Cohen et al. “Patent Trolls” Study Uses Incomplete Data, Performs Flawed Empirical Tests, and Makes Unsupportable Findings

PDF summary available here

I.   Introduction

A recent draft study about patent licensing companies entitled “Patent Trolls: Evidence from Targeted Firms is making the rounds on Capitol Hill and receiving press coverage. This attention is unfortunate, because the study is deeply flawed and its conclusions cannot and should not be relied upon. If the draft paper is ever published in a peer reviewed journal, it will certainly need to be greatly revised first, with its most notable results likely changing or disappearing.  In sum, the study should receive no credit in policy debates.

The study, by Lauren Cohen, Umit G. Gurun, and Scott Duke Kominers, finds that non-practicing entities (NPEs) are “opportunistic” because they target defendants that (1) are cash-rich (particularly compared to practicing entity patentees), (2) operate in industries that “have nothing to do with the patent” in suit, (3) are staffed by small legal teams, and (4) are busy with numerous non-IP cases. Additionally, the authors conclude that defendants that lose in patent litigation with NPEs on average have marked declines in subsequent R&D expenditures, on the order of $200 million per year. On this basis, the authors suggest “the marginal policy response should be to more carefully limit the power of NPEs.” One of the authors has been circulating this unpublished study to congressional staffers to make the case that NPEs have a large negative effect on US innovation.

II.   Critique of the Study

Professor Ted Sichelman, University of San Diego School of Law, and an expert in empirical studies of patent litigation, critiques the most recent, publicly available version of the Cohen et al. study in detail in his response paper, “Are Patent Trolls ‘Opportunistic?”.[1] He finds that the study’s dataset is incomplete and unrepresentative, its theoretical model is flawed, and its empirical models are unsound. Professor Sichelman concludes that neither their findings nor policy prescriptions are justified. Major weaknesses in the study are as follows:

  • The study’s public firm defendant dataset in current version of paper is incomplete and unrepresentative
  • The study relies on proprietary, unverified coding from PatentFreedom that groups together numerous NPE types (including individuals, R&D shops, and IP holding companies of operating companies), but in making its policy recommendations, the study assumes all NPEs are patent aggregators
  • The study’s finding that NPEs sue cash-rich defendants may simply be driven by the fact that NPEs tend to target software, Internet, and finance-related companies for reasons unrelated to cash holdings, but these companies simply happen to have larger cash-holdings than the average publicly traded company
  • When comparing NPE behavior to that of operating companies, the study improperly includes operating company suits in which the patentees primarily seek injunctions, which are not cash-driven suits
    • Our belief is that NPEs and operating companies alike that primarily seek royalties are likely to seek defendants with enough cash to pay likely damage awards and—like a seller of goods ensuring that a buyer has sufficient cash to pay for those goods—there is nothing “opportunistic” in this behavior
  • NPEs asserting patented technology that is different from the primary industry of the accused infringer are typically not going “after profits unrelated to the patents”
    • For instance, the use of patented computer hardware, software, or technical equipment may occur in any industry and provide a competitive advantage relative to others using non-patented technology
  • The study’s datasets and variables to determine the size of law firm and the number of pending cases are incomplete and flawed
  • The authors’ finding that R&D of accused infringers is differentially affected by a “loss” is based on a very small dataset of “wins” (n=35)

In sum, there is no support for the study’s policy recommendation “to more carefully limit the power of NPEs.” In this regard, we reiterate our view that any plaintiff targeting defendants with enough cash to satisfy a damages judgment is simply ordinary litigation behavior. According to Professor Sichelman, there is “massive risk aversion by many small NPEs” and “large uncertainty in [patent] cases” that may cause any patentee primarily seeking money damages to assert its patents against defendants who can pay their bills.

Finally, in making their policy proposals to restrict NPEs, Cohen et al. rely on the discredited study of Bessen and Meurer (2014) to argue that NPEs do not channel a large percentage of funds received back to inventors. As Schwartz and Kesan (2014) have shown, Bessen and Meurer’s study is inapplicable to most NPEs, because only 12 publicly traded aggregators were examined, and even for those 12 aggregators, Schwartz and Kesan persuasively argue that Bessen and Meurer’s findings are wrong. Indeed, there is ample evidence that many patent aggregators return 50% of net recoveries in litigation or licensing (i.e., after paying for attorneys’ fees and related costs) and that many NPEs are individuals, R&D shops, and other entities that effectively keep 100% of the net returns from recoveries.

As such, the Cohen et al. (2014) study should receive no credit in congressional policy debates. Indeed, another leading academic at a recent conference expressed surprise and dismay that this early-stage study was being circulated by its authors throughout Congress.


[1] The authors presented new material in response to Sichelman’s critique at a recent conference, but as far as we know, they have not made any of it available to the general public. As such, we focus on Sichelman’s critique of the most recent, publicly available version of the study.