Today's Date: November 29, 2014 Search | Home | Contact | Services | Patent Attorney | Patent Search | Provisional Patent Application | Patent Application | Software Patent | Confidentiality Agreements

Questionable Science Used to Misguide Patent Policy


Written by Ron Katznelson, Ph.D.
President, Bi-Level Technologies
Posted: October 24, 2013 @ 6:26 pm

Follow Me!
Tell A Friend!

In their article entitled The Private and Social Costs of Patent Trolls,[1] James Bessen, Jennifer Ford, and Michael Meurer present a study on patent litigation involving Non Practicing Entities (NPEs), which they define as firms that do not produce goods but rather acquire patents in order to license them to others.  Bessen et al.’s conclusions are startling.  The loss to defendants involved in NPE patent suits during the last four years “exceeds $83 billion per year, over a quarter of U.S. industrial R&D spending per annum;”[2] and NPE patent litigation constitutes a “very large disincentive to innovation.”[3]  Bessen et al.’s article was prominently featured on the cover of the Winter 2011-2012 issue of Regulation magazine with a cover illustration of oversized humanoids with visible malign intent, armed with clubs, holding up innocent travelers for payment at a bridge, wherein the cover is titled “Patent Trolls – How NPEs harm innovation.”[4]

In my full article “Questionable science will misguide patent policy,” I expose fundamental flaws in the methods that Bessen et al. apply in their studies and explain why their fantastic cost estimates should be dismissed as extremely biased and unreliable, and why their conclusions should be discarded as misleading for patent policy.  An abridged version follows.

Bessen et al.’s stock return event studies on patent litigation

Bessen et al.’s thesis is predicated on “event studies” of lawsuit filings—what happens to an alleged infringer’s stock price around the filing of a patent infringement lawsuit, after taking into account general market trends and random fluctuations of the individual stock.  Without providing any proof, these authors argue that during these “events,” stock value declines that are otherwise unaccounted for by estimated market trends (called “Abnormal Return”), reflect “the costs of lost business, management distraction and diversion of productive resources that might result from the lawsuit, possible payments needed to settle the suit, and the reduction in expectations of profits from future opportunities that are forestalled or foreclosed because of the suit.”[5]

Unfortunately, this event study method suffers from two fatal flaws:

(a) it treats a loss in market capitalization—a loss to the firm’s shareholders —as if it measures a loss to the firm itself; and

(b) even if the change of shareholder wealth were relevant, this method tracks the stock value effects of patent litigation only upon filing of the lawsuit but ignores any subsequent related stock value corrections or gains upon disposition of the lawsuit in an announced settlement or a final verdict.

As to my point (a), my full paper cites studies showing that the change in market capitalization associated with certain kinds of bad news, such as the filing of a patent lawsuit, exceeds, often by an order of magnitude, any informed estimate of the capitalized loss in the firm’s reasonably expected future earnings associated with the bad news.  In any event, as my point (b) states, Bessen et al.’s method fails to properly account for the actual change of shareholder wealth due to the patent litigation.

Bessen et al.’s persistent misuse of stock event studies on patent litigation

Bessen et al. derive the Cumulative Abnormal Return (CAR)—the detected Abnormal Return in the alleged infringer’s stock accumulated during a 5-day period around the lawsuit filing event.  They find ensemble CAR value of -0.32% when averaged over the ensemble of 4,114 NPE suits.[6]  They derive from such stock declines the market capitalization losses, which they call patent litigation losses.

When the Abnormal Return on a stock of a firm accused of patent infringement declines by X percent on the day of filing the lawsuit, for Bessen et al. the stockholders have definitively lost X per cent due to patent litigation even if it had not concluded.  If the stock then rises by Y percent on the day a settlement of the lawsuit is announced, Bessen et al.’s method ignores it, but one may ask whether the firm’s stockholders really lost X percent due to patent litigation?  Alternatively, if the Abnormal Return changes by Z percent on the day a verdict in the lawsuit is announced, have the firm’s stockholders really lost X percent due to patent litigation?  The answer to both of these questions is clearly NO.  Bessen et al.’s method misrepresents the true change in shareholder wealth due to patent litigation by selectively choosing to measure only one component.

Examining only a few lawsuits by the NPEs that Bessen et al. have identified in their Costs of Patent Trolls study help demonstrate the flaw in their method.  Three such NPE cases are analyzed in the Appendix of my full paper in which CAR values were calculated according to the method that Bessen et al. used.  These CAR values are evaluated not only for the announced event of filing the lawsuit, but also for announced settlements or other legal disposition events in these very same cases.

