will be able to pick the remedy that maximizestheir individual profits. As will become clearer duringour discussion, the incentives that patent indemnificationrules create for these firms do not render being infringeda dominant strategy.

Finally, however, there also exist firms holdingpatents to minor technological solutions that can serveas technological complements to a core invention. Thesepatents may result from the firms’ own (technically unso-phisticated) research, or may have been bought (fordetails see Section 4.3). Some of them concentrate their“R&D” efforts on particular minor complementary tech-nological components to a core invention held by amanufacturing firm. Oftentimes, the technological sub-tlety of these firms’ inventions is considered marginaland they could easily be substituted by alternative tech-nological solutions so that ex ante licensing negotiationswith manufacturers will most likely lead to negligibleprofits for these innovators. Their activities can only beunderstood when incorporating the passive infringementrationale into their strategy. As a matter of fact, thesefirms hope to be infringed and do everything they canto keep their patent-protected technology as invisible aspossible until it is illegitimately used by a manufacturer.Such unpleasant surprises for the manufacturer can occurfor three reasons: the patent was a “submarine”,29 themanufacturer expected that the patent would never beused in litigation against him/her, or the patent slippedthe manufacturer’s attention despite monitoring. While
submarine patents are restricted to the U.S., both otherreasons are relevant globally (we will elaborate on theseissues in Section 4.2).
29 So-called “submarine patents” are possible due to the policy of theUSPTO to disclose patent applications only at the time when the patentis granted. This policy was generally in force until 1999, and since thenhas been restricted to cases in which the applicant states that no patentapplication for the same subject matter “has been or will be madeoutside the United States” (US Patents Act, Section 122 (2) (B) (i)).So, even in 2006, trolls – if they target the US market exclusively – caneffectively hide their patent from potential infringers for a long time.We thank one of our referees for sharing this thought with us. A closelyrelated practice is the filing of “continuation patents” (Graham, 2004),which is also restricted to the US. Continuations share the priority dateof the parent application, but may be submitted any time before theparent application and all related (earlier) continuations are decidedupon. Thus, also continuations may serve trolls in hiding their patentsfor long periods of time.

icy 36 (2007) 134–154

As the following model and discussion will show,these firms can generate enormous profits from bettingon being infringed. They are the “sharks”—and it is boththe patent law and its interpretation by the courts thatforms their basis of existence.

3. Infringement rules and marketcharacteristics—a model

In order to illustrate the fundamental mechanismsthat lead to the existence of the “shark” business wedevelop a simple microeconomic model. Despite itssimplicity it captures most of the parameters describ-ing the managerial perspective delineated in Section2.2, particularly technology monitoring efforts, inventaround costs, and institutional legal details regardingindemnification calculations.30 To focus on the mainmechanisms of the shark business and to keep the modeltractable and transparent, some of the relevant variables(e.g. product complexity) are not explicitly parameter-ized and some simplifying assumptions are introduced.31

As Section 4 will show, however, the effects ofthese non-parameterized variables can be discussedverbally.

3.1. Sharks’ rationales and actions

In order to study a fully-fledged “game” betweensharks and their prey we would need to model the (strate-gic) rationales of both players as well as the reactionconnectivity of their actions. In the following, we take

the shark business profitable). We therefore assume thata small capacity-constrained patent holder of a non-

30 We know of no other theoretical contribution parameterizing all ofthese variables in order to better understand the mechanisms that drivepatent infringement. The paper that comes closest to our approach is byBessen and Meurer (in press). In their article, the authors include twoof the aforementioned parameters as variables in a model of endoge-nous patent dispute resolution; namely, these are invent-around andtechnology monitoring. While we share the perceived importance ofincluding these two variables in the theory of patent infringement withBessen and Meurer (in press), the goals, foci and results of the twopapers differ substantially. This is not least because the introductionof damage award calculation details is necessary to understand theprofitability of the shark business.31 For example, we do not model the uncertainty potential infringers

face when assessing whether they actually illegitimately use a thirdparties’ technology or not once they have discovered it (see Bessenand Meurer, in press, for an elaboration of this idea). We will criticallyreflect on the impact of these assumptions during our discussion andin the conclusions.

rch Pol

sattlfipitctpadtt

3

iroAstlenpnmhrl

bcmtpw2tIpt

se

duce the following further notation. LF denotes thelicense fee to be paid by M in case a licensing contract isclosed (node A in the decision tree). The damages to be

33 A similar, though more general, shape of the function p(x) describ-


ophisticated technology has no options but to eitherpproach a manufacturer before infringement (in ordero license out his/her technology), or, conversely, to presshe manufacturer for royalties after infringement (i.e. actike a shark). We neglect the possibility that the smallrm, before patenting, may have the additional option oficking the technological area where it wants to makets investment; nor do we consider the possibility thathe small patent holder may lose its infringement case inourt (see Bessen and Meurer, in press, for an elabora-ion of this aspect). We assume, however, that the smallatent holder can anticipate his/her prey’s actions quiteccurately. Being a shark will consequently be his/herominant strategy if profits from being infringed exceedhe profits from offering the technology to a manufac-urer before infringement.

.2. Manufacturers’ rationales and actions

A company M (“manufacturer”) is considering enter-ng a certain market. Producing the respective goodequires solving a technical problem, to which technol-gy T is the most obvious, but not the only solution.s part of its new product development process, M puts

ome effort into checking if T is patented. Depending onhe result of this monitoring, its prior beliefs about theikelihood of T being patented, cost factors, and salesxpectations, M decides whether to enter the market orot, and if so, with what technology. If T really has beenatented by a firm PH (“patent holder”), this firm couldegotiate a license with M, or it could sue M for infringe-ent if T was used illegitimately. We assume that PH

as no production capacity of its own. This setup cor-esponds to Section 2.2.2, with M having a (potentiallyarge) manufacturing capacity and PH having none.

In more detail, the manufacturer’s logic is describedy the decision tree depicted in Fig. 1 (which the sharkan anticipate). From the perspective of the “innocent”anufacturer it is nature that, in Stage 1, decides whether

echnology T is patented (probability pp) or not.32 M’srior belief is that T is patented with probability ppM,hich will in general be different from pp. In Stage, M decides how much effort x (measured in mone-
ary units) to put into checking whether T is patented.n Stage 3, nature decides if – in the case that T isatented – M finds out about this fact. This discoveryakes place with probability pfind(x) = 1 − e−ax, where a
32 Note that this seeming exogeneity is unrealistic if we assume thatharks make dedicated investment in certain technological areas; how-ver, also aforementioned, to start with we neglect this complication.

icy 36 (2007) 134–154 143

is a constant parameter.33 M is aware of this probability.Depending on the outcome of its patent search, M per-forms Bayesian updating to adapt its beliefs about thepatent protection of T, to either 1 (if a patent has beenfound) or to p’pM(x) = ppM e−ax/(ppM e−ax + 1 − ppM) (ifno patent has been found). The latter term thus gives theperceived probability of patent protection conditionedon negative results after incurring the search cost x.

