Love on University Patents

Brian Love's paper Do University Patents Pay Off?  Evidence from a Survey of University Inventors in Computer Science and Electrical Engineering, 16 Yale J. L. & Tech. __ (forthcoming) is available on ssrn.  Here is the abstract:

Studies of the costs and benefits of university patent ownership have, to date, focused on life sciences technology. Increasingly, however, many of the most lucrative university-owned patents relate to computing and telecommunications, not genes or pharmaceuticals. In 2007, a University of California spin-off named Eolas settled a patent suit with Microsoft for $100 million. In 2010, Cornell University won a $184 million jury verdict against Hewlett-Packard in a case that later settled on confidential terms. And most recently, in 2014, Carnegie Mellon University received a $1.5 billion judgment — one of the largest patent damages awards in history — in an ongoing suit against Marvell Semiconductors. 

As universities shift their focus in the patent arena, so too must those studying tech transfer. Commentators generally agree that the costs and benefits of the patent system vary greatly across industries and many place the high-tech and bio-tech industries at opposite ends of that spectrum. Accordingly, universities would be well advised to reassess the costs and benefits of their own tech transfer programs as they shift from bio-tech to high-tech. 

This Article examines the pros and cons of university patenting in the high-tech field by reporting the findings of a survey of professors at major U.S. universities who teach and research in the areas of electrical engineering and computer science. Among other findings, survey responses suggest that: Patenting high-tech inventions made on university campuses may not be a profitable undertaking, even at those universities best-positioned to profit from tech transfer. Based on the patenting and licensing activities of survey respondents, I estimate that university patent programs earn a negative 3.5% rate of return on high-tech patents. The prospect of obtaining patent rights to the fruits of their research does not motivate university researchers in high-tech fields to conduct more or better research. Eighty-five percent of professors report that patent rights are not among the top four factors motivating their research activities. Moreover, fifty-seven percent of professors report that they do not know how, or if at all, their university shares licensing revenue with inventors. University patent programs may, instead, actually reduce the quantity and quality of university research in high-tech fields by harming professors’ ability to obtain research funding, to collaborate with faculty from other institutions, and to disseminate their work to their colleagues. University patent programs are, at best, a modest benefit to professors seeking to commercialize high-tech academic research. Entrepreneurial professors report that these programs hinder their ability to work as a consultant with companies that show interest in their research, and fewer than half of university spinoff founders report that the ability to patent their research affirmatively helped their commercialization efforts.

As Professor Love explains, he sent out surveys to "2,387 tenured and tenure-track faculty members affiliated with the nation's top twenty ECE and CS departments" and "collected 269 responses . . . .  Respondents in this sample are highly representative of the target population with respect to publicly-observable characteristics" such as "duration of work experience, gender, and rate and quantity of patenting and entrepreneurial activities" (pp. 10, 13-14).  At page 16-18, he further explains:

Approximately two-thirds of respondents reported that they have been named as an inventor or co-inventor on a U.S. patent application resulting from their university research, with a median of four applications per respondent. Of those who had filed an application covering their university research, more than four-fifths reported that at least one of their applications had resulted in an issued U.S. patent, with the median respondent-patentee reporting that his or her research had resulted in three issued patents. Of respondents whose research had been patented, about two-thirds reported licensing at least one of their patents, with the median respondent-licensor reporting that licenses to his or her patents have, to date, earned his or her university a total of $30,000 in royalties.

. . . aggregate licensing revenue is heavily dependent on a small number of licenses that are extreme outliers.  More than seventy percent of the reported royalty total was generated by less than three percent of licensors, and the top nine percent of licensors were responsible for over eighty-five percent of all reported licensing revenue. By contrast, almost twenty-five percent of all respondent licensors reported that their licenses have, at least to date, failed to earn any royalties. Another nine percent of licensors each reported earning $5,000 or less for their universities.

Beginning at page 19, Professor Love sets out his methodology for estimating the cost of prosecuting the relevant applications, based on his analysis of a sample of those applications and data reported by the AIPLA on median costs of prosecuting applications in the U.S.  His conclusion is, as stated above, that the median rate of return on patenting in the fields of electrical engineering and computer science at the top twenties universities in these fields is negative 3.5% (p.27).  As for why universities pursue patents in these fields if the return is likely to be negative, Professor Love suggests, among other things, that they may be "basing patent policy on extreme outliers" (p.42), e.g., cases like those mentioned in the abstract in which very large damages were awarded.

Overall, it appears to me that Professor Love's research is thorough and his methodology fully explained in the text and accompanying appendix.  I don't do empirical work like this myself, and perhaps others who do will find fault with one thing or another or suggest areas for improvement.  But it seems to me from my initial read-through that Professor Love has made a substantial contribution to our understanding of the costs and benefits of university patenting in these particular fields.