The CAR values obtained in Table 1 in the full paper are negative on filing of the lawsuit in all three cases, ranging from -0.26% to -2.45%.  However, the opposite had happened upon case dispositions.  In one case, upon a verdict of patent invalidity and non-infringement, the detected CAR value of alleged infringer Microsoft was nearly +3%.  In another case, despite a verdict finding the asserted patent valid and infringed, the detected CAR value of alleged infringer Yahoo! was positive at +0.11%, followed by a correction CAR of +1.33% upon a reversal in-part by the appellate court, holding the patent not invalid but not infringed.  In a third case, where filing of the lawsuit against Micron Technology was associated with a detected CAR value was -2.45%, the subsequently detected CAR was more than +9% upon the settlement announcement.  This settlement granted Micron a license to a large patent portfolio including many more patents than those asserted against it.  As seen in the Appendix, the net CAR values in these three examples were significantly more favorable to the shareholders of the alleged infringers than the initial negative CARs upon filing the lawsuits.  This demonstrates that some circumstances may result in net positive CAR associated with the patent litigation.

There is a more fundamental objection to Bessen et al.’s method.  The filing of a lawsuit is only a single step in the bargaining process between the patentee and the firm, at the end of which the parties would account for any transactional costs (including litigation costs) and would exchange rights and valuable considerations.  Contrary to Bessen et. al.’s counterfactual categorization, the firm’s market capitalization changes associated with patent litigation are not “costs” but are part of transfers in the language of economists.  Over 80% of patent lawsuits settle.  We should expect that at least the removal of uncertainty upon settlement would result in favorable stock value corrections for defendants firms.  Indeed, event studies cited in my full paper find generally that the defendant firms’ shareholders benefit from a significant wealth increase when settlements are announced.  The Micron case is by no means unique; in other settled lawsuits, firms obtain a broad license to a full portfolio of patents, which may impart a valuable advantage to the firm over its competitors who have not obtained such license.  Selectively looking only at one event in the patent rights bargaining process is meaningless as an indicator of the actual effects of patent litigation and the resultant change in shareholder wealth.

Questionable science in support of the “patent troll” narrative

The application of Bessen et al.’s stock value event studies to assess the social costs of patent litigation is highly questionable on the additional grounds that it ignores fundamental economic effects of patent enforcement.  Observed negative stock price reaction does not necessarily include positive externalities for third parties when wealth is redistributed to third parties such as competitors of the defendant firm, or other third parties licensed under the asserted patents.

Other important positive externalities of patent enforcement, including by NPEs, may be realized when competitors are encouraged to design-around the asserted patent.  Incentives to design around patents usually materialize only upon patent enforcement lawsuits, but when design-arounds are commercially successful, they may result in substantial increases in social welfare: design-around patents have been documented to spur new manufacturers’ entry to the market, unleash fierce price competition, and reduce deadweight losses of the patentee’s monopoly pricing.[7]  From a dynamic efficiency perspective, the greatest potential social welfare enhancement due to the designs-around appears downstream over many years even in areas other than the patented technology.[8]

Despite the glaring deficiencies of their method discussed above, Bessen et al.’s Costs of Patent Trolls article, with its implausible NPE patent litigation cost estimates of over $80 billion per year was widely publicized.  Support for the Costs of Patent Trolls study was obtained from entities known to have an interest in the direction of the results.[9]  The resultant high “cost” estimates of the article were used extensively by several large technology corporations and organizations who cultivated the patent “troll” narrative to promote legislation that would weaken patent rights, ostensibly to curb purported “abusive patent litigation.”  This year, the Costs of Patent Trolls received recognition as a legitimate source of patent litigation cost information in a publication of the Congressional Research Service.[10]  The study’s results have been adopted as partial justification for Presidential executive actions to curb the “private costs of lost opportunities to commercialize technology” in the White House report entitled “Patent Assertion and U.S. Innovation.”[11]  Unfortunately, none of these Government reports question the flawed method of the Bessen et al. study.

Conclusion

Bessen et al.’s event studies are unreliable.  Nevertheless, this questionable science has been used to frame the debate about patent legislation.  Reliance on unreliable scholarship will undermine the achievement of the economic and innovation goals that the Administration and Congress share.

 


[1] James Bessen, Jennifer Ford, and Michael J. Meurer, The Private and Social Costs of Patent Trolls, Boston University School of Law Working Paper 11-45, (September 19, 2011) (hereinafter “Cost of Patent Trolls”), at http://www.bu.edu/law/faculty/scholarship/workingpapers/Bessen-Ford-Meurer-troll.html ; a version of this paper also published in 34 Regulation 26 (Winter 2011-2012), at http://object.cato.org/sites/cato.org/files/serials/files/regulation/2012/5/v34n4-1.pdf

[2] Id. at 17.

[3] Id. at 21.

[5] Bessen et al. (2011) at 16.

[6] Bessen et al. (2011) at 32.

[7] Ron D. Katznelson and John Howells, “Inventing-around Edison’s incandescent lamp patent: evidence of patents’ role in stimulating downstream development,” The Fifth Annual Searle Center Conference on Innovation and Entrepreneurship, Northwestern University, Chicago (June 2012), available at http://j.mp/Inventing-around-Edison. (Reviewing the social welfare benefits of design-around including pharmaceutical patents; in the incandescent lamp industry, Edison’s competitors remained in the market using their design-around solutions throughout the Edison patent term, from which it may be reasonably inferred that their amortized design-around R&D costs were sufficiently low as they are presumed to have set their prices to make adequate profits at the lower prices).

