In 2013, Howard Markel published a “perspective” in the New England Journal of Medicine on the Bayh-Dole Act, “Patents, Profits, and the American People–The Bayh-Dole Act of 1980.” In his article, Markel argues that Bayh-Dole should be “recalibrated,” in part to deal with the high price of “lifesaving technologies,” along with problems of sharing, undue emphasis on applied research at the expense of basic research, and conflicts of interests for doctors. “All Americans should be able to share in the bounties of federally funded biomedical research.” It’s difficult to argue with any of this.
What concerns me, however, is how Markel describes Bayh-Dole. It’s not that he doesn’t have citations for his treatment, but rather that he apparently accepts the assertions of those “sources” though they lack any substantive basis. Since the bulk of Markel’s argument consists of retelling a history of Bayh-Dole, we may as well give some context to that history in the interest of providing a sense of perspective.
History is often told from the perspective of the victor. In the case of Bayh-Dole, the victor has been the patent brokers who went up against public policy to ensure that pharmaceutical companies could maintain monopolies over new therapeutic compounds that might be beneficial in fighting disease. But let’s work through Markel’s retelling and point out some of the problems with the victors’ account. The starting point is this statement regarding Bayh-Dole:
the law reversed decades of government policy by allowing scientists, universities, and small businesses to patent and profit from discoveries they made through federally funded research
The government policy that controlled was the Kennedy patent policy (1963), subsequently modestly revised by Nixon. That policy encouraged flexibility within a uniform framework of principles. The government was directed to acquire invention rights when the purpose of the contract was to
(1) create products or methods for commercial use by the general public
(2) explore fields that directly concern public health or welfare
(3) study a field where the government is the principal developer or user
(4) operate a government-owned research facility or coordinate the work of others
In these cases, the contractor would get a non-exclusive license to any invention that was made, and one non-exclusive license meant that the next license, too, would have to be non-exclusive. It’s rather difficult, you see, for the government operating under this policy to grant exclusive licenses to anyone other than the contractor that hosts the inventive work. Thus, non-exclusive methods are baked into the executive branch patent policy from the get-go. Nice.
The conditions in (1) include the public in the U.S. “or abroad”–considering the effect of patenting in non-US jurisdictions where US citizens may be–and also to uses that may be required by federal regulation. Thus, “commercial use” is not merely a matter of profiting from the sale of products, or profiting from the licensing of patents to permit the sale of products, and certainly not from speculating on the possibility that at some future point a patent can be monetized by taxing those that might eventually sell products. Commercial use, here, means any use that involves commerce–so that the use might be entirely “internal” but allow someone (a person, a company) to engage in commerce or to offer a product or service required by regulation.
Conditions in (2) speak directly to work such as searching for new compounds with therapeutic benefits–to prevent, to cure, to mitigate disease, for instance. It is here, at (2), that we will see the collision between federal policy and the pharmaceutical industry. The government, in funding the search for therapeutic compounds–medicinal chemistry in its glory years of the 1950s and 60s–encroaches on territory that the pharmaceutical companies wanted to control with patent monopolies. “Patent medicines” and all that. Thus, when the HEW insisted that inventions in this area made with federal support should be owned by the government and either dedicated or licensed to the public–in keeping with the Kennedy patent policy–pharma was horrified and organized a boycott. How do you think things ended up?
However, a contractor could under the Kennedy patent policy could “in exceptional circumstances” acquire more than a non-exclusive license if it is in the public interest, either when the contract was awarded or after an invention has been identified when the invention is not a direct goal of the contract and greater rights are needed to “call forth risk capital.”
But if the contract is for government procurement and the contractor has technical competence and an established “non-governmental commercial position” in that area, then the contractor should “normally acquire the principal or exclusive rights,” so long as the government gets a non-exclusive license.
