`
`
`
`
`
`
`
`
`M NPEE DISP
`PUTES
`
`
`
`Boston Univversity Schoool of Law Laaw & Econommics Researrch Paper Noo. 12-34
`
`
`
`Berkmman Center for Interneet and Sociiety, Harvarrd Universitty
`
`
`
`
`
`
`
`This paperr can be doownloaded wwithout chaarge at:
`
`
`
`
`
`THE DIRRECT CCOSTS
`
`S FROM
`
`
`
`
`
`
`99 COORNELL L. REV. (forthccoming 20144)
`
`
`
`
`
`
`
`
`(JJune 25, 20112, revised Juuly 2013)
`
`
`
`Jamees Bess
`en
`
`
`
`Booston Univeersity Schoool of Law
`
`
`
`
`
`
`
`Michaeel J. Meuurer
`
`
`
`Booston Univeersity Schoool of Law
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`http://wwww.bu.edu/law/faculty//scholarshipp/workingpaapers/20122.html
`
`
`
`
`
`
`
`

`

`The Direct Costs from NPE Disputes
`
`
`
`99 CORNELL L. REV. (forthcoming 2014)
`
`
`
`
`
`By James Bessen and Michael J. Meurer**
`
`
`
`Abstract:
`
`In the past, “non-practicing entities” (NPEs), popularly known as “patent trolls,” have
`helped small inventors profit from their inventions. Is this true today or, given the
`unprecedented levels of NPE litigation, do NPEs reduce innovation incentives? Using a
`survey of defendants and a database of litigation, this paper estimates the direct costs to
`defendants arising from NPE patent assertions. We estimate that firms accrued $29
`billion of direct costs in 2011. Although large firms accrued over half of direct costs,
`most of the defendants were small or medium-sized firms. Moreover, an examination of
`publicly listed NPEs indicates that little of the direct costs represents a transfer to small
`inventors.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
` Lecturer, Boston University School of Law; Faculty Associate, Berkman Center for Internet and Society,
`Harvard University.
`** Abraham and Lillian Benton Scholar and Professor of Law, Boston University School of Law. We thank
`David Anderson and the staff at RPX for invaluable assistance. We thank Brian Love, Michael Risch, and
`Catherine Tucker for helpful comments, and Cong Yao, supported by funds from the Coalition for Patent
`Fairness, for able research.
`
`

`

`1. Introduction
`
`The American patent system has experienced an explosion of patent litigation
`
`over the past decade initiated by parties called Non-Practicing Entities (NPEs).1 The term
`
`Non-Practicing Entity identifies parties who own and sometimes assert patents, but do
`
`not practice technology covered by their patents.2 Commentators agree that there has
`
`been an explosion of NPE patent litigation, and that NPE lawsuits differ in important
`
`ways from other patent lawsuits,3 but they disagree in their normative assessments of this
`
`phenomenon. We believe that this explosion is troubling, and herein we present evidence
`
`that NPE litigation imposes substantial direct costs on high-tech innovators with little
`
`apparent offsetting benefit to inventors or innovators4 from assertion of NPE patents.
`
`
`1 James Bessen, Jennifer Ford & Michael J. Meurer, The Private and Social Costs of Patent Trolls,
`Regulation, 26 (Winter 2011-2012); Patent Freedom. 2012. “Litigations Over Time,” available at:
`https://www.patentfreedom.com/about-npes/litigations/; Feldman, Robin, Ewing, Thomas & Jeruss, Sara,
`The AIA 500 Expanded: Effects of Patent Monetization Entities, UC Hastings Research Paper No. 45, 7
`(2013) available at SSRN: http://ssrn.com/abstract=2247195 (patent monetization entities filed 58.7% of
`the patent lawsuits in 2012)
`2 The troll label is applied to NPEs that behave opportunistically or cause social harm. But see, Highland
`Plastics, Inc. v. Sorensen Research and Development Trust, CV 11-02246 (C.D.Ca. Aug. 17, 2011)
`(denying motion to strike “patent troll” from the complaint, because it “is a term commonly used and
`understood in patent litigation and is not so pejorative as to make its use improper)
`http://www.iplawalert.com/uploads/file/Highland%20Plastics%20v%20Sorensen%20Rsrch.pdf. Colleen
`Chien coined the term Patent Assertion Entities (PAEs) to specifically identify NPEs who assert patents
`rather than play some other intermediary role in the market for patent rights or the market for technology.
`Colleen Chien, Patent Assertion Entities, Presentation to the Dec 10, 2012 DOJ/FTC Hearing on PAEs,
`http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2187314.
`3 John R. Allison, Mark A. Lemley & Joshua Walker, Extreme Value or Trolls on Top? Evidence From the
`Most-Litigated Patents, 158 U. Penn. L. Rev.1, 12-15 (2009); John R. Allison, Mark A. Lemley & Joshua
`Walker, Patent Quality and Settlement Among Repeat Patent Litigants, 99 Geo. L. J. 677, 686-689 (2010);
`Bessen et al, supra note 1 at 29; Feldman et al. supra note 1, at 8 (“Of the 10 parties who filed the
`greatest number of patent litigations in the years we studied, all were patent monetization
`entities.”)
`4 We use the term inventor to refer to the creator of a new technical idea that may be eligible for patent
`protection. We use the term innovator to refer to a party who develops technical ideas into new technology
`with commercial value.
`
`
`
`2
`
`

