Fiona Murray has a rare background for an economist: she did her PhD at Harvard in Applied Science. Her work generally revolves around the intersection of commercial science, with its secrecy and licenses, and “open science” following the usual academia playbook. In particular, the three present papers all involve a genetic technique which helps researchers make mice ultra-susceptible to cancer, among other diseases: the Oncomice. A Supreme Court case in 1980, in coordination with the Bayh-Dole Act that encouraged university-private sector partnerships, led the Oncomouse method to be patented with exclusive licensing rights held by DuPont. The company insisted on fairly extensive restrictions if you wanted to use mice whose genes had been altered with the DuPont method: you had to buy the mice at high cost, you could not breed them freely, you had to accept oversight of your scientific studies, etc. Of course, there are many other genetic modification techniques, many other animals related to humans, and many other scientific questions that scientists could choose to work on if the Oncomouse license was too strict. How did this restriction on open science affect science further on down the road?
Let me briefly give the headlines from three papers on this question. First is “Do formal intellectual property rights hinder the free flow of scientific knowledge?” by Murray and Stern, published in 2007 in JEBO. Second, “Of Mice and Academics: Examining the Eff ect of Openness on Innovation” by Murray, Aghion, Dewatripont, Kolev and Stern, a 2009 working paper. Third, “The Oncomouse that Roared,” sole-authored by Murray, in a 2010 issue of the American Journal of Sociology; I, for one, appreciate the good Peter Sellers pun in the paper’s title.
The 2007 paper takes the Oncomouse as anecdote and examines whether patents, in general, on academic techniques lead to less development of further research in that line. In particular, the authors look at publications in a well-known applied science journal Applied Biotechnology. More than half of the papers published in this journal are later associated with a patent by their authors, and nearly everything in the journal plausibly discusses something that could be patented; research in this intersection of basic and applied technology is what Stokes famously called “Pasteur’s Quadrant.” What is great about this paper is the identification: until 2001, patent applications in the US were secret, and patents averaged three years of lag between application and grant. The speed of research in biotech is much faster than in economics, so within three years of publication of an academic article, there is already significant followup research. If patents are an “anti-commons” and the surprise grant of a patent makes researchers wary of following up the associated academic article, then the grant of a patent should lead to a collapse in citations for its associated article. The empirical results suggest that patenting causes between 10 and 15 percent of future expected citations to never occur. The effect is only seen with papers that have at least one public sector or university author, which makes sense given the lack of “surprise” when a private sector firm patents tech that it develops. Though there is no general welfare calculation, I interpret these results are hugely worrying for supporters of Bayh-Dole style policies.
The 2009 paper stresses that, in the authors’ words, “an increase in openness does not simply lead to a temporary increase in follow-on research but instead has an increasing impact over time.” The methodology is very similar to the recent AER on Biological Resource Centers by Furman and Stern which I discussed recently on this site. Essentially, in the late 1990s, the new head of the NIH negotiated simpler, more open, more standard licenses that researchers could use if they wanted to do research using Oncomouse methods (along with one other similar genetic technique). There are also genetic techniques for mice whose patents were never really enforced, such as the University of Utah’s Knock-Out method. The authors gather citations over time for each independent article using mice modified with these technologies. They then look for a change in the (nicely semiparametric) functions of lifetime citations after the NIH negotiations. The increase in openness as a result of the NIH negotiations led to a 20 to 40 percent increase in lifetime citations for treated mouse-articles, an effect that grew over time (science is cumulative!). The effect was mostly seen by new institutions and new researchers using the now “more open” mice. These mouse-articles have mean citations in the hundreds, so the 20 to 40 percent increase is, at first glance, hugely important economically.
Finally, the 2010 AJS provides the extremely interesting anthro- and sociological backstory of the Oncomouse. These types of studies are fun to read as well; phrases like “retired school teacher, turned fancy-mouse breeder, Miss Abbie Lathrop” rarely appear in Econometrica! Murray describes the extremely tight-knit community of mouse genetics laboratories going back to the early 20th century, with traditions of exchange in new strains of mice that, while fairly open, still relied to some extent on non-financial currencies like coauthorship or “gift exchange” in later collaborations. The patenting and harsh license restrictions on the Oncomouse caused a near-riot in the community, and led the NIH to negotiate in the late 90s for more “academic” licensing terms. What is interesting, though, is that this pushback did not lead to a rejection of patenting by other mouse geneticists: indeed, patenting is now common in that field. In the case of defensive patenting to protect an anti-commons, there is no blatant hypocrisy. But beyond this, even when patents are used as currency, the manner in which they are used is still determined by community norms more than it is by the law, and standards of openness and sharing are not totally determined by the legal strength of patents. In many cases today, patented mice are shared freely in the academic community and sold at a price to industry, with follow-on inventions, protected by the patent, more lucrative to private firms wishing to apply academic findings to new products. And indeed, this was exactly the justification of Bayh-Dole. More work needs to be done on whether Bayh-Dole is “good policy”; my guided intuition is still that it is a net negative for social welfare, but the verdict is sufficiently murky that any decent social science on welfare effects would be greatly appreciated by the innovation/science studies community.
http://www.nber.org/~marschke/mice/Papers/murraystern.pdf (2007 JEBO final version)
http://www.economics.harvard.edu/faculty/aghion/files/Of%20Mice%20and%20Academics.pdf (July 2009 Of Mice and Academics working paper)
http://fmurray.scripts.mit.edu/docs/THE_ONCOMOUSE_THAT_ROARED_FINAL.pdf (2010 AJS final working paper)
(To the extent that anyone knows of particularly interesting research in progress on the intersection of open science and the real economy, this is a current research interest of mine and I would love to discuss it – shoot me an email at the address on the “About” page to your left)
A Fine Theorem 