If the contractor doesn’t have an established commercial position, however–as is the case with most universities–agencies are instructed to make any determination of ownership after an invention has been made, “in a manner deemed most likely to serve the public interest as expressed by this policy statement.” Especially important is whether the contractor intends to “bring the invention to the point of commercial application”–that is, so that “the invention is being worked and that its benefits are reasonably accessible to the public.” It was the delay in federal agencies deciding on ownership in such cases that created the impetus for Bayh-Dole, but only after other efforts had failed.
Under the Institutional Patent Agreement program, begun in the early 1950s and then suspended, only to be revived in 1968 by the NIH, nonprofit organizations could enter into a master agreement with the NIH (and later also with the NSF) so that determinations of ownership were made upfront, by contract (despite the Kennedy and then Nixon patent policies requiring determinations to be made after an invention was identified). Thus, for the sixty plus nonprofits that had an IPA with the NIH, the contractor could own patents on inventions made with federal support–subject to various requirements to protect the public interest, with the implication being that the patent system alone was not sufficient to so protect the public where the purpose of the research was not government procurement from commercial sources, but rather government subvention–assistance to support work undertaken overtly with a public interest as its goal.
Thus, Bayh-Dole did not “reverse” government policies on patents. That’s part of the nonsense of the faux history of Bayh-Dole. Bayh-Dole made one part of government policy arbitrary–determination of whether the government should require ownership of a given invention when the contractor did not have an established commercial position–and made that arbitrary determination government wide, and then eliminated many of the protections for the public interest that went with that determination. In essence, Bayh-Dole replaced “in a manner most likely to serve the public interest” with “it serves the public interest that institutions that host federally funded research should have a free hand in exploiting for profit any patents on any inventions that may be made in that research.” That is, Bayh-Dole determines “public interest” by rule rather than by judgment, and the rule is that the public interest is served when a university owns an invention made with federal funds. And by implication, the public interest is not served when an invention is owned by its inventors under federal common law, or by the government, except as the garbage bin for otherwise worthless inventions, to keep them out of the hands of their inventors and not in competition with university patent licensing efforts.
The Bayh–Dole Act is beloved by the biotechnology and investment communities. In 2002, the Economist called it “possibly the most inspired piece of legislation to be enacted in America over the past half-century” because it “helped to reverse America’s precipitous slide into industrial irrelevance.”
The Economist reversed itself three years later in an article titled “Bayhing for blood or Doling out cash?” And even if an anonymous writer for The Economist asserts that Bayh-Dole has done something, it would have been really keen if there had been any data to back the claim. But there wasn’t at the time, and still isn’t today, because the reports of use that Bayh-Dole anticipates that federal agencies may request are kept secret. A quip in a British op ed piece doesn’t establish anything as a fact, and if that’s what people are basing their views of public policy on, well, gosh, that’s something. For all that, Senator Long called Bayh-Dole “the worst bill I’ve ever seen”–and that was before the 1984 amendments removed the restrictions on university patent behaviors that got the bill passed in the first place. Oh, yes, as Markel points out, Bayh-Dole was a political favor for Senator Bayh, who had lost his election. Bayh-Dole, the parting gift to the patent broker lobby.
There’s nothing to indicate Bayh-Dole did anything to “reverse America’s precipitous slide”–or that there really was much of slide at all. We might say, then, that The Economist at one point approved of Bayh-Dole and based that approval on an assertion about a change in America’s “industrial relevance.” But there’s no good evidence that Bayh-Dole had led to greater industrial activity in America. For all that, from 1968 to 1978, nonprofits had first dibs on pretty much all NIH-sponsored inventions and what did they do? They “commercialized” 4 inventions out of 96–about 5% of what they touched. How did Bayh-Dole, implementing roughly the same terms (however distorted) that the IPA program used, do anything different? Even now, university commercialization rates are sub 1%. Where are all the manufacturing jobs brought about by Bayh-Dole? Non-existent.