`

`In this Article we present results from a unique survey of firms targeted by NPE
`
`patent assertions.5 We augment the survey results with information derived from a
`
`comprehensive database of NPE litigation, and information derived from financial
`
`disclosures by publicly traded NPEs. We find that: (1) the estimated direct accrued costs
`
`of NPE patent assertions total $29 billion in 2011; (2) much of this burden falls on small
`
`and medium-sized companies; (3) publicly-traded NPEs likely cost small and medium-
`
`sized firms more money than these NPEs transfer to inventors; and (4) the distribution of
`
`costs imposed by NPEs is highly skewed, probably because NPEs pursue a range of
`
`different business strategies.
`
`The survey we will describe is unique in three ways. First, it includes defendant
`
`companies that are privately held, including small firms. Second, it reveals information
`
`about costs associated with cases in which NPE patents are asserted but which are
`
`resolved before a lawsuit is filed. Finally, it provides aggregated information about NPE
`
`patent license fees. These kinds of information have not been available, in part, because
`
`the terms of patent licenses are often secret, and in part because previous surveys have
`
`simply not asked about assertions that did not advance to the filing of lawsuits.6 The costs
`
`disclosed by this survey are significant and should play a prominent role in policy debates
`
`about the treatment of NPE patent lawsuits.
`
`Our survey results are largely consistent with the only other study of NPE
`
`litigation costs, a study we completed recently with co-author Jennifer Ford.7 In contrast
`
`to the $29 billion annual cost figure estimated in this Article, we previously estimated the
`
`
`5 The survey was conducted by RPX, a firm that helps companies manage risk from exposure to patent
`litigation. The Coalition for Patent Fairness paid RPX to defray part of the expense of conducting this
`survey.
`6 Mark A. Lemley & Nathan Myhrvol, How to Make a Patent Market, 36 Hofstra L. Rev. 257, 258-59 (2008).
`7 Bessen et al, supra note 1.
`
`
`
`3
`
`

`

`annual cost of NPE litigation to publicly traded American firms to be about $80 billion.8
`
`The previous analysis used a slightly different data set, a very different empirical
`
`approach and a different concept of “cost.” Rather than surveying defendants and asking
`
`them to report costs, we observed the stock market reaction to the filing of an NPE
`
`lawsuit against a defendant firm. We estimated litigation cost by analyzing stock price
`
`movements associated with lawsuit filing.
`
`We are not surprised that the survey generated lower costs than the stock market
`
`event study, because the survey measures only direct costs from NPE patent assertions
`
`while the earlier study measured total costs. Direct costs include the cost of outside legal
`
`services, licenses fees, and other direct costs incurred in response to NPE litigation risk.
`
`Indirect costs captured by our event study methodology include the opportunity costs of
`
`the effort exerted by legal, managerial, engineering, and scientific personnel inside the
`
`firm, and other business disruption costs such as loss of goodwill, loss of market share, or
`
`disruption of innovative activities.
`
`This new study also complements our earlier study by providing information on
`
`companies that are not publicly listed, including small companies. This information helps
`
`reveal the extent to which NPEs help small and medium firms realize profits from their
`
`innovations and the extent to which small and medium firms, to the contrary, incur costs
`
`as the targets of NPEs.
`
`NPEs are individuals and firms who own patents but do not directly use their
`
`patented technology to produce goods or services, instead they assert them against
`
`companies that do produce goods and services. In the past, some NPEs have played a
`
`valuable role in bringing innovations from small inventors to market. Some inventors
`
`8 Id. at 31.
`
`
`
`4
`
`