Stage 4 is only relevant if a patent exists and M hasfound out about it. In this case, M and the patent holderPH negotiate whether to stipulate a licensing contractand at what fee. In all other cases, no action is taken atthis stage. Finally, in Stage 5 M decides whether to enterthe market, and with what technology. We assume thatM has the option of substituting T with an alternativetechnology Tia. Compared to using T, this invent-aroundcauses additional (fixed)-development costs of cia since,as we assumed, T is the most obvious solution to theproblem. Without restriction of generality, we set thedevelopment cost of T to zero since we treat this caseas our benchmark (no matter whether T is a sophisti-cated core technology or not, see Section 2). The heightof the invent-around costs parameterizes the sophistica-tion of the technology. For simplicity we assume, in casea patent on T has been identified and using Tia insteadis considered, that M can verify at a fixed cost (con-tained in cia), and with certainty, that Tia is not patentprotected. While this assumption differs from our mod-eling of how M checks the patent protection of T, it is asecond order effect and neglecting it allows us to keepthe model tractable.

We do not model market interaction, but instead sim-ply assume that M sells Q units of the good at a pricep, with variable cost cv of production and zero fixedcosts (apart from development costs).34 If infringementoccurs, we assume that it is detected with probability pd.

In order to quantify the expected payoffs, we intro-

ing the probability of success of monitoring is used by Crampes andLanginier (2002) who focus on the patent holder’s monitoring effort.34 Instead, we could introduce a fully blown market interaction stage.

In this stage, M would either sell as a monopolist to a set of buyersdefined by a demand curve or M would compete with one or more otherfirms. However, what is relevant to our analysis is solely the outcomeof the market stage in terms of price P and quantity Q, since the patentholder does not appear in the market interaction. Thus, modeling thisstage explicitly would only burden our model and distract from theactual issue.

144 M. Reitzig et al. / Research Policy 36 (2007) 134–154

tree sho

vto 5%. That is, the curve Π0 equals 5% of revenues.Parallel to the top lines in Fig. 2a and b run the curvesΠ0 − cia, in both figures, which become relevant to M

Fig. 1. Excerpt of the “game” between shark and manufacturer. Theindicate information sets.

paid by M in case of enjoined infringement (nodes B, E)are denoted by d. Finally, pdM is the probability, as per-ceived by M, that infringement will be discovered. Justas M’s prior belief ppM about the probability of patentprotection on T may differ from pp, also pdM may differfrom the “true” value pd Denoting by Π0 ≡ Q(p − cv)the gross profit (excluding all costs except variable costsof production) M makes on the market (Stage 6 of thetree), we obtain the following equation for M’s expectednet profit Π, conditioned on the search effort x (seeAppendix A for more details):

E[Π |x ] = ppM(1 − e−ax)

× max{Π0 − LF, Π0 − d, Π0 − cia, 0}+(1 − ppM(1 − e−ax))

× max{p′pM(pdM(Π0 − d) + (1 − pdM)Π0)

+(1 − p′pM)Π0, 0} − x (1)

From Eq. (1), one can calculate M’s optimal searcheffort x*, given its beliefs. We refrain from rendering thisrather complex expression. Still, the formal presentationshows two main aspects.

ws the decision-making rationale of the manufacturer. Dashed lines

First, M might underestimate the probability of exist-ing patent protection (ppM < pp) as well as that ofinfringement being discovered (pdM < pd), leading to asuboptimal (too low) level of monitoring.35

Second, and most importantly: the cost of inventingaround the patented technology becomes relevant as athreat point in various ex ante licensing negotiation sce-narios. As Eq. (1) shows, in a real ex ante (i.e. beforeinfringement) licensing negotiation, M will at most bewilling to pay d or cia as a royalty to PH, depending onwhich of the two figures is smaller. Fig. 2 illustrates thisconsideration showing three different profit curves forM in each of the figures (a) and (b).

Gross profit Π0 as a function of quantity sold, Q, isshown as the top line in both Fig. 2a and b. For illustra-tion purposes, we set the contribution margin (p − c )/p

35 Note that we arrive at this result without even assuming that theshark strategically patents in the technological domain of the manufac-turer. The effect may be even stronger if we relax this latter assumption.

M. Reitzig et al. / Research Policy 36 (2007) 134–154 145

F enariosp

itCicpe“tDirspn

i

4

cpicetu

4i

F

ig. 2. Payoffs of M as a function of output quantities, for different scossible outcomes of ex ante licensing negotiations.

n case it has discovered the patent on T and decideso use the non-infringing technology Tia instead (node). Finally, M’s profit curves for the case of enjoined

nfringement (Π0 − d) are shown for different damagealculation rules: Fig. 2a shows the case of “infringer’srofits” (Π0 − dip), in which, by definition, M’s net profitquals zero. Fig. 2b, in contrast, depicts the case of areasonable royalty” (Π0 − dsr), where we assume thathe court applies a standard royalty rate of 2% of sales.espite their general importance as the third possible

ndemnification (see Section 2.1), “lost profits” do notequire separate treatment in our model and hence noeparate illustration, since they refer to the shark’s “ownroduction” (see Section 2.1), which is set to 0 by defi-ition.

The implications of our model findings are discussedn the following.

. Results and discussion

We now employ the model developed above to dis-uss the main questions of our article, namely whyatent sharks exist, why the shark business is seem-ngly growing, which countermeasures manufacturersan take, and, finally, which policy debates appear rel-vant. Wherever relevant, we will briefly elaborate onhe importance of the findings for particular jurisdictionssing our overview presented in Table 1.

.1. Patent sharks—maximizing profits from suing
nadvertent infringers
In the middle branch of our tree (nodes E and F inig. 1) the potential infringer is not aware of the patent

(excluding monitoring cost x). Shaded areas and broken lines indicate

on technology T held by firm 1 (we briefly commentedon the possible reasons for his/her surprise in Section2.2.2 and elaborate on this rationale in Section 4.2). If,as we assume, T is the cheapest and most obvious solu-tion to the technical problem at hand, then the likelihoodof firm M inadvertently infringing T rather than uncon-sciously inventing around the patent is high (that is, firmM is likely to end up in node E). As we will show, exist-ing patent indemnification remedies induce incentivesfor the patentee PH to trap manufacturers in such sit-uations (node E) and act as trolls. In more detail weargue that courts’ unwillingness to consider hypotheti-cal invent-around costs as a benchmark for the size ofdamage awards in tort cases is the key to the success ofPH’s strategy.