If Bayh-Dole was to have any noticeable effect, it would have to be in non-defense (where contractors could own), non-health care (NIH IPAs allowed contractors to own), non-science (NSF IPAs allowed contractors to own) areas of federally supported research. What does that leave us with? Energy, space, agriculture. Whatever has happened has happened at the same time as Bayh-Dole, or even in spite of Bayh-Dole, but no one has shown that Bayh-Dole has caused much of anything with regard to America’s industrial economics. The burden of demonstration is on those advocating for Bayh-Dole. Support your claims with data. But you can’t. Isn’t. Markel gets this part right:
The law certainly contributed substantially to the increase in patents awarded to universities over the past three decades — from 380 in 1980 to 3088 in 2009.
That’s because prior to Bayh-Dole, most patents weren’t awarded directly to universities, but went to university-affiliated research foundations such as WARF or to Research Corporation. Bayh-Dole didn’t have to upset this balance–but patent brokers used it to break up Research Corporation’s dominance in the field. If anything, university patenting become much less selective, more fragmented, less connected to industry, and much less accomplished.
As for patenting, we need some perspective. University and nonprofit share of total patenting has gone from 0.1% in 1976 to 1.1% in 2016. Meanwhile government owned patenting has decreased as a share of patents overall, from 2.6% in 1976 0.28% in 2016. University and nonprofit patents with a government interest are up 10x relative to total patents and government patents are down 10x relative to total patents. But look at the actual percentage–we’re talking about finding a significant effect from 1% of the patents issued. That in itself would appear to be a difficult task, but keep in mind that universities appear to have “commercialized” only about 0.5% of what they have claimed, so we are talking about finding an effect in 0.5% of 1.1% of all U.S. patents… at most… a very small number of patents… at best about 2,400 patents out of 5,600,000 patents. Are we really ready to make a case that a couple of thousand patents over 40 years has made a significant difference in overall American industrial anything relative to the other 5.6 million patents–even if only 5% of the inventions covered by those patents were ever used?
We don’t even see much effect for Bayh-Dole when it comes to university patenting. There are jumps in university patenting, but much later. And patenting, on its own, is a measure of the amount of money available to patent, and not directly of innovation (which often has no need of patenting, or is a result of avoiding patents). Further, the existence of patents is a direct measure of new technology that is *not* available for broad use. That’s the nature of a patent–the right to exclude all others. Counting patents shows, absent any contrary evidence, what is excluded from general use. Where’s the information that such exclusion by nonprofits, who almost never use what they patent, has resulted in use by others–you know, the “practical application” nodded to by Bayh-Dole. Hasn’t happened the way it has been claimed.
More difficult to confirm is industry’s estimate that between 1996 and 2007, university-based research-licensing agreements contributed $47 billion to $187 billion to the gross domestic product.
So true. The claim is not “industry’s”–it is a claim manufactured by the Association of University Technology Managers a lobbying organization of mostly university licensing professionals (whose dues are paid mostly by their university employers). How do they arrive at their numbers? They take the total reported university licensing revenue and assume that all income represents sales (doesn’t). Then they multiply this revenue by the reciprocal of a typical royalty rate (to create a fictitious “royalty base” which they take to be total sales–cannot possibly be, since much royalty income is in the form of payment for patenting costs, upfront licensing fees, lump sum payments such as at milestones, realized equity, and in settlements). Then they can apply modeled “multipliers” for the effect of spending that fictitious total sales figure in the economy. It’s all balderdash–clever, but entirely made-up.
Oh, the US GDP between 1996 and 2007, about $154 trillion dollars. The “industry” asserts a maximum effect of $187 billion using its fictional model. Let’s see, that amounts to contribution of 0.1%. And that’s the best case that the “industry” can put forward, even with made-up calculations. At $47 billion, the effect rounds to 0. There is effectively no effect from Bayh-Dole on the American economy, on American manufacturing, or on American innovation. Any effects we might want to look for are going to be local, or not measured by such coarse things as the number of patents or the amount of cumulative licensing income. How many inventions are excluded from general use does indicate innovation. How much money people have paid for access does not track use but rather requirements for payment. And when the figures for either amount to a tiny, almost undetectable, portion of the overall patenting and GDP figures, it’s absurd to think that these are arguments for Bayh-Dole’s success.