`

`lack the resources and expertise needed to successfully license their technologies or, if
`
`necessary, to enforce their patents. NPEs provide a way for these inventors to earn rents
`
`that they might not otherwise realize, thus providing them with greater incentives to
`
`innovate.9 But in the past, also, some NPEs have used patents opportunistically. During
`
`the late nineteenth century, “patent sharks” were widely seen as extracting money from
`
`innocent individual farmers and railroad companies.10
`
`However, while NPEs have been around for a long time, over the last few years,
`
`NPE litigation has reached a wholly unprecedented scale and scope.11 In 2011, 2,150
`
`unique companies were forced to mount 5,842 defenses in lawsuits initiated by the
`
`actions of NPEs.12 Moreover, the number of defenses has been growing rapidly (see
`
`Figure 1). Part of this growth has been fueled by new sources of funding and new
`
`business models.13
`
`2. Literature Review
`
`Large scale NPE patent litigation is a recent development, thus the empirical
`
`literature is limited, but growing rapidly. Our NPE lawsuit event study is the most closely
`
`
`9 See Ashish Arora, Patents, licensing, and market structure in the chemical industry, 26 Research Policy
`391, 395-97 (1997); see generally, Naomi R. Lamoreaux and Kenneth L. Sokoloff, Inventors, Firms, and
`the Market for Technology in the Late Nineteenth and Early Twentieth Centuries, In Learning by Doing in
`Markets, Firms, and Countries, edited by Naomi R. Lamroeaux, Daniel M. G. Raff, and Peter Temin.
`(1999).
`10 Gerard N. Magliocca, “Blackberries and Barnyards: Patent Trolls and the Perils of Innovation,” 82 Notre
`Dame L. Rev. 1809, 1829, 1832 (2007).
`11Patent Freedom. 2012. “Litigations Over Time,” available at: https://www.patentfreedom.com/about-
`npes/litigations/; RPX. 2012. “US Litigations Involving NPEs,” available at:
`http://www.rpxcorp.com/index.cfm?pageid=45.
`12 Id. These figures come from the RPX database described below. About 4% of these defenses were
`mounted as declaratory actions rather than infringement suits; these were nevertheless initiated by the
`NPEs. The figure for 2011 reflects, to some extent, an effort by NPEs to initiate litigation before the
`America Invents Act took effect, restricting multi-party lawsuits. Nevertheless, the trend shown in Figure 1
`illustrates rapid growth before 2011.
`13Andrei Hagiu & David B. Yoffie, The New Patent Intermediaries: Platforms, Defensive Aggregators, and
`Super-Aggregators, 27 J. Econ. Perspectives, 45, 52 (2013); Executive Office of the President, Patent
`Assertion and U.S. Innovation, 5-6, June 2013.
`
`
`
`5
`
`

`

`related piece of earlier research; in it we found that the annual wealth lost from NPE
`
`lawsuits was about $80 billion for publicly traded U.S. firms.14 In theory, this cost could
`
`be composed mostly of transfers in the form of royalty payments to NPEs. Indeed, a
`
`number of papers argue that NPEs play a socially valuable role by enabling small
`
`inventors to realize greater profits from their inventions.15 These papers, however, do not
`
`provide empirical evidence to support that assertion.
`
`Our 2011 paper rejected that possibility based on the evidence available to us; we
`
`concluded that much of the cost born by technology companies as they defend against
`
`NPE lawsuits is a social loss and not a mere transfer. The survey results we describe
`
`below provide strong additional support for our view that much of cost imposed on
`
`defendants is a social loss. In particular, the current study finds that NPEs impose costs
`
`not only on large technology companies, but also on very many small and medium firms,
`
`making it even less likely that innovative start-ups are net beneficiaries of NPE activity.
`
`One other researcher has quantified the costs to defendants from NPE litigation.
`
`Catherine Tucker examines the effect of a lawsuit by an NPE (Acacia) against several
`
`firms that make medical imaging software.16 She compares the impact of the lawsuit on
`
`sales of both medical imaging and text-based medical software produced by the targeted
`
`firms. She also compares the sales by the targeted firms to the sales of medical imaging
`
`software made by other firms in the industry who were not targeted with a lawsuit. She
`
`14 Bessen et al, supra note 1 at 31.
`15 See generally, Spencer Hosie, Patent Trolls and the New Tort Reform: A Practitioner's Perspective, 4 J.
`L. & Policy Info. Soc. 75 (2008); Sannu K. Shrestha, Trolls or Market-makers? An empirical analysis of
`nonpracticing entities, 110 Colum. L. Rev. 114 (2011); James F. McDonough, III, The Myth of the Patent
`Troll: An Alternative View of the Function of Patent Dealers in an Idea Economy, 56 Emory L. J. 189
`(2006); Marc Morgan, Stop Looking Under the Bridge for Imaginary Creatures: A Comment Examining
`Who Really Deserves The Title Patent Troll, 17 Fed. Cir. Bar J. 165 (2008); Nathan Myhrvold, The Big
`Idea, Harvard Business Review, March 2010, pp. 40-50.
`16 Catherine Tucker, Patent Trolls and Technology Diffusion, December 24, 2011, available at:
`http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1976593.
`
`
`
`6
`
`