To better understand this rationale, we need tojuxtapose the outcome of realistic ex ante licensing nego-tiations between M and PH with the fictitious ex posttreatment of inadvertent infringement cases in court. Wecommence by linking back to our model and describ-ing profit-maximizing rationales for manufacturers insituations of complete information about patented tech-nology.

In the scenario where M has discovered that technol-ogy T is patent protected and M enters into licensingnegotiations with the patent holder PH, the outcomeof the bargaining process depends on both players’threat points. For the purpose of our paper, two majorscenarios must be distinguished: either that inventingaround the patent is a feasible alternative to paying dam-
ages/licensing, or that it is not.
When inventing around the patent is more attractivefor M than paying the (anticipated) damages d, then Mis willing to pay a royalty of at most cia to PH, while

rch Pol
PH demands a licensing fee that is at the very least posi-tive. PH cannot credibly threaten to leave the negotiationtable since, in this case, M would enter the market usinga non-infringing technology. Hence, depending on thetwo parties’ respective negotiation power any licens-ing fee between zero and cia is possible, as shown bythe shaded areas in Fig. 2a and b. The dashed lineswithin the shaded areas indicate the Nash bargainingsolution, which divides the surplus equally between theplayers.

For those cases in which inventing around the patent isnot an option, outcomes differ between the two indem-nification regimes as depicted in Fig. 2a and b. Whena standard royalty rate is applied (Fig. 2a), then playerPH can credibly threaten to terminate the negotiations,since in this case M would still enter the market usingthe infringing technology. Hence, the licensing fee willbe equal to the anticipated damages, dsr (the “classical”equilibrium outcome as assumed in most prior literature,see Schankerman and Scotchmer, 2001). In contrast,when damages equal infringers’ profits (Fig. 2b), thenthe threat to stop negotiating is again not credible, sincein this case M would not enter the market (yielding aprofit of zero for PH). Hence, as in the case above, alloutcomes between the two threat points (zero, Π0) arepossible, depending on the players’ negotiation power.

Interestingly, however, the following insight holdsirrespective of the relevant indemnification rule and thedistribution of bargaining power between the differentplayers: above a certain threshold quantity of sold goods,M’s profits in the case of successful ex ante licensingnegotiations theoretically exceed his/her (counterfac-tual) profits in case of conviction of patent infringement.Moreover, this difference increases with M’s outputquantity. Also, it will be the more pronounced the lowerthe cost cia of inventing around, in other words, the moreobvious the troll’s technology.

Sharks can come into existence because the law, atleast currently, refuses to accept this fundamental logicas delineated in the last paragraph. In the case of infringe-ment, no matter whether inadvertent (middle branch ofour tree) or wilful (upper part of the tree), the patenteeenjoys the freedom to pick the remedy that maximizeshis/her profits. Depending on the jurisdiction of concern(see Table 1) the patentee will be able to claim the fullinfringers’ profits, which might far exceed the realisticroyalty fee. But even if the law “only” offers a reasonableroyalty fee as compensation for the infringement, “being
infringed” may be far more profitable than entering reallicensing negotiations ex ante. This is due to the problemthat courts refuse to assess counterfactual invent-aroundcosts during the trial—leading to a distorted calculation
icy 36 (2007) 134–154

of the “reasonable royalty rate” that may again exceeda realistic ex ante licensing fee by far. Take the follow-ing case as an example and concentrate on the damageaward figures (rather than the well-known “submarine”tactics).

In 1990, individual inventor Jerome H. Lemelsonappealed at the US District Court for the NorthernDistrict of Illinois, Eastern Division against an earlierjudgment according to which Mr. Lemelson had beengranted damage awards for the non-willful infringementof his patent on a coupling technology used by Mattel,Inc. in one of the corporation’s toy trucks. Mr. Lemelsontried to prove that Mattel, Inc. had willfully infringedon his patent, and in accordance with US law he soughtto be reimbursed with a triple licensing fee. Mr. Lemel-son’s idea of multiple damages at the appellate courtwas denied; however, the single royalty rate he wasgranted for the inadvertent (!) infringement by MattelInc. still amounted to 24,780,000 US$. The royalty ratewas calculated using the standard remedy calculationsfor royalty rates as a standard 4.5% industry percentageof all truck toy sales of Mattel Inc. between 1971 and1986. From an economic standpoint, this result seemsvery odd. As a matter of fact, if the aforementioneddamage award captured the hypothetical ex ante bargain-ing process correctly, this would mean that Mattel, Inc.would have had a willingness to pay roughly 25 millionUS$ to Mr. Lemelson. Given the status of Lemelson asan individual inventor, his need to access complementaryassets to produce a competitive toy truck himself, MattelInc.’s likely low costs of developing an alternative cou-pling mechanism, and the obvious lack of willfulness onthe part of the infringer, we leave it to the reader to judgewhether he/she is convinced that the actual reimburse-ment Mr. Lemelson received reflects an economicallysuitable damage award or whether the result is an out-come of a standard application of a legally accepted, buteconomically incommensurate, remedy calculation.

Towards the end of this section, we would like to pickup on two issues which are also nicely illustrated by theaforementioned example.

Often, cases of this kind are used as examples toillustrate the value of hiding patented technology (sub-marines). A closer look, however, reveals that keepingan invention secret is not a sufficient condition for trollsto run a profitable business. Hiding does increase theodds that the potential infringer will overlook the patent,which is a necessary condition for the troll to succeed.
However, only the unrealistic treatment of fictitious exante licensing negotiations gives value to “hiding tech-nology”. Conditional on the patent not being found bythe infringer, this inadequate legal treatment constitutes

rch Pol

at

mcaHtto4wiTs

4c

ttIctpfawiofi

eam

hiMmmtLrssi

f

uct components that inventing around (though ex antecheap) becomes extremely costly ex post (see Bekkerset al., 2002).

38 It appears an interesting question both for practitioners and for legalscholars if cross-licensing agreements can contain provisions grant-


necessary and sufficient condition for the troll businesso be profitable.

Moreover, at this point we briefly recall that in ourodel we have abstracted from the fact that a typi-

al “troll patent” usually covers only one technologymong many others contained in a complex product.owever, as the example also shows, sharks do not need

o worry too much about the marginality of their inven-ions. Lemelson’s invention was only one among manythers used in the product, and still led to a royalty rate of.5%. Thus, even under more realistic assumptions thane could model, being a “troll” emerges as the legit-

mate dominant strategy for small low-tech inventors.his does not mean that it is not problematic, as we willhow more clearly in the following.