Indeed, the law’s many critics question how much it has actually benefited the economy (as opposed to individuals and shareholders) and the extent of its social costs.
Help me out–where are these many critics? I’d like to meet even a few of them! But we don’t even have good quantitative data on Bayh-Dole’s social costs. Just reports from the field and reasoning about public purposes and the amazing thirty-five year dearth of anything positive being reported by universities. The universities won’t even break out the effect of patents on subject inventions relative to all inventions that universities claim ownership of (patents on federally supported inventions now constitute about half of all university patents, based on federal interest statements in issued utility patents).
Bayh–Dole’s inspiration was not a perceived need to transform the conduct of research but the economic doldrums of the 1970s.
This is the story put forward by Senator Bayh. But as best I can tell, it was just a cover story. There was no likelihood that a handful of patents a year would make much difference to the American economy. The motivation for Bayh-Dole was the determination to replace the canceled IPA program that had shuttled federally supported research inventions to pharmaceutical companies via university exclusive licenses that preserved (at very low cost) patent monopolies in newly discovered compounds. That, combined with the idea that if universities had financial incentives, they would create their own licensing programs, patent more things, and make money for more university research, which in turn would create more inventions and start a virtuous cycle of money-making, research, inventions, patents, and more money-making.
In 1978, a group of constituents representing Purdue University lobbied Bayh to seek ways of reaping profits from government allocations for university-based research, arguing that although the United States spent billions of dollars annually funding more than half of all academic research and owned 28,000 patents, it had little to show for the investment.
It was the director of the Purdue Research Foundation, and the problem had to do with an invention that been supported by the Department of Energy that the agency was dragging its feet in making a determination of ownership. You can read Joseph Allen’s account here. At the time, the federal government spent around $7 billion on basic research at universities–so, yes, technically “billions” of dollars, but not really. The 28,000 patents is something of a mythology, as checking the references to citations to 28,000 figure never seems to end up citing a report with actual data. But even so, as Markel points out, Rebecca Eisenberg shows that many of these patents were ones that contractors had already passed up. More to the point, only 325 of these patents were from HEW research (here’s Eisenberg):
For example, 325 of the 28,000 patents in the government’s portfolio were from HEW, and seventy-five (or twenty-three percent) of these HEW patents were licensed as of the end of fiscal year 1976.
That is, in the health area, the federal government’s licensing rate for patents was 23%–about the same as the rate claimed for universities’ licensing agents (the research foundations and Research Corporation) working with non-federal inventions. The commercialization rate–somewhat different than the licensing rate–for universities handling federally supported inventions was closer to 5%. One might think that the federal government was doing way better than the universities prior to Bayh-Dole, at least in the area of health care. But that’s not how politics spun the quant, obviously.
Since it typically takes millions of dollars to transform a discovery into a profitable product, industry advocates argued, few firms would pursue something that could be reproduced by a competitor once the first company had succeeded.
This is a strange argument, but one that keeps getting repeated. Profitable products are not the only use for discoveries. Discoveries can often be used directly by anyone capable. That’s the case, especially, for method inventions. There’s nothing to “develop.” That’s also the case, largely, with things like disease assays, which follow a procedure, and software, which implements algorithms. Further, often discoveries may improve or extend a known product, and therefore do not require “millions of dollars” in development because most of that work has already been done.
