AAU, APLU, and others aim to “bolster” federal technology transfer, 4

We are dealing with the bombast that AAU and other “higher education associations” put forward as advice to NIST with regard to how the federal government might better manage its own technology transfer. Instead, the HEAs seek to improve their members’ own technology transfer. The HEAs adopt the fine bureaucratic strategy of insisting they are doing everything well and Bayh-Dole is working as intended while complaining that they don’t have enough money, that the requirements are too demanding, that patent law is a problem, and that the tax code lacks the proper incentives. That is, the university approach to patent monopolies on publicly supported research is not working, but it is not the fault of the HEA members, even though they have adopted a historically failed approach, spent wickedly on it, and ignore all the key requirements of Bayh-Dole while they are at it.

Let’s continue. The HEAs are going to argue for new metrics to account for what universities do in technology transfer, since the metrics that the universities have put out during the Bayh-Dole era apparently are inadequate. But first the HEAs insist on using those old metrics.

The HEAs claim in their advice to NIST that universities received 6,452 US patents in 2016. A PTO search for utility and plant patents assigned to universities shows 10,978 US patents. But only 3,264 recite government funding. The HEA patent figures aren’t even close, and why do the HEAs cite all university patenting when the issue at hand isn’t all university patenting, but rather federal technology transfer–the inventions owned by the federal government, or made at federal labs?

Lost in all of this is the observation that universities get nearly 2/3 of their research funding from the federal government, but only about 1/3 of their patents include a federal funding notice. Isn’t that an amazing result? The HEAs want NIST to look at all university patenting, but not at the oddity that the non-federal funded research is twice as productive in producing patents as is the federally funded research. Now producing patents is itself not a good measure of anything related to technology transfer, innovation, or public benefit–but it does suggest that there are university practices at work that favor dealing with inventions that don’t involve federal funding.

The HEAs add data from AUTM’s licensing survey on startups and a very sketchy claim made about the economic impacts of university “licensing activity.” The AUTM startup information is not audited for accuracy, involves multiple universities reporting the same startup as their own, and fails to exclude shell companies created by universities to pad their startup figures (as have University of Washington and University of Utah). As for the economic impact estimates, the model used takes as input the reported licensing income, treats that as a percentage of total income, allocates the income as paying for jobs, and claims the economic impact of all those people spending that income. It’s an exercise in multiplication, a mathematical propagandum, a political bluff with no published attempt (that I can find) to validate the model. The output of the model gets passed around among the various front groups that defend the university/pharma patent monopoly pipeline approach.

The telltale signs that the model is political bluff include the “up to” 4.3 million jobs and “up to” $591 billion in GDP. I haven’t found a full account of the economic “model” that AUTM or BIO relies upon to get their figures, but here’s how AUTM’s earlier model “worked.” Take $200,000 of licensing income reported by a university. Assume that this income reflects an earned royalty on sales. Assume the royalty rate is 2% of adjusted gross income. That indicates gross sales of around $10m (200,000 x 100/2). Now assume that that some portion of the sales revenue is profit, some goes to materials and energy, and the rest goes to labor. So let’s say the profit margin is 10%, fixed costs are another 10% and the rest, 80% is labor–that’s $8m. How many tech workers can that $8m support? Let’s say a tech worker makes on average $100,000, including benefits. That would mean “up to” 80 jobs are supported by the income indicated by the license payment. Again, $100,000 of licensing income indicates “up to” 80 tech jobs.

Now, some reality. Let’s say that university U licenses an invention exclusively to a big corporation. Let’s say a hefty $100,000 up front, deductible from future royalties, reimbursement of patenting expenses, and a 2% earned royalty on sales. The company has the university file five foreign patent applications in addition to the US application, and at the end of the first year of the license sends the university $200,000. $100,000 is the upfront fee. $100,000 is reimbursement of patenting expenses. Half of this income comes from company earnings on other products–there’s no sales, so the income cannot represent new jobs. But the income could support 1 university administrative position. The other half of the payment goes to a law firm for patenting costs. The law firm in turn sends out $15,000 each to five foreign affiliates, and each of these uses a portion of that money to pay for technical translation services for the national phase patent applications. Let’s say that the payment to each foreign affiliate then supports 1/10th of two full time jobs. That’s 2/10ths of a full time job times 5 affiliates, or one non-US full time equivalent job. Let’s say that the $25,000 retained by the law firm represents another 1/4 of a full time job (though for law firms we have partners, associates, and paralegals at significantly different pay scales all getting their share). We can account, then, for “up to” 6.25 jobs, five of them foreign, one in the university, and a quarter of a job in a U.S. law firm.