.2. Why “being infringed” is a strategy of growingoncern

In our eyes, the Mattel case gives reason to believe thathe calculation of standard royalty rates creates incen-ives for small patentees to be infringed by large firms.t is true that the particular setup of the aforementionedase is US-specifie and that, in order for a similar caseo happen nowadays (i.e. after the U.S. adjusted theiratent disclosure policies) the plaintiff would have toorego filing for related patents outside the U.S. On bal-nce this means that “submarine” patenting strategies,hich have always been restricted to the U.S., still work

n this country basically as they did before. Trolls are nownly restricted in so far as they can no longer additionallyle for (non-submarine) patents outside the U.S.

The increasing shark activity outside the U.S.,36 how-ver, illustrates that submarine patents and continuationsre not to be the only source of “surprise” for large R&Danufacturers. There are two other potential reasons.First, a published patent or patent application may

ave slipped the manufacturer’s attention despite mon-toring. Given the overall surge in patenting (see also

acdonald, 2004, p. 141; Blind et al., 2006, p. 656 ff.),37

onitoring costs for existing patented technology haveassively increased. It thus seems fair to assume that

he overall risk of neglecting prior art has risen (seeemley, 2000; Quillen and Webster, 2000 for an elabo-

ation of this argument; these thoughts were furthermore
upported by interviews that the authors conducted witheveral Heads of patent departments of leading R&Dntensive firms). Inadvertent infringement is particularly
36 See our introductory cases in Footnotes 6 and 7.37 See the statistical bulletins of the European and the US Patent Officeor impressive evidence.

icy 36 (2007) 134–154 147

likely if the manufacturer develops a technical compo-nent in a complex technological industry where several(patented) inventions enter the final product (see Mergesand Nelson, 1990; Kash and Kingston, 2001; Ziedonis,2004), and that particular technical component does notreflect one of the firms’ core R&D products. Patents ontrivial or already existing inventions are particularly hardto monitor, since the engineer intending to use themmay not perceive them as patentable at all and hencemay not find a patent search worthwhile. Manufacturingoutside a firm’s home territory or importing the afore-mentioned technological components increases the riskof inadvertent infringement even further.

Secondly, another potential reason for “litigation sur-prises” is that the manufacturer did not consider a patentknown to him – or even all patents within a certaintechnology class – a threat at the time he/she startedmanufacturing since it was owned by a “non-litigious”patentee. Such “non-aggression” pacts may be explicit,based on cross-licensing, or implicit, based on industryculture and mutual deterrence. Troll surprise may thenarise when the patent changes ownership (often due to thepatentee’s bankruptcy),38 when it is licensed to a troll, orwhen the patent holder changes its business model from“producing” to “trolling”.39 And as a matter of fact wethink therefore that, even though submarine patents maybe abolished in the near future, the overall importanceof the shark phenomenon increases.

Sharks’ chances of running a profitable business inthese industries are furthermore enhanced by the irre-versibility of substituting infringing components withalternative technologies after a certain point. To managethe complexity of firm boundary spanning R&D projects,hardware and software standards have assumed majorimportance in these technologies. Once certain specifi-cations of a standard are frozen, the substitution of oneinfringing technology with an easy-to-invent alternativeentails the adjustment of such a large number of prod-

ing the licensee an option to buy the patent in case of the licensor’sbankruptcy. We thank an anonymous reviewer for sharing this idea withus. Such provisions would help to forearm against trolls in the case of(explicit) cross-licensing. However, all implicit “non-aggression pacts”would have to be made explicit lest the door for trolls remains open,entailing potentially onerous transaction costs.39 Prominent examples of the three types are Acacia Technologies

(acquisition of patents), InPro (licensing), and Patriot Scientific Cor-poration (changed business model).

rch Pol

their comparative advantages; namely, using their politi-cal influence to lobby for regulations which put an end tothe business of “being infringed”.43 This option may be


Our model captures the above considerations. Mmight underestimate the probability of existing patentprotection (ppM < pp) as well as that of infringementbeing discovered (pdM < pd). Both assumptions seemquite realistic given that the activity of patent trollshas increased considerably in recent years. Hence, largemanufacturers might not yet be accustomed to the num-ber of “troll patents” and the degree of monitoring trollsexert. Since monitoring is one of the troll’s main businessactivities, and since monitoring a few potential (large)infringers is both feasible and potentially most profitablefor the troll, the “true” probability pd is likely to be closeto 1.

Underestimating the above probabilities implies thatM chooses too low a value for its monitoring effortx*, as Eq. (1) shows. When ppM becomes smaller, thenthe benefits of searching for the patent – and thus, ifit is found, of being able to maximize profits within the(advantageous) first line of Eq. (1) – decrease. When pdMbecomes smaller, then the outcome of the second line’smaximization becomes more attractive, again makingmaximization along the first line’s entries relatively lessdesirable. In addition, E[Π|x] is overestimated (for anygiven value of x, and in particular for x*), potentiallyleading to market entry decisions which should not havebeen taken.40

Finally, the creation, in 1982, of the US Court ofAppeals for the Federal Circuit (CAFC) is generally seento have strengthened the position of patent applicantsand patent holders (e.g., Lerner, 1994; Lanjouw, 1994;Lanjouw and Shankerman, 1997; Kortum and Lerner,1999). This fact is illustrated by a CAFC decision in2002, which basically dismissed obviousness as an argu-ment for rejecting a patent application (Hall and Harhoff,2004). The authors quote deputy commissioner EstherKepplinger saying this ruling means that “we cannotreject something just because it is stupid”.41 Given thefocus of patent trolls on technically simple technologies,this ruling further simplifies their activity.

4.3. Potential counteractions by manufacturers

From the above discussion it appears intuitive thatmanufacturing firms are well-advised to prepare for

40 More details are provided in Appendix A. We do not examine theeffects of uncertainty associated with the outcome of court cases andassume that patent holders will always win. Including the possibilitythat courts may “mistakenly” rule in favor of the infringer (see Bessenand Meurer, in press) would likely lead to downward corrections ofoptimal values for x.41 Source: David Streitfeld, “Note: This Headline is Patented” (L.A.

TIMES, 07 February 2003).

icy 36 (2007) 134–154

shark attacks. Understanding the strategies that sharkspursue is key to designing effective counteractions.

One of the approaches that manufacturers can pursueis to increase their monitoring efforts. Being faced with apotentially increasing “infringement” business, the allo-cation of resources to this activity may require a timelyrevision. “Freedom to operate” reports need to becomean even more crucial element in functional IP manage-ment, eventually co-determining the choice of a firms’entire technological trajectory.

But monitoring will always be imperfect, even if database research is supported by exploiting social networks(see Singh, 2005; Sorenson et al., 2005) especially ifthe “creativity” of sharks increases. Take the followingas an indication: with dedicated investment funds todaystarting to buy patent portfolios of bankrupt estates withthe intent to pressurize potential infringers it will be everharder for a manufacturer to foresee where the dangerseventually lurk.42

For these reasons R&D intensive manufacturers arewell-advised to create independence from particulartechnological solutions. Designing technological stan-dards and products in such a fashion that non-coreelements can be substituted by a range of alternativesappears more important than ever before. Building tech-nological solutions based on open-source standards willfurther reduce the likelihood of ending up in inadvertentinfringement, for two reasons. First, not only the focalfirm, but also other parties have an interest in search-ing for potentially infringed patents. Second, disclosinga development as open source immediately turns it intoprior art, such that (at least in principle) no patents on theinventions contained in the development can be grantedanymore.