[8] Id. at Section 4.1 and subsections therein including Appendix B.

[9]  Bessen et al. disclose that they obtained some support for their study from the Patent Freedom and the Coalition for Patent Fairness.

[10] Brian T. Yeh, An Overview of the “Patent Trolls” Debate, Congressional Research Service, R42668 (April 16, 2013) at http://www.fas.org/sgp/crs/misc/R42668.pdf.

[11] Executive Office of the President, Patent Assertion and U.S. Innovation, (June 2013), available at http://www.whitehouse.gov/sites/default/files/docs/patent_report.pdf


11 comments
Leave a comment »

  1. “Bessen et al.’s event studies are unreliable.”

    Ron,

    You’re being charitable regarding Bessen & Company. Their academic diatribe is disingenuous and biased to the core. Frankly, Bessen & Company should be ashamed and held accountable for such poor and flawed scholarship and research (if you can even call it scholarship/research). Unfortunately, Bessen & Company sing the PC “song” when it comes to patents. And thanks for again exposing how misguided and flawed these studies really are.

  2. It might be fun to apply the same methodology to a totally unrelated type of event and reproduce the results.

  3. Kudos once again, Mr. Katznelson. If only objectivity were as prized by our lawmakers.

  4. Ron – Thank you for an excellent piece of work. I am hopeful that, with enough truth, we can stem the tide of the continuing attack upon the American Patent System.

    Best,
    Randy Landreneau

  5. Thank you all for your comments. I posted a revised version of the paper at http://j.mp/Junk-Science , which now includes in the introduction some historical context – quotations from 1917 alleging that the Wright Brothers were “patent trolls.”

  6. While I agree with nearly everything you say here Ron I don’t see what is so new or important about what you’re presenting. Everyone who read the initial work understood the facts you’re now presenting may well be the case. You seem to be saying that while the observations seem negative you’ve really got to watch the stock prices longer to have a good handle on the situation. I’m fairly sure everyone realizes that already, but in the absence of such work, we have to make do with what we have. Perhaps you’re the man to fill that niche, and I look forward to your paper regarding it if you are. Likewise, I think everyone understands that there are likely some positive externalities for third parties and also design arounds. But in view of a lack of research into the economic effects of those things we must, as with the other topic, make do with the information we have. If you’d like to do papers on those subjects I’d be interested to read it.

    If you have an extensive paper on each of these subjects then I’m interested to read all about it. But I don’t take from your abr. article that you get all that deep into either subject in your full article. But I will check it out and hope to be pleasantly surprised.

  7. John Smith: “but in the absence of such work, we have to make do with what we have.”

    John, you apparently misapprehend the scientific method. The content of “what we have” is determined and controlled by the researchers’ design of the experiment – the data they deem necessary to collect in order to draw the inferences. The researchers purport to have inferred the actual costs to shareholders due to patent litigation. In order to draw such an inference, it is the researchers’ burden – not mine – to design their experiment and collect all data necessary for estimating actual costs to shareholders. They have not done that, but made their false inference anyway. The argument that “we have to make do with what we have” cannot justify falsehoods and making it in such circumstances will make you flunk any science course.

    So “in the absence of such work,” the only thing we can “make do with what we have” is to infer the decline in shareholder wealth only upon filing of suit – not the actual litigation cost to shareholders. That was the point of my paper, in case you missed it.

  8. I found John Smith’s post to be borderline offensive. It was almost as if Mr. Smith did not want to understand the clear thrust of the article here.

    That Mr. Smith’s post is so easily dismissed as being off-point is its only ‘saving’ grace.

  9. Ron,

    Excellent analysis on what (as you cogently point out) are flawed and biased conclusions. I wonder if anyone has ever done an analysis on the benefits of a robust patent protection regulatory scheme? Of course, the skeptic in me suspects that analysis (if a benefit was proven up) would be, at best, ignored and, at worst, somehow twisted to further support an erosion of patent holder rights and remedies.

    But, I digress. The purpose of my comment was to compliment on an excellent article so, well done.

  10. Thank you B for your acknowledgements. As to “an analysis on the benefits of a robust patent protection regulatory scheme,” I am not aware of much empirical work on the subject. The related question of what social welfare benefits emerge from the existence of strong patent rights when those are enforced is answered in my paper on Edison’s patent with ample evidence of substantial benefits not only in enhanced competition and price reductions, but also in downstream technological developments ( http://j.mp/Inventing-around-Edison ).

  11. Ron thanks for reminding us all that the volatility of a stock price is subject to all manner of factors.

    The plunge in the value of most stocks in the fall of 2008 would hardly have been based primarily on a given patent suit even if that patent was importantly to the stock companies bottom line.

    Your logic clearly exposes Bessen et al’s stock capitalization metric as severely flawed.