The economic model, however, proposes “up to 80 jobs” based on that payment of $200,000. Reality is up to 6.25, and 5 of those are not even in the U.S–and all of them are on the university income side of the deal, not jobs out there in company licensing land. Let’s say the economic model overestimates jobs by 10x and places the jobs on the wrong side of the licensing deal without revealing that.

Now consider a bigger deal. Let’s look at Emory’s deal with Gilead and Royalty Pharma for $525m. In this deal, which is much more complicated than the rough outline here, Emory licenses its invention (drug name Emtriva) exclusively to Triangle Pharmaceuticals, a startup with a bunch of former Wellcome executives and scientists working to develop various drug candidates. Triangle gets acquired by Gilead in a deal worth $464m. Then Gilead, Royalty Pharma, and Emory do a deal for $525m. In this deal, Gilead and Royalty Pharma buy out Emory’s future royalties, with Gilead paying 65%, Royalty Pharma paying 35%, and with Emory and some of its inventors given the option to acquire an interest in Royalty Pharma. Under the deal, Gilead will then pay future royalties to Royalty Pharma. In this deal, what would the AUTM simple economic model produce? There’s $525m of reported income. If it were from sales, with a 2% royalty, we are looking at $26.25b. Reduce 20% for profits and costs, and we have $21b. At $100,000 a tech worker, that’s “up to 210,000 jobs.” Of course, “up to” makes it all uncontestable–no one says there really were 210,000 jobs, nor does the model say, actually, where the jobs are located–on the company side of the deal (Gilead, Royalty Pharma) or the university side of the deal (Emory).

The invention that Emory licensed has indeed produced income for both Gilead and Royalty Pharma. It’s just that now that income will not be reported as part of any government reporting under Bayh-Dole because Emory has received the income that it will receive for the license. How many jobs are supported by the sales income at Gilead? How many at Royalty Pharma, which gets the Emory share of royalties? How many at Emory, taking its lump sum payment? We don’t know–but we can be sure that the $525m did not support “up to” 210,000 jobs. That money came from other sales and investment income at Gilead and Royalty Pharma.

Some of the Emory’s income will go to its inventors (current policy is 25% of income over $4m), and some to the Chemistry department (Emory reported 10%). Here’s how an Emory official described the use of income from the deal:

“A large proportion of recent royalties, including those from Emtriva, are being used to implement the relevant components of Emory’s new strategic plan, which includes faculty recruitment, financial aid, and new initiatives in predictive health, global health, neurosciences, and computational and life sciences,” said Michael Mandl, executive vice president for finance and administration.

There would appear to be faculty and administrative jobs involved, though it is not clear just how many. Let’s say Emory held $350m after various payouts and invested that, spending the income. At an 8% return, that would be around $28m a year, and at $200,000 a year for biomedical faculty (average in 2016 was about $165,000 for professors at Emory, add benefits) means about 140 jobs. Round it up to 200. Well, that’s not 200,000, but we haven’t accounted for the jobs at Gilead. According to this web site, Gilead has about 10,000 employees and has an operating profit of 54% across all its sales. Even if all of Gilead’s income came from Emtriva, we’d still have only 10,000 jobs, not 210,000. According to this site, the average salary at Gilead is about $108,000–in line with our estimates. And we don’t know how many jobs were lost at other companies as Gilead’s new drugs came onto the market. After all “jobs supported” ought to be a net of the income generated less the lost income at other companies that compete in the same market.