Finally, however, large innovators may exploit one of

42 To understand this logic consider the following: in complextechnologies (Merges and Nelson, 1990) large R&D intensive manu-facturers often enter multilateral cross-licensing agreements with otherplayers (see Hall and Ham Ziedonis (2001) for a study in the semicon-ductor industry). In the case of inadvertent infringement by either oneof the parties, disputes will often be resolved backstage; hence, “over-looking” another player’s IP is far less dangerous than that of a shark.If, however, a player goes bankrupt and sharks manage to buy this ex-player’s portfolio out of the bankruptcy estate, the ex-player’s initiallyharmless patents become a real danger for the remaining manufacturer.43 For example, Microsoft and other large corporations are lob-

bying legislation for a patent reform which was introduced in theHouse of Representatives in June 2005. Among other things, it pro-poses to “limit a patentee’s access to injunctions by requiring a

rch Pol

awt

4i

pppowRibllt

vtiiwptoie

ctsfist

lhipol

(

tN(((e


very feasible one, since, as the following Section 4.4ill show, there are various objectively good arguments

o support such an initiative.

.4. The policy side—considerations for an optimalnfringement rule

Whereas “being infringed” is an interesting androfitable strategy for small innovators, it poses greatroblems to large R&D intensive manufactures, andotentially to society. From a policy perspective, anptimal patent indemnification rule should maximizeelfare that is generated for all parties, including sharks,&D intensive manufacturers, and consumers. Design-

ng such an optimal indemnification rule would beeyond the scope of this paper. Nevertheless, we wouldike to introduce some basic considerations that will shedight on some of the deficiencies of the current regula-ions.

The economic purpose of the patent system is to pro-ide incentives for innovation by allowing the patenteeo control the use of the patented technology for a lim-ted period of time. The social gains derived from thesencentives and the patent system’s disclosure function areeighed against the inefficiencies resulting from marketower, the cost of the patent system, and the restric-ions imposed on subsequent innovators.44 The questionf how to strike the optimal balance – in particular thessue of patent length and patent breadth – is complexven in a world without patent infringement.45

The trade-off, however, does become even more diffi-ult in a world where infringement can take place. If weake the – plausible – short cut that the patent grantingystem should define how much an innovator is rewardedor his/her invention, then allowing (not prosecuting)
nfringement cannot be considered socially beneficialince it would distort optimally chosen innovation incen-ives due to the patentee’s shaken trust in the system.
ikelihood of irreparable harm” (Steptoe and Johnson LLP, 2005,ttp://www.steptoe.com/publications/PI10264.pdf). While not obviat-ng the troll business altogether, this proposal would weaken theirosition considerably. Also, a post-grant review in the style of thepposition procedure at the European Patent Office is favored by theobbyists (news.zdnet.com, 13 September 2005).44 See, e.g., Blair and Cotter (2001, 45 and 46), Gallini and Scotchmer2002) and Henkel and von Hippel (2003).45 The economic analysis of incentives to innovate and the role ofhe patent system goes back at least to Arrow (1962), Nelson (1959),ordhaus (1969), and Schmookler (1966). See Gallini and Scotchmer

2002) for a comprehensive discussion, and Grossman and Helpman1991) in the context of economic growth. See Gilbert and Shapiro1990) and Klemperer (1990) for an economic model assessing theffects of patent breadth.

icy 36 (2007) 134–154 149

Thus, if some kind of ‘infringement’ (use of the patentedtechnology by others than the holder) was deemed over-all beneficial by policy makers, then it should make senseto define the underlying patent more narrowly in thefirst place (thus defining an otherwise infringing act as alegitimate one).46 Following this rationale, we can sub-scribe to the view that damages should at least coverthe losses the innovator incurred due to infringement.That is, he/she should be put into a position “but for” theinfringement.

The question then is what the patentee’s damagesamount to? Here, we need to distinguish two major sce-narios. When joint profits of patent holder and infringerare decreased by the infringer’s use of the technol-ogy, then the patentee’s “lost profits” are unambiguouslydefined as the difference between its profits without andwith the use of the technology by the infringer.

However, when their joint profits are increasedthrough the infringement – as is the case, e.g., forresearch tools (Schankerman and Scotchmer, 2001) aswell as in our model – then the situation is less clear-cut,and we have to consider two further sub cases.

First, consider the case of full information andnon-substitutable technologies (sub case 1). Absentinfringement, the two parties would have negotiated alicensing contract. As to the licensing fee, we lean onthe rationale put forth by Schankerman and Scotchmer(2001) arguing that the outcome of ex ante licensingnegotiations will depend on what the law promised thepatentee as a remedy in the case of infringement. Sincethe latter refers to the outcome of hypothetical ex antelicensing negotiations, the ex post remedy and the ex antelicensing fee will be self-enforcing. Infringement shouldnot take place in equilibrium (see the upper branch ofour tree). Quite obviously, from a real-world perspectivethis theoretical view is unsatisfactory, if only for the factthat it cannot explain infringement as anything but an
“out-of-equilibrium” event.
The situation differs when incomplete informationand substitutive technologies are introduced (sub case

46 In the extreme case, “defining the patent more narrowly” may meannot granting it at all—that is, in general terms, to increase the thresh-old level of non-obviousness and the quality of prior art search. Asmentioned earlier, many troll patents have been characterized as triv-ial and/or non-novel, and some have indeed been invalidated for thatreason (e.g., the U.S. Patent Office found that prior art “completelyanticipated the broadest claims of the patent” No. 4,698,672 owned byForgent Networks, see http://www.pubpat.org/Chen672Rejected.htm).Thus, increasing patent quality clearly is another important avenue forobstructing the troll business. Of course, a more restrictive grantingpolicy must be implemented carefully to avoid “good” applicationsbeing erroneously rejected alongside trivial ones.

http://www.steptoe.com/publications/PI10264.pdf

http://www.pubpat.org/Chen672Rejected.htm

rch Pol

manufacturing R&D intensive firms, due to balloon-ing numbers of patent applications, probably led to theincrease of sharks’ relevance for innovators. It facilitates

47 It strikes us as extremely counterintuitive to base the royalty ratecalculation in the case of inadvertent infringement on the averagepercentage of standard industry contracts. The latter are outcomes ofreal licensing negotiations, while an inadvertently infringed patent hasnever been subject to such negotiations. Most likely, a problematicselection separates the different cases, however. A patent holder whoanticipates the outcome of ex-ante licensing negotiations to lie belowthe average industry royalty rate has an incentive to avoid such negoti-ations; instead he/she will aim at being awarded damages ex-post andact as a troll. This incentive is larger the higher the damage awards are,and should be particularly pronounced whenever “infringer’s profits”


2). With incomplete information, inadvertent infringe-ment does take place in equilibrium (node E in thetree). At the same time, it implies that the assess-ment of hypothetical ex ante licensing negotiationsby courts becomes much more difficult, since aninadvertent infringer did not even have the chanceto enter such negotiations. Had he/she done so,he/she would have considered the option to use asubstitutive technology instead. As we have arguedabove, neglecting this option of inventing aroundthe patented technology can lead to highly exag-gerated estimates of hypothetical ex ante licensingfees.