Finally, and most importantly, if there is still “millions of dollars” of development work to be done, then there is more intellectual property yet to be developed, and any patented invention in such a line of development can prevent others from copying that invention. Thus, exclusive commercial positions, where development is required, do not need to arise from an initial access to a discovery, even when a new product must be created from mere research findings. Anyone can develop, and when they do, they will invariably have to invent, and the first to invent will control that line of development. If there are other lines of development, then there will be competition. Otherwise, the competition is to be first to establish any significant development necessary to create a commercially profitable product. That’s no different than a company doing its own research in a race with other companies. And if there’s no significant development necessary, then also it’s not likely a product will take millions of dollars to develop. The patent then plays a different role: to prevent others who might also readily develop a product that doesn’t take millions of dollars from doing so. That is, to preclude competition, not to enable the necessary investment to make a product at all.
So the argument about millions of dollars of development doesn’t make any sense unless the development is all testing and not inventing. Even in drug development, one of the places in which development is expensive and copying relatively inexpensive, there are still things to work out having to do with the therapeutic dose, the method of delivery, testing for efficacy, and the like. These things, too, may be inventive and create the desired exclusive commercial position with regard to how any given company settles on doing things. To my mind, the argument about “millions of dollars” is unfounded. But it keeps getting repeated, in part perhaps because companies do spend “millions of dollars” developing new products. They just don’t need to have an exclusive license to a fundamental enabling discovery to justify the expense. But no doubt most companies would jump at the chance to exclude their competition where the expense to develop is not all that great and the profit potential is high, especially if the price to be charged can be based on a patent monopoly without competition.
Nevertheless, his signature opened the floodgates to a river of money that has become more turbulent over time.
There has not been that much money, actually, except in some rare–once every thirty years–kind of deals. What got opened were the floodgates to on a river of administrative windfall-lust. Just start a licensing operation and you could have piles of money–and have the cover that every dollar was “in the public interest.” Just hire a few of these nice patent brokers and their attorney friends, and the rest will take care of itself in just, um, a few years, yeah, like five, or twenty, or thirty maybe, sure, thirty at the max. If it takes much longer than that, it’s because the university hasn’t invested enough money in the effort, or faculty have been uncooperative, or the federal government still hasn’t provided enough research money. Anything other than that the dual monopoly approach (take all inventions, license to create private monopolies) doesn’t work as advertised.
And there’s not much that anyone can point to that this stream or creek of money has accomplished, other than a few rich inventors, a building here or there, and a bunch of well paid patent attorneys and tech transfer professionals. The attorneys and administrators have done well, regardless of the outcomes of patent licensing deals.
Markel asks some good questions. But here’s one that’s not relevant to Bayh-Dole:
How can scientific discovery proceed if all innovations and research tools are patented and the discoverers control access to them?
This is, in its way, the same issue that Michael Eisen is addressing now, with his concern for patents and the pursuit of knowledge. The answer is simple: it is not the patent that causes the problem, it is how the patent right is deployed. Back in the day, when Archie Palmer compiled the then-current list of reasons why universities should not be into patenting, one of the reasons to patent was to preserve the research and public commons against monopolist attacks. This is the same approach used by “free” open source (“free” as in speech, not beer) with copyright–use copyright to create conditions to defend against the exploitation of copyright against the goals of making a source code and its derivatives openly available. Similar methods are used in patent commons (such as under the Apache licensing protocols) and in standards–contribute your patents and in exchange gain access to everyone else’s patents, but if you assert a claim against the standard, you lose your rights in everyone else’s patents.
As long as there are copyrights and patents, then there is a need for copyrights and patents to resist exploits that, while “legal” may not be at all in the interests of those creating a commons. And while commons are not themselves always benign (patent pools can be used by monopolist interests as well), they certainly have their place and are often strongly aligned with academic goals. The trick in managing patents to encourage the expansion of the frontiers of science is to focus primarily on *not controlling access* to them, except when someone attempts to prevent others from enjoying the benefits of that access. Such thinking fuzzes the brains of patent attorneys raised with the idea that to have a patent means having the right to threaten to destroy one’s competitors, unless they pay, if pay is more desirable that destruction. A patent on an academic discovery is used, not as an ordinary patent, but as a patent to resist the effects of ordinary patents on a research commons or on the public domain. It’s just that university administrators–most, not all–refuse to rise to the occasion and use academic patents in this way. Even when Bayh-Dole still contains traces of the gestures to such an approach as the default.