`

`finds that sales of medical imaging software declined by one-third for targeted firms. She
`
`attributes the sales decline to a “lack of incremental product innovation during the period
`
`of litigation,” and she conjectures that incremental innovation was deterred by concerns it
`
`would create additional risks in the on-going litigation.17
`
`Two other strands of previous research are especially relevant to this project.
`
`First, earlier work has quantified legal fees associated with patent litigation. We collected
`
`data about legal fees that were made public in court decisions concerning fee-shifting in
`
`patent cases.18 Also, the American Intellectual Property Law Association conducts a
`
`biannual survey of its members and includes questions about fees in patent lawsuits.19
`
`The sources are helpful and we report some of their findings below, but they do not
`
`contain information about NPE litigation in particular, and they do not contain
`
`information about assertions that never reach the filing of a lawsuit.
`
`
`
`A better-developed strand of literature reports various characteristics of NPE
`
`litigation.20 While not measuring costs, these studies do shed light on the question of
`
`whether the private losses to firms targeted by NPE patent assertions also tend to be
`
`social losses. The answer appears to be yes. NPE patent litigation has all the hallmarks of
`
`patent notice failure that distorts the patent system and makes it impede technological
`
`
`17 Id. at 1.
`18 James Bessen & Michael J. Meurer, The Private Costs of Patent Litigation, 9 J. L. Econ. & Policy, 59,
`80-81 (2013).
`19 American Intellectual Property Law Association, AIPLA Report of the Economic Survey (2011).
`20 This includes Allison et al. (2010), Chien (2009), Brian J. Love, An Empirical Study of Patent Litigation
`Timing: Could a Patent Term Reduction Decimate Trolls Without Harming Innovators? Santa Clara Law
`School Working paper (2011); Michael Risch, Patent Troll Myths, 42 Seton Hall L. Rev. 457 (2012):
`Schwartz (2012), and Tucker (2012). Other studies have looked at the characteristics of NPE patents
`asserted in lawsuits, including Sannu K. Shrestha, Trolls or Market-makers? An empirical analysis of
`nonpracticing entities, 110 Colum. L. Rev. 114 (2011).), Allison et al. (2009), and Timo Fischer & Joachim
`Henkel, Patent Trolls on Markets for Technology: An Empirical Analysis of Trolls’ Patent Acquisitions,
`working paper (2011).
`
`
`
`7
`
`