The fact that these latter cases of inadvertent infringe-ment are not treated realistically by the courts mayexplain why we observe significant numbers of patentinfringement cases, even though classical theory (subcase 1) cannot explain them. In our eyes, there is there-fore a pressing need for a theoretical elaboration ofthe economics of patent infringement. While the detailsof this research must be left to future studies, we do,with all due modesty, strongly encourage legal policymakers to consider these downsides of current indem-nification practice, especially in light of the rising riskof unconscious infringement. Introducing hypotheticalinvent around costs as an element in the equation to cal-culate damages in inadvertent infringement cases strikesus as necessary. We recommend such a reconsidera-tion despite the notorious difficulties that courts facewhen assessing counterfactual situations (like hypothet-ical invent around scenarios). By passing its Directive2004/48/EC on the enforcement of intellectual prop-erty rights, the European Parliament and the Councilsignaled their insight and their wish that patent indemni-fication needs to become “more realistic”. In particular,Article 13 1a explicitly encourages the consideration of“all appropriate aspects” to determine the proper dam-age award. This, however, should also entail aspectsregarding the infringed (and not only the infringing)party and their ulterior motives for being infringed.Quite clearly, however, the consideration of such addi-tional parameters must not relieve manufacturers oftheir responsibility to keep monitoring prior art technol-ogy altogether. Trading off these parameters to designan optimal infringement rule will be a challengingtask.

5. Conclusions and outlook on future research

Motivated by the vast gap between the managerialrelevance of patent trolls on the one hand and the theo-retical understanding of the profitability of their business

icy 36 (2007) 134–154

on the other, this paper sought to answer four dif-ferent questions; namely, why sharks exist, why theirimportance has increased, what countermeasures man-ufacturers can take, and what policy debates should beheld. We addressed these questions by mapping interna-tional legal indemnification rules for patent infringementwith managerial rationales of capacity-constrained hold-ers of (simplistic) patents who – by assumption – havetwo choices: they may enter licensing deals about theirtechnologies with potentially interested manufacturersshortly after discovery/patenting, or they may wait to beinfringed (i.e. act like a shark).

By doing so, we were able to show that a necessaryand, conditional on the patent not being found by thepotential infringer, sufficient condition for these firms toact most profitably as sharks is the inadequate (unrealis-tic) treatment of hypothetical ex ante (i.e. before patentinfringement) licensing negotiations between the patentholder and infringer in courts. In other words: undercurrent indemnification regulations, “being infringed” isthe dominant innovation exploitation strategy for small,capacity-constrained firms owning trivial patents. Thiscentral finding seems to hold across all the jurisdic-tions we studied, although it is particularly relevant incountries such as Germany where infringers’ profits areawarded as one potential remedy. However, as we alsodemonstrated, even in countries where reasonable roy-alties are the only possible indemnification for a shark,the mistakenly high benchmark using standard indus-try rates overcompensates the troll and renders beinginfringed valuable.47

We further argued that the increasing technologymonitoring efforts for victims of trolls, namely large

can be awarded (see Fig. 2b). In contrast, a patentee holding rightsto a technically sophisticated core technology will often not be sat-isfied when awarded a standard royalty rate ex-post, such that forthose patent holders “being infringed” becomes less attractive thannegotiating ex-ante.

rch Pol

‘gce(mmappcepttulwittfaccp

aqr

caceAprscwarmhr

buo

c

Acknowledgments


trapping’ manufacturers by ‘hiding’ patented technolo-ies in confusing patent thickets—a second necessaryondition for sharks to operate. Moreover, the strength-ning of patent holder’s rights in certain jurisdictionse.g. the US) most likely enabled sharks to operateore profitably, too. We illustrated that R&D intensiveanufacturers are well-advised to revisit their budget

llocations for technology monitoring efforts and toatent alternatives to their core inventions (in com-lex industries), and that concerted lobbying efforts tohange patent indemnification laws may be promising;specially since, as we showed, sharks potentially dissi-ate social value by reducing manufacturers’ incentiveso innovate. And since they do so, sharks are a mat-er of concern for policy makers who – in our eyes –rgently need to revisit the practice – rather than theaw itself – of patent indemnification. In more detail,e suggest that inadvertent infringers’ trade-offs before

nfringement be more realistically captured by courtshan is currently the case. The dangers associated withhe malassessment of counterfactuals (like the manu-acturer’s invent around option ex ante) – a classicalrgument potentially to be brought forth against our con-lusion – does, in our eyes, not vindicate the simplisticurrent practice by the courts that strengthens the trolls’ositions.

As is common in research, this paper left us withs many questions as it did answers. Some of theuestions strike us as relevant avenues for furtheresearch.

One trajectory of research is theoretical and ofoncern for theoretical scholars in the areas of lawnd economics. In Section 4.4, we showed that theurrent theory on patent infringement stops short ofxplaining the large number of infringement cases.ccording to Schankerman and Scotchmer (2001),atent infringement does not take place in equilib-ium. While we acknowledge their circular logic ofelf-enforcing royalty rates and damage awards in thease of complete information about prior technology,e also showed that inadvertent infringement followsdifferent logic. Elaborating on this line of thought may

efine our theoretical understanding of patent infringe-ent (see also Bessen and Meurer, in press) and may

elp to suggest a differentiated optimal indemnificationule.

A second trajectory of further research is empiricaly nature. Here, various questions appear intriguing to
s, of which we briefly mention the two most interestingnes.
As Table 1 of this article shows, national idiosyn-rasies in the jurisdiction of patent indemnification exist.

icy 36 (2007) 134–154 151

If our model assumptions are correct and of relevance,then we would expect that shark strategies as well as theirprofitability differ from country to country. “Infringers’profits” (as long as Directive 2004/48/EC is not fullyadopted by all the EC member states) in Germany mayrepresent better bait than reasonable royalties in, say,France. Despite the theoretical possibility of sharks suingmultinationals in various countries of jurisdiction, wewould expect to see a concentration on certain nationalmarkets where, among other things, patent indemnifi-cation rules and practice would create incentives forinitiating a troll business.