The way universities have implemented Bayh-Dole, discoverers don’t get to control access to “innovations”–meaning inventions made with federal support don’t gain the benefit of academics choosing whether to make their discoveries broadly available or hold them close. University administrators take these inventions, often citing Bayh-Dole as the reason, and treat them as potential profit-generating “assets” to be held in a “portfolio” which in turn is “successful” if one invention a decade scores a lucrative deal. But Bayh-Dole does not vest title with universities, does not mandate that universities take title, does not authorize a portfolio approach, does not require commercialization, and does not treat patents on subject inventions as ordinary inventions, and does not encourage exclusive licenses.
If academic inventors were given the freedom to choose, many would choose not to patent–and thereby increase the public domain, make research exchange simple and make technology adoption direct. If academic inventors were given the freedom to choose, when they did opt for patenting, they also would choose invention management organizations that had a decent profile for the technology and industry represented by the invention. Inventors would not be stuck with their own institution, regardless of its capabilities and then implicated in a conflict of interest between trying to make money and also being a trustee safeguarding the integrity of the research in the public interest. And if academic inventors had the freedom to choose, they could negotiate with those agents for how an invention would be managed, under what conditions the agent could be dismissed and if so, how rights in an invention would be disposed. That’s a tough negotiation when the agent is also the university that employs you. Finally, if academic inventors are given the choice, they may well choose to have the federal government act as the trustee for their inventions–that federal government management, especially in the form of non-exclusive licenses available to all, has the advantage of recognizing inventive work with a patent, disclosing the invention in the patent literature, and managing any resulting patents without a profit motive. That sort of management, it turns out, is strongly aligned with academic values (at least, for many academic researchers).
That’s my experience, anyway, in nearly twenty years of working with academic researchers. They make good choices relative to the research, to their colleagues, to their institutions, and with respect to industry. Some want money. Others don’t. No arbitrary policy can begin to anticipate their needs or the variations in the circumstances of their inventions.
Here’s Markel’s conclusion:
It’s time for Congress to recalibrate Bayh–Dole.
Clearly, nothing happened on this in 2013 or 2014 as a result of Markel’s article. That’s unfortunate–we need more articles like this one, but in forums monitored by the people who control university patent policies and practices, or who can influence those people. Those would be the major donors to the university, regents, football coaches, and state legislators, and senior administrators. Maybe the AAUP would gather its gumption and go after patent policies that abrogate academic freedom by forcing faculty inventors to publish in the patent literature.
But I will make a different point. University patent policy and practice has next to nothing to do with Bayh-Dole. Bayh-Dole doesn’t require university administrators to behave as they do with regard to patents on federally supported inventions–not in claiming inventions, not in what they choose to file patent applications on, not in how they conduct those patent prosecutions (in secret, to get cleverly broad claims), not in how they hold patents for licensing, not in the structure of the deals they demand. Next to none of this depends on Bayh-Dole. Recalibrating Bayh-Dole won’t do a thing to change the course university administrators have chosen for their invention management programs. Even repeal of Bayh-Dole won’t change a thing at universities, other than perhaps pulling some inventions away from badly misconceived and mismanaged university patent licensing programs. Policies and practices will persist. Bureaucrats, once entrenched, are loathe to risk their livelihoods by giving creative people their freedom–even if that freedom is the freedom promised in the form of academic freedom; even if that freedom is what makes academic researchers distinctive as independent investigators supplied with institutional scale resources.
Until federal patent policy recognizes the freedom of the academic investigator, and the academic inventor (not always the same thing), and recognizes as well the role of the public domain, commons, and platforms in developing the foundations for new opportunities in technology, we will stay on the present university path to the Moloch state, one in which administrators interfere in every possible discovery to attempt to extract payments before anyone may use or adopt what’s been discovered.