`

`progress.21 In Patent Failure we show that the U.S. patent system works well for
`
`chemical and pharmaceutical inventions because the system provides clear notice to the
`
`world of the scope and existence of patent-based property rights.22 For most other
`
`inventions, especially software and business methods, notice failure means that
`
`innovative firms are targeted in patent infringement suits through no fault of their own.23
`
`
`
`Notice failure is likely for NPE lawsuits. Sixty-two percent of the time they
`
`feature software patents24 which are notoriously difficult to interpret. Allison, Lemley,
`
`and Walker study patents litigated multiple times and usually asserted by NPEs; they find
`
`that software patents account for 94% of such lawsuits.25 The patents asserted in NPE
`
`lawsuits are often subject to lengthy prosecutions which delays public access to information about
`
`patent claims.26 Rather than transferring technology and aiding R&D it appears that NPEs
`
`usually arrive on the scene after the targeted innovator has already commercialized some
`
`new technology.27
`
`3. Data
`
`a. Survey
`
`Between February and April 2012, RPX invited about 250 companies to
`
`participate in a survey of their NPE-related costs. The pool of invitees included RPX
`
`clients and non-client companies with whom RPX has relationships. Most invitees were
`
`21 Bessen et al, supra note 1 at 34; Executive Office of the President, supra note 12, at 8; Brian T. Yeh, An
`Overview of the “Patent Trolls” Debate, Congressional Research Service, 9 (April 16, 2013).
`22 James Bessen & Michael J. Meurer, Patent Failure: How judges, bureaucrats and lawyers put innovators
`at risk 46-72 (2008).
`23 Id. at 47-51.
`24 Bessen et al, supra note 1 at 29.
`25 Alison et al., Most Litigated, supra note 3, at 695-96 (74.1% of most litigated patents and 93.7% of the
`assertions).
`26 Allison et al., Extreme Value, supra note 3, at 13; Love, supra note 20 at 21; Risch, supra note 20 at 490.
`27 Federal Trade Commission, The Evolving IP Marketplace: Aligning patent notice and remedies with
`competition 75-80 (2011).
`
`
`
`8
`
`

`

`technology companies, but certain non-technology companies with NPE exposure were
`
`also invited (for example retailers with e-commerce exposure). Participants provided
`
`information to the extent that doing so was consistent with their obligations to third
`
`parties. The information was aggregated and anonymized such that individual data was
`
`not disclosed.28
`
`Participants filled out a standardized Excel template that included a range of
`
`questions about their NPE-related costs. The instructions for the template asked that
`
`participants include certain statistics estimating all of their direct (external spend) NPE-
`
`related costs from 2005-2011. An NPE was defined to include patent assertion entities
`
`and other parties using the same definition as the NPE Lawsuit Database (see below). A
`
`list of each participant’s NPE litigations from that database was provided to ensure
`
`alignment between the survey response and database. Templates were submitted by
`
`email or directly into a secure online data room. To the extent possible, an RPX study
`
`team reviewed the submission for quality and completeness. If needed, the company was
`
`asked certain follow-up questions. Finally, RPX aggregated the submitted data within a
`
`secure computing environment. The resulting dataset forms the basis of the data tables
`
`provided in this document.
`
`Of the 250 companies invited to participate, 82 provided data on lawsuits and of
`
`these, 46 also provided data on non-litigation patent assertions and related costs.
`
`b. NPE Lawsuit Database
`
`In addition to the survey, we also used a comprehensive database of NPE
`
`litigation developed by RPX. These NPE litigation statistics are based on cases coded
`
`
`28 Although RPX provided data for this study, RPX did not exercise control over the substance of our text.
`
`
`
`9
`
`

`

`“830 Patent” in the PACER database which is maintained by the Administrative Office of
`
`the U.S. Courts.29 In case counts, RPX excludes misfiles, non-patent, false marking and
`
`other non-core patent infringement cases. When a case is transferred, RPX counts it as
`
`one case and allocates it to the venue to which it was transferred. When several cases are
`
`consolidated into one, RPX counts it as one case but with multiple defendants. When a
`
`case is severed RPX counts it as separate cases. In defendant counts, RPX rolls up
`
`operating company subsidiaries into a parent entity (Samsung Group and Samsung
`
`Electronics count as one defendant).30
`
`RPX defines NPEs to include patent assertion entities, individual inventors,
`
`universities, and non-competing entities (operating companies asserting patents well
`
`outside the area in which they make products and compete). RPX identifies NPEs
`
`through a manual review process. In this review process, RPX reads patent complaints
`
`found in PACER and checks information in the complaint against its NPE database. RPX
`
`also checks its database of plaintiff counsel, searches public filings and performs web
`
`research. Some of the factors they consider when determining whether a company is an
`
`NPE (or more specifically a patent assertion entity “PAE”) include: Is the entity the same
`
`as or share a substantial financial link with a known PAE? Is there any evidence that the
`
`company sells a product or offers a service? Does the entity webpage prominently
`
`mention technology, licensing, and patents; and not offer a product or sales? Does the
`
`complaint indicate whether the entity has a product in-market or in-development that is
`
`being harmed by infringement? Are the lawyers involved known to specialize in
`
`
`29 This database does not include patent disputes before the International Trade Commission.
`30 Declaratory actions are included in case counts unless otherwise noted.
`
`
`
`10
`
`