Finally, we deem it an extremely interesting ques-tion to further inquire empirically into the increasingprofessionalism of the trolls. R&D manufacturers areobserving with growing concern how parts of patentportfolios are strategically bought out by dedicatedinvestment funds, for example during bankruptcy pro-ceedings. As mentioned in Sections 2.2.2 and 4.2, therehave been cases where several patents, originally ‘harm-less’ for all manufacturers since they had been used inmultilateral cross-licensing negotiations between the dif-ferent players prior to one of them filing Chapter 13, hadbecome dangerous all of a sudden when they fell intothe hands of the non-manufacturing investment fund.While the details of these particular cases are often notpublic, various other indications of such an increase inshark professionalism have been discussed in the popularmedia. One of the firms receiving ambivalent criticismis Nathan Myhrvold’s “Intellectual Ventures”. Specializ-ing in the exploitation of inventions without engaging inproduction themselves, Intellectual Ventures both per-forms internal research and buys third parties’ patentsin industries that are of relevance to them. These latteractivities, officially dedicated to forearming clients ofIntellectual Ventures (namely R&D intensive manufac-turers!) against potential sharks, have been criticized astroll-like activities by the firm itself.48 No matter whichstandpoint one takes, the rising intensity and complex-ity of the patent acquisition and sales business and theengagement of non-producers lift this source of “patentsurprises” to a whole new level, massively complicatingthe problem of cross-licensing, whose empirical rele-vance is, in our view, important to study.

The authors gratefully acknowledge the travel sup-port received by the Danish Social Science Council

48 See Newsweek electronic edition, 22 November 2005 (accessibleat http://www.msnbc.msn.com/id/6478691/site/newsweek/).

http://www.msnbc.msn.com/id/6478691/site/newsweek/

rch Pol

1 − ppM(1 − e−ax) yields a quadratic polynomial in e−ax


within the RIPE network program. The authors wouldlike to thank various colleagues who provided thoughtfulcomments on earlier versions of this paper, particularyAshish Arora, Andrea Fosfuri, Henrik Lando, MariaTheresa Larsen, and Olav Sorenson. The usual dis-claimer applies.

Appendix A. Appendix

In the following, we elaborate on the derivation andanalysis of Eq. (1), which has the following shape:

E[Π |x ] = ppM(1 − e−ax)max{Π0 − LF, Π0

−d, Π0 − cia, 0} + (1 − ppM(1 − e−ax))

× max{p′pM(pdM(Π0 − d) + (1 − pdM)Π0)

+(1 − p′pM)Π0, 0} − x (1)

Given M’s prior belief that a patent exists with proba-bility ppM and that, conditional on the patent’s existence,M will discover it with probability (l − e−ax), the termppM(l − e−ax ) denotes the overall probability (as per-ceived by M before searching) of M finding a patent onT if it expends the search effort x. Having found a patent,M maximizes its profits according to the first line of Eq.(1).

Correspondingly, the term (1 − ppM(1 − e−ax)) in thesecond line describes the probability (as perceived byM before searching) that M will not find a patent onT, given its effort x. Conditional on not having found apatent, M updates its belief about the probability that apatent nonetheless exists, following the rules of Bayesianupdating, to

p′pM(x) = ppM e−ax

ppM e−ax + 1 − ppM

If no patent exists (probability (1 − p′pM)), then

the expected payoff from entering the market simplyamounts to Π0. If a patent exists, then the expectedpayoff is given by (pdM(Π0 − d) + (1 − pdM)Π0), withpdM denoting the probability, as perceived by M, that itsillegitimate use of the patented technology will be dis-covered and infringement will be enjoined. The secondline of Eq. (1) can be simplified, yielding

E[Π |x ] = ppM(1 − e−ax)

× max{Π0 − LF, Π0 − d, Π0 − cia, 0}+(1 − ppM(1 − e−ax))

× max{Π0 − p′pMpdMd, 0} (1′)

icy 36 (2007) 134–154

From this equation one can calculate M’s optimalsearch effort x* by differentiating with respect to x, set-ting the result to zero, and solving for x. In doing so,one has to distinguish between the cases that maximiza-tion in the second line of Eq. (1′) yields either 0 orΠ0 − p′

pMpdMd. In case the maximization yields 0, weobtain

d

dxE[Π |x ]

= appM e−axmax{Π0−LF, Π0 − d, Π0 − cia, 0}−1.

Setting this expression to zero and solving for xyields x* (the equation holds in the range where Π0 −p′

pMpdMd < 0):

x∗ = a−1 ln(appM max{Π0−LF, Π0−d, Π0 − cia, 0}).

It becomes immediately clear that a too low priorbelief about ppM on the part of M implies a too lowsearch effort. The case Π0 − p′

pMpdMd > 0 is a morecomplex. Using the abbreviation

B = max{Π0 − LF, Π0 − d, Π0 − cia, 0},

we obtain the following equation for the derivative ofE[Π|x]:

d

dxE[Π |x ] = appM e−ax(B − Π0 + p′

pMp′dMd)

+(1 − ppM(1 − e−ax))pdMddp′

pM

dx− 1

= appM e−ax

(B − Π0

+pdMdppM e−ax

1 − ppM(1 − e−ax)

)

+adppMpdM e−ax 1 − ppM

1 − ppM(1 − e−ax)

−1.

Setting this term to zero and multiplying by

solving for e−ax, taking the logarithm, and dividing by−a yields x*. However, the resulting equation is ratherinvolved, and actually more difficult to interpret than theoriginal Eq. (1). We thus refrain from displaying the fullexpression for x*, instead providing in the main text adiscussion of Eq. (1).

rch Pol

R

A

A

A

A

B

B

B

B

B

C

C

C

C

E

G

G

G

G

G

H

H


eferences

grawal, A., Garlappi, L., 2002. Public sector science and the strategyof the commons. Academy of Management Best Paper Proceed-ings.

rora, A., Fosfuri, A., Gambardella, A., 2001. Markets for Technology:The Economics of Innovation and Corporate Strategy. MIT Press,Cambridge, MA.

rora, A., Ceccagnoli, M., 2006. Patent protection, complementaryassets, and firms’ incentives for technology licensing. ManagementScience 52 (2), 293–308.

rrow, K.J., 1962. Economic welfare and the allocation of resourcesfor invention. In: Nelson, R. (Ed.), The Rate and Directionof Inventive Activity. Princeton University Press, Princeton,NJ.

ekkers, R., Duysters, G., Verspagen, B., 2002. Intellectual propertyrights, strategic technology agreements and market structure: thecase of GSM. Research Policy 31 (7), 1141–1161.