Perhaps the point is clear. “Up to 4.3 million jobs” is impossible to verify and is only true because “up to” means there’s really no factual claim being made at all. It’s advertising copy made to sound impressive, leaving the impression that all those university licensing deals are creating jobs. The overestimation appears to be on the order of 200x. Perhaps we might estimate 21,500 jobs supported. It’s a nice number for those involved. But in terms of an impact on the national economy, there’s no particular reason to be impressed or deluded.

More telling is that the HEAs’ metrics are directed at all university licensing activity, not simply the Bayh-Dole portion, even though the discussion is supposed to be about improving federal technology transfer. Nearly two-thirds of current university patenting activity does not cite federal funding. The patent metrics suggest that universities are doing much less patenting of federally supported work. One wonders if that is Bayh-Dole working as intended. No doubt fewer patents are better–but is that what the HEAs mean?

Finally, compare the HEAs’ metrics with U.S. totals–962 companies, many of them shell companies with no operations, no significant investment, and no products–in 2015, more than 650,000 new companies were formed in the U.S. The formation of a company is a paper transaction, as is the issuance of a patent. Neither has anything directly to do with practical application of a subject invention–use of the invention with benefits available to the public on reasonable terms. In fact, to the extent that companies are formed that don’t produce a product or patents are issued that don’t result in promoting utilization of inventions, these paper activities are waste, distractions, off-topic. Whether people make money from starting companies or from dealing in patents is beside the point.

These efforts, if they fail to produce practical application, are expensive overhead. Furthermore, if these startups and patents do result in practical application, we are still left with the question of whether practical application would have been achieved more quickly and with less expense and complication if a university had not insisted on licensing patents sweetheart and exclusively to its own startup or had not insisted on using patents in the first place. Even if one can show that in the presence of Bayh-Dole, certain beneficial things have happened, one still has not shown that those things happened because of Bayh-Dole, or that they have happened through an advantageous pathway to move publicly supported research inventions into public use. Public benefit may arise despite university patents and startups. It takes a bit of calculated cleverness to claim that because Bayh-Dole, these good things happened, when only about 1/3 of university utility patent activity presently involves federally funded inventions, and only 42% averaged over the Bayh-Dole era. That is, federally supported invention activity at universities is growing at a lower rate than is activity involving non-federal inventions.

Now for a fascinating conclusion following from all this bluff activity. The HEAs describe the work of two committees at AAU and APLU, respectively, that studied university technology transfer:

After extensive review, both working groups reaffirmed that the primary goal of university technology transfer operations is to advance the public interest.

It took extensive review for HEA administrators to come up with this? Holy moly! It’s difficult to comprehend what went into the “extensive review” at AAU and APLU. Were the working groups grappling with the possibility that technology transfer should not advance the public interest? That would be rather interesting. Now here’s a possibility: perhaps if university administrators were forbidden to claim as a goal “advancing the public interest” we might have better technology transfer. That’s right. Claiming a practice as a virtue suppresses discussion–who can be against virtue? In effect, aligning a given practice with a general claim of advancing the public interest serves to demonize anything that would stand against the practice. If a university’s invention management practice advances the public interest, it should be up to the public to point that out–the statements in support of a practice should be the result of evidence not aspiration. A university can report what it has done, but there’s no good point in ascribing nice-sounding intentions to the corporate fiction of the institution. That’s like telling people what to think. If a university’s technology transfer program is so awful that people have to be told what to think, or bluffed into delusion, then, really, should we be listening to what university administrators or their front groups proclaim?

A second problem with a university claiming that it acts to advance the public interest in invention licensing is that doing so obscures the prevailing actual practice, which is to attempt to advance a private interest–that of the exclusive licensee/assignee of an invention and its patent rights. The licensee/assignee gains the benefit of a patent monopoly created as a result of an invention made in a federally funded project and the basic deal is that the licensee/assignee will then share some small portion of the exploited monopoly back with the university, and the university will share some still smaller portion back with the inventors while pocketing the rest, and this activity advances the public interest. The HEAs in effect claim that private exploitation of patent monopoly positions advance the public interest. The only difference offered by Bayh-Dole is that university administrators have found a way to be complicit in that private exploitation, bureaucratic parasites. Nothing about paying a university a 1% royalty on sales contributes to a company’s practical application of a subject invention. And everything in Bayh-Dole that would favor public interest is called “chilling” to the success of the law. 