Some readers of this blog may have already seen the response to Professor Love's article by Bob Zeidman, which was published last week on the IPWatchdog Blog, so I'll say a few words about Mr. Zeidman's analysis as well.  Mr. Zeidman titles his post "Sloppy, Misleading Yale Paper Challenges University Patenting," and he begins by stating that he attended a conference at Stanford earlier this year (titled "Patent Trolls and Patent Reform") at which "professors from elite universities around the country explained why patents, and those who license or litigate them, had made the United States such a plodding, backward nation that is desperately trying to catch up with progressive countries like China, Russia, and Europe."  (Europe's a country?)  Anyway, Mr. Zeidman asserts that Professor Love's paper was the worst at the conference, "a ridiculous example of how our universities are putting out 'research' that is terribly shoddy, detached from the real world, and simply reinforces generally faulty assumptions about how the world works."  His first bone of contention is that Professor Love "sent out only 2,387 questionnaires to tenured and tenure-track faculty members at select elite universities," "a miniscule .15% of the 1,565,504 professors in the United States as of 2011," citing a U.S. Department of Education Report titled "Employees in Postsecondary Institutions, Fall 2011 and Student Financial Aid, Academic Year 2010-11."  If I'm reading this document correctly, however, it states (table 1, p.4) that there were 1,565,504 "[s]taff whose primary responsibility is instruction, research, and/or public service" at Title IV institutions in the United States in the fall of 2011.  Presumably this figure includes professors from all disciplines (English, philosophy, law, etc.), so it's a little hard to see what the relevance of the number is to the subject Professor Love is studying, which focuses on electrical engineering and computer science in particular.  (If there were 1,565,504 professors of computer science and engineering, that would be one for about every 200 Americans!)  Moreover, Professor Love's focus on professors at the top twenty universities in these fields would seem, if anything, to bias the rate of return upwards, since you'd expect the top twenty universities would gain a disproportionate share of whatever benefits are to be derived from university-derived patents in these fields (a point that Professor Love makes at pp. 10-12).

Moving on, Mr. Zeidman states that "rather than relying on financial data that could have been obtained from university records and finance departments, Professor Love relied on asking simple questions of professors who have no particular access to that information."  Perhaps I'm wrong, but I'm not at all sure that one can simply obtain information on these universities' rates of return on specific types of patents by, say, checking to see how much licensing revenue they derive from all of their patents in a given year.  The University of California 2012 Technology Transfer Annual Report, for example, which Professor Love cites at p.21 n.71, reports technology transfer income and expenses for the University of California system for 2012 but unless I'm missing something it's not broken down by field of technology (although it does list the top-earning inventions at p.20).  Again, Professor Love's focus is on patents in computer science and electrical engineering, not in other fields.  Moreover, as Professor Love explains (p.26 n.81), under U.S. law universities are obligated to share licensing revenue with university inventors. so you'd expect those inventors who are getting substantial royalties to be aware of what they're getting and what it's based on.  On this issue, however, Mr. Zeidman believes that the question Professor Love posed to the respondents--"[A]bout how much total licensing revenue have your university patents earned?"--was ambiguous, since it might be asking how much the university earned or how much the researcher earned.  Looked at in context, however, I'm not persuaded that there is an ambiguity.  In the appendix, Professor Love sets out the questions and the order in which they were presented, as follows:

Have any of your university patents brought in licensing revenue for your
university?
 Yes
 No
 One or more of my university patents was licensed, but those license(s)
never generated any revenue for the university.
 One or more of my university patents has been licensed, but those
license(s) have not YET generated any revenue for the university.
If so, how many of your university patents have been licensed?
___________________________________________
Also, if so, about how much total licensing revenue have your university patents
earned? (A gross approximation is sufficient.)
___________________________________________
According to your university’s policies, are you entitled to a share of the revenue
your patents earn the university?
 Yes
 No
 Don’t know
If yes, do you know (without looking it up) what percent of the revenue you are
entitled to receive?
 Yes
 No
If yes, what percent or other arrangement?

Taken in context, I don't see the ambiguity.

Finally, Mr. Zeidman states that the "most significant, glaring problem" is Professor Love's statement at p.2 that "[F]ifty-seven percent of professors report that they do not know how, or if at all, their university shares licensing revenue with inventors."  He emphasizes the point by stating that "Professor Love has concluded that universities do not earn returns on their investment in patents and at the same time concluded that more than half of all respondents to his survey admitted they do not know anything about what the survey was purporting to study."  But Professor Love addresses this very point at p.31 n.99:

These licensors’ lack of knowledge about revenue sharing arguably calls into question the amounts of overall revenue they reported earlier in the survey. However, as discussed supra in Part II.A, total licensing revenue is heavily driven by a small number of large licenses. The majority of licensors who did not know precisely how royalties are shared also reported that their licensed patents had not earned any royalties to date or, perhaps in an abundance of caution, did not answer the survey questions asking for licensing revenue. The rest reported relatively small sums of revenue that, even if trebled, would raise the estimated rate of return shown in Table 4 by just one-tenth of 1%. 

Anyway, as I said, I'm not an empiricist, and for all I know there may be better ways of trying to estimate what Professor Love is trying to estimate.  Based on his careful explanation of what he has done, however, his analysis certainly seems plausible to me, and Mr. Zeidman's critique strikes me as off the mark.