essen, J., Meurer, M., in press. Patent Litigation with EndogenousDisputes, American Economic Review Papers and Proceedings.

lair, R.D., Cotter, T.F., 2001. Rethinking patent damages. Texas Intel-lectual Property Law Journal 10, 1–93.

lind, K., Edler, J., Frietsch, R., Schmoch, U., 2006. Motives to patent:empirical evidence from Germany. Research Policy 35, 655–672.

rinkhof, J., 2000. The enforcement of patent rights in the Netherlands.International Review of Intellectual Property and Copyright Law31, 706–722.

asucci, G., 2000. The enforcement of patent rights in Italy. Inter-national Review of Intellectual Property and Copyright Law 31,692–705.

ohen, W.M., Nelson, R.R., Walsh, J.P., 2000. Protecting their intellec-tual assets: Appropriability conditions and why U.S. manufacturingfirms patent (or not). NBER Working Paper 7552.

rampes, C., Langinier, C., 2002. Litigation and settlement inpatent infringement cases. RAND Journal of Economics 33,258–274.

ornish, W., Llewelyn, D., 2000. The enforcement of patents in theUnited Kingdom. International Review of Intellectual Property andCopyright Law 31, 627–645.

dison, T., 1898. Interview in Scientific American 78 (2): 19. Reprintedon the www at http://www.myoutbox.net/posa78n.htm, accessedon January 4, 2006.

allini, N., 1992. Patent policy and costly imitation. RAND Journalof Economics 23, 52–63.

allini, N.T., Scotchmer, S., 2002. Intellectual property: when is itthe best incentive system? In: Jaffe, A., Lerner, J., Stern, S. (Eds.),Innovation Policy and the Economy, vol. 2. MIT Press, Cambridge,MA.

ilbert, R., Shapiro, C., 1990. Optimal patent length and breadth.RAND Journal of Economics 21, 106–112.

raham, S., 2004. Continuation, Complementarity, and CapturingValue: Three Studies Exploring Firms’Complementary Uses ofAppropriability Mechanisms in Technological Innovation. Ph.D.Dissertation, University of California at Berkeley.

rossman, G., Helpman, E., 1991. Innovation and Growth in the GlobalEconomy. MIT Press, Cambridge, MA.

all, B., Ham Ziedonis, R., 2001. The patent paradox revisited: an
empirical study of patenting in the U.S. semiconductor industry,1979–1995, Rand Journal of Economics 32, 101–128.
all, B.H., Harhoff, D., 2004. Post-grant reviews in the U.S. patentsystem—design choices and expected impact. Berkeley Technol-ogy Law Journal 19 (1), 1–29.

icy 36 (2007) 134–154 153

Heath, C., 2000. The enforcement of patent rights in Japan. Inter-national Review of Intellectual Property and Copyright Law 31,749–770.

Heath, C., Henkel, J., Reitzig, M., 2005. Patent Indemnifi-cation Rules—and International Legal Comparison. Munich/Copenhagen, Mimeo.

Henkel, J., von Hippel, E., 2003. Welfare Aspects of User Innovation,Working Paper, MIT Sloan School of Management, Cambridge,MA.

Kash, D., Kingston, W., 2001. Patents in a world of complex technolo-gies. Science and Public Policy 28 (1), 11–22.

Klemperer, P., 1990. How broad should the scope of patent protectionbe? RAND Journal of Economics 21, 113–130.

Kortum, S., Lerner, J., 1999. What is behind the recent surge in patent-ing? Research Policy 28 (1), 1–22.

Lanjouw, J.O., Lerner, J., 2001. Tilting the table? The predatory useof preliminary injunctions. The Journal of Law and EconomicsXLTV, 573–603.

Lanjouw, J.O., 1994. Economic consequences of a changing liti-gation environment: the case of patents. NBER Working Paper4835.

Lanjouw, J.O., Shankerman, M., 1997. Stylized facts of patent lit-igation: value, scope and ownership, N NBER Working Paper6297.

Lemley, M., 2000. Rational Ignorance at the Patent Office, BerkeleyProgram in Law and Economics Working Paper Series, 19.

Lerner, J., 1994. The importance of patent scope: an empirical analysis.Rand Journal of Economics 25 (2), 319–333.

Lowe, R., Ziedonis, A., 2006. Overoptimism and the performanceof entrepreneurial firms. Management Science 52 (2), 173–186.

Maloney, D., 2000. The Enforcement of patent rights in the UnitedStates. International Review of Intellectual Property and CopyrightLaw 31, 723–748.

Marshall, H., 2000. The enforcement of patent rights in Germany.International Review of Industrial Property and Copyright Law31, 646–676.

Macdonald, S., 2004. When means becomes ends: considering theimpact of patent strategy on innovation. Information Economicsand Policy 16, 135–158.

Merges, R.P., Nelson, R.R., 1990. On the complex economics of patentscope. Columbia Law Review 90, 839–916.

Mowery, D., Nelson, R., Sampat, B., Ziedonis, A., 2001. The growthof patenting and licensing by U.S. universities: an assessment ofthe effects of the Bayh-Dole act of 1980. Research Policy 30,99–119.

Nelson, R., 1959. The simple economics of basic research. Journal ofPolitical Economy 67, 297–306.

Nordhaus, W.D., 1969. Invention, Growth, and Welfare: A TheoreticalTreatment of Technological Change. MIT Press, Cambridge, MA.

Petit, L., 2000. The enforcement of patent rights in France. Inter-national Review of Intellectual Property and Copyright Law 31,669–691.

Quillen, C.D., Webster, O., 2000. Continuing Patent Applications andPerformance of the U.S. Patent Office, Mimeo.

Reitzig, M., 2004. The private values of “Thickets” and “Fences”—towards and updated picture of the use of patents across indus-
tries. Economics of Innovation and New Technology 13 (5), 457–476.
Schankerman, M., Scotchmer, S., 2001. Damages and injunctions in theprotection of intellectual property. RAND Journal of Economics32, 199–200.

http://www.myoutbox.net/posa78n.htm

rch Pol
Schmookler, J., 1966. Invention and Economic Growth. Harvard Uni-
versity Press, Cambridge, MA.
Singh, J., 2005. Collaborative networks as determinants of knowledgediffusion patterns. Management Science 51 (5), 756–770.

Sorenson, O., Singh, J., Fleming, L., 2005. Science, Social Networks,and Spillovers. Working Paper, London Business School.

icy 36 (2007) 134–154

Teece, D.J., 1986. Profiting from technological innovation: Implica-
tions for integration, collaboration, licensing and public policy.Research Policy 15, 285–305.
Ziedonis, R., 2004. Don’t fence me in: fragmented markets for tech-nology and the patent acquisition strategy of firms. ManagementScience 50 (6), 804–820.

on sharks, trolls, and their patent prey—unrealistic...

Documents