`

`representing NPEs? Is this entity known as an NPE or as an established operating
`
`company?31
`
`This definition of NPE is broader than some other definitions. There is no
`
`consensus among researchers on the proper definition of NPE. Schwartz and Kesan have
`
`criticized our reliance on a broad definition because it reaches plaintiffs like universities
`
`who are more meritorious in some sense.32 They argue that because the database
`
`includes lawsuits filed by universities, and other supposedly meritorious plaintiffs, it
`
`overstates the costs incurred by “bad” trolls.33 It is surely difficult to attempt to
`
`distinguish “good” NPEs from “bad”—some people argue that universities sometimes are
`
`bad players who occasionally abuse overly broad patents. But the difficulty of divining
`
`the true nature of NPEs does little to distort our conclusions for two reasons.
`
`First, relatively little of the patent litigation we study comes from universities—
`
`only about one percent of the NPE lawsuits.34 Instead, the lawsuits in the RPX database
`
`were overwhelmingly filed by “patent assertion entities.” The RPX database, in fact,
`
`closely matches other efforts to categorize litigants. The counts of lawsuits are very
`
`similar to those compiled by Patent Freedom. Also, Colleen Chien checked the RPX
`
`database against her own categorization of 1,000 lawsuits and found little difference.35
`
`So our definition of NPE is hardly “unconventional” as Schwartz and Kesan claim.
`
`
`31 There are a range of views among scholars and policy-makers about the appropriate definition of NPE,
`and different analysts are likely to assembly different NPE litigation databases. Based on our experience
`researching patent litigation, we believe that the RPX database yields statistics that are consistent with
`information about NPE patent litigation from other sources.
`32 David L. Schwartz & Jay P. Kesan, Analyzing the Role of Non-Practicing Entities in the Patent System,
`99 Cornell L. Rev. x (2014).
`33 Id. at x.
`34 Colleen Chien, Patent Trolls by the Numbers, Patently-O Patent Blog,
`http://www.patentlyo.com/patent/2013/03/chien-patent-trolls.html
`35 Id.
`
`
`
`11
`
`

`

`Moreover, changes to our database, such as excluding universities, are likely to have only
`
`a small impact on our aggregate estimate of direct costs.
`
`Second, although universities perform research that is extremely valuable to
`
`society and although most university licensing is done in a socially efficient manner,
`
`when universities engage in patent litigation, especially against defendants who have
`
`inadvertently infringed, they nevertheless create social costs. The problem of cost that we
`
`identify is not based on the identities, motives, or other activities of the NPEs, but is
`
`based, instead, simply on the excessive litigation the NPEs create. From this perspective,
`
`we are no less sanguine about excessive litigation among practicing companies, a point
`
`we highlight in our book, Patent Failure.36 And so our estimate of $29 billion implies
`
`socially wasteful litigation-related expenditures and reduced innovation incentives even if
`
`it includes university litigation. We discuss this topic more in Section 6.
`
`c. Sample characteristics
`
`Table 1 compares characteristics of the survey sample with RPX’s database of
`
`NPE lawsuit defenses. Data for the survey are on the left while data for the entire
`
`database of NPE defenses are on the right. The 82 surveyed companies collectively
`
`mounted 1,184 defenses in NPE lawsuits beginning between 2005 and 2011. Of these,
`
`784, or 66%, were resolved by adjudication or settlement and did not involve
`
`indemnification or other factors that cause costs to be atypical.
`
`Note that a possible truncation bias arises because so many lawsuits were
`
`unresolved at the time of the survey. Because lengthier disputes tend to be more costly, at
`
`least as far as legal costs are concerned, and because the number of lawsuit filings has
`
`
`36 Bessen and Meurer, supra note 22 at 120-146.
`
`
`
`12
`
`