The primary public policy gesture in Bayh-Dole is authorizing private interests to preempt public objectives in funding faculty-led university-supported “extramural” research.

Bayh-Dole makes optional the enforcement of Bayh-Dole’s public covenant apparatus, creating the appearance of public protections while protecting private interest preemption.

Worrying compliance with the law without actual enforcement enhances the perception that universities comply with Bayh-Dole and its standard patent rights clause, including the public covenant apparatus.

Without actual operation of Bayh-Dole’s public covenant apparatus–compliance, march-in, and use of the government’s license to make, use, and sell–Bayh-Dole amounts to a private appropriation of public work, with administrative parasitism to facilitate the process and disenfranchise faculty from the process, while creating the illusion for the public that there are safeguards in place to prevent the very exploitation that Bayh-Dole authorizes.

The rare case was that a private interest would agree to do work on behalf of the public, and enjoy a patent monopoly for a time to accomplish that work. That was the rare case. Bayh-Dole made that case the default, made it difficult to vary from that default, and stripped the default of anything that would distinguish work on behalf of the public–any profit-seeking activity takes precedence over any public objective. Or, another way, private profit-seeking through a patent monopoly is the public objective.

I don’t  have a problem with private profit-seeking. I don’t have a problem with patent monopolies. These things have their goods and bads. Decent people make decent use of them, and assholes spoil most anything. Perhaps you like the combination–private profit-seeking, using patent monopolies, without any special conditions, based on publicly supported research or development. In my experience, that combination has proven destructive more often than that anyone has got their desired profits. Across most inventions arising in federally supported research, the approach does not merely not work–it is damaging, stifling, wearying.

Perhaps it’s as simple as that when the money gets to be too easy and to much, then things start to get slack. In a bureaucracy, administrators show up to routinize the slackness, and then “corruption” (called “conflict of interest” here) becomes deviation from the policies that authorize the management of the slackness. Seems what humans do. We might say that Bayh-Dole shows up as a clever scheme to siphon off federal money and public opportunity because the federal government allocates sufficient money to university research that creates sufficient value to justify the scam. Take that as a rhetorical hypothesis offered as an alternative to the HEAs’ account of the intention behind Bayh-Dole. If we really do want the public policy of preemption of public objectives by private interests whenever they obtain ownership of a patentable invention made in a project with federal support, then there is absolutely no reason for Bayh-Dole to include a public covenant or to involve payments to university administrations. Everything else is grossly parasitic and works against the success of the public policy preemption.

As it is, Bayh-Dole’s effect is to change the public purpose of the research itself after the fact of the grant award–so that the research is in fact conducted so that a future, single private interest will have the opportunity, with minimal (none, actually) public accountability, to preempt any other rationale given for making the grant. Under Bayh-Dole, there’s no consequence for failing to use an exclusively licensed subject invention. There’s no consequence for using an exclusively licensed subject invention without a public benefit. There’s no consequence for using an exclusively licensed subject invention but not making the public benefit available on reasonable terms. The Bayh-Dole apparatus that appears to favor public interest does not operate. That’s not because everything that gets done by university administrators under Bayh-Dole is always in the public interest but rather because no one really cares about the public interest. It’s a dead concept in practice that sounds good in aspirational political rhetoric. Private interests displace public ones. University administrators ought to be required to admit this much or have evidence to demonstrate their claim for advancing the public interest rather than a private interest or even the university’s own business interests.

But for that, people would have to put some effort into accepting what a public interest is, and how it might be advanced. What university administrators mean is that the public benefits when a private company makes a profitable drug available for sale. Advancing the public interest is to make otherwise public assets available as monopolies to private companies. That’s the “advance.” Whether those companies actually make product is not material. The public interest has been “advanced.” That is, a public asset has been brokered into a private asset. That’s the advance. Most university licensing offices end their description of the technology transfer “process” with an exclusive license (that’s functionally an assignment). They don’t care in practice (other than for the money) whether anything ever gets “commercialized.”

This entry was posted in Bayh-Dole, Bozonet, Policy, Technology Transfer and tagged , , , , , , , , . Bookmark the permalink.