`

`risen sharply in recent years, cost estimates based only on resolved lawsuits might be
`
`understated.37
`
`We divided the companies into sub-categories based on their revenue in the most
`
`recent year reported (small/medium at under $1 billion or large at over $1 billion) and
`
`whether they were in the broad software industry, including e-commerce and finance, or
`
`instead in a hardware industry (everything else).38 The latter distinction might be
`
`significant because most hardware industries involve greater sunk capital costs than do
`
`software industries or finance and for this reason hardware industries may be more at risk
`
`of hold up.39
`
`The right panel shows that small and medium firms dominate the universe of NPE
`
`defendants. Small and medium-sized companies make up 90% of the defendant firms,
`
`mounting 59% of the defenses. Firms making less than $100 million in revenue account
`
`for 82% of the defendants and 50% of the defenses.40
`
`As the Table shows, our survey sample consists of companies that are larger,
`
`more likely to be public and which experience relatively more lawsuits than the average
`
`NPE defendant firm. In the rows that control for size and industry sector, survey firms
`
`appear to experience about twice as many lawsuits as companies in the comprehensive
`
`database. This is not surprising, however, it raises the possibility that our sample might
`
`be unrepresentative of the broader population, possibly experiencing costs that are greater
`
`
`37 Jay P. Kesan, & Gwendolyn G. Ball, How Are Patent Cases Resolved? An Empirical Examination of the
`Adjudication and Settlement of Patent Disputes, U. Illinois Law & Economics Research Paper No. LE05-
`027. (2005).
`38 To preserve data confidentiality, statistical analysis was performed by RPX personnel working under our
`direction.
`39 Readers should be mindful of the distinction between the industry of the defendant and the technology
`covered by the patent asserted by the NPE. In particular, it is important to recognize that problematic
`software patents are often asserted against hardware manufacturers.
`40 This estimate assumes that firms with unreported revenue have revenue less than $100 million.
`
`
`
`13
`
`

`

`or smaller than those of the universe of sued companies. Below we do some tests to see
`
`whether the survey appears to have unrepresentative costs.
`
`4. Findings
`
`a. Mean and median costs
`
`Table 2 provides estimates of mean legal41 costs, licensing costs and total costs
`
`(the sum of these) with standard errors in parentheses. The last column also shows
`
`median total costs.
`
`Median total costs per litigation defense fall roughly around half a million dollars,
`
`smaller for small and medium firms, larger for big ones. However, mean total costs are
`
`much higher, nearly eight million dollars for our survey sample. This difference implies
`
`that the distribution of costs is highly skewed, as we explore below. Thus one must be
`
`particularly careful in extending judgments about the costs of litigation based on small
`
`samples. While “typical” costs might only be a few hundred thousand dollars, mean
`
`costs—reflecting the large costs in a small number of very costly lawsuits—are an order
`
`of magnitude higher.
`
`Mean total costs are, not surprisingly, significantly greater for large companies
`
`than small and medium companies. This difference is significant at the 1% level.
`
`The first column reports the legal component of costs. Mean legal costs per
`
`defense range from $420,000 for small/medium companies to $1.52 million for large
`
`companies.
`
`
`41 In the survey estimated legal costs for a particular case were specified as: “Value of any legal costs
`related to this matter through December 31, 2011. Include outside counsel (lead and local), experts,
`discovery costs, prior art searching, jury consultants, graphics, other expenses, and other related costs.
`Include any costs that were ultimately recouped or expected to be recouped by indemnification agreements
`or other mechanisms. Exclude in-house legal costs.”
`
`
`
`14
`
`

`

`Column 2 of Table 2 reports the dollar amounts paid to the plaintiff to settle the
`
`case.42 Column 3 reports the total costs, the sum of legal and settlement costs. The mean
`
`settlement costs for small/medium companies are $1.33 million and for large companies,
`
`$7.27 million. Mean total litigation costs are $1.75 million for small/medium companies
`
`and $8.79 million for large companies.
`
`Legal costs are about a third as large as settlement costs, or about one quarter of
`
`total litigation costs (slightly larger for small/medium companies).43 This implies that a
`
`substantial part of direct costs of NPE litigation is a deadweight loss to society.44
`
`Also note that NPE litigation is relatively more costly to smaller companies. In
`
`our sample, the large company litigation costs were five times as high as small/medium
`
`company litigation costs. But (see Table 1) the mean revenue of large companies in our
`
`sample is nearly seven times the mean revenue of the small and medium companies. This
`
`means that, roughly speaking, smaller companies pay more in direct NPE litigation costs
`
`relative to their size.
`
`Hardware firms have higher costs than software firms. This difference is
`
`significant at the 5% level. Since hardware firms generally have greater sunk costs than
`
`software firms, this difference is consistent with the inter