Getting at the truth about Bayh-Dole’s impact, Part 5

Now we get to the crunch of Catherine Kirby’s blog article–published at a Rice University web site for entrepreneurship–with the section “Did the Bill Work?”

Since the passage of the Bayh-Dole Act, more than 5,000 new companies have formed from federally funded university research.

Follow the link to the source–a propaganda piece by AUTM:

Since the enactment of Bayh-Dole, more than 5,000 new companies have formed around university research.

Notice the misquote in the patch writing here. AUTM does not report how many companies formed around federally funded research. AUTM does not report how many inventions claimed by universities are subject inventions. Bayh-Dole keeps that information secret. AUTM doesn’t ask for it. Instead, AUTM counts the total number of startups reported to it “since” Bayh-Dole. The impression is that these companies must be the result of Bayh-Dole. There’s no evidence presented for a connection, however.

At the University of Utah, they went on a company-creation tear, forming 20 companies a year for five years–most of which were shell companies with no employees, no operations, no funding. But Utah was for a time viewed as the startup leader in the country. Just an administrative illusion to secure millions in state economic development investment to realize even more “potential.” Creating a company is as easy as filing paperwork with the state. Licensing to such companies is also easy, so long as there isn’t anyone on the other side of the table with risk capital in play. Even inducing speculative investors to buy into such a company is modestly possible, especially in places with uncanny wealth. But getting a company to achieve practical application of a subject invention–that part eludes most folks fixated on exclusive licensing. It’s much easier to create a Potemkin village with the illusion of economic development than to actually contribute to economic development through the beneficial use of inventions made with federal support.

And as for the 5,000 figure, sadly, AUTM doesn’t even bother to consider whether multiple universities are reporting the same startup companies. A number of university inventions are joint between researchers at different institutions. Each university will report the startup as its own. Repeat across all such joint inventions. Similarly, a startup may need to get licenses from multiple universities. I saw one startup that had to get licenses from 20 universities, all of which no doubt counted the company as their startup. So the AUTM figure over-reports startups.

One more thing. Look at the AUTM wording: around university research. AUTM isn’t counting subject inventions, patentable inventions in general, or even new technology, or even anything licensed. They just say “around university research.” That could be most anything–even folks reading published articles. That all may be fine and good, but the context is that somehow Bayh-Dole has something to do with this output, that patents on subject inventions are in play, that these patents have been licensed, and that the companies have operations, and are developing these inventions, and still exist. The word play by AUTM is intended to deceive. This isn’t even putting a good spin on the numbers. It’s cooking the numbers first, and then attributing to Bayh-Dole activity that has no immediate connection with Bayh-Dole. We might say this is corrupt. Or we might observe how easily AUTM’s text is able to deceive a student writing a post for an entrepreneurship center into thinking that the recited figures are reliable and arise because of Bayh-Dole. Rotten AUTM apples.

To put the inflated and misleading 5,000 company figure into perspective, contemplate for a moment this chart:



This chart is from the U.S. Bureau of Labor Statistics. It runs only from 1994 to 2015, whereas the AUTM count is from 1980 to 2010. But the time line is not so important as the scale. Each year we see on the order of 600,000 new company starts. In a recent thirty year period, we might expect 15 million new companies. And AUTM overcounts and switches counts to get to 5,000 in thirty years. The new company formation around research is a rounding error on a rounding error. 5,000 over 15,000,000 rounds to like, utter insignificance.

There may be a few hundred companies, starting with Google, that have developed around subject inventions during the Bayh-Dole era and that have fielded licensed inventions resulting in commercial products with public benefits available on reasonable terms. Those companies are no doubt generally a good thing, and there’s no harm in encouraging company starts around research or taking some pride in those companies when they do succeed. Good.

But to attribute that activity to Bayh-Dole–that takes some doing. If all we want is company formation around university research, regardless of federal funding or subject inventions or patent licenses, then we might posit that most university startups that have taken a license to a patent on a subject invention (a posi) have operated despite Bayh-Dole. That is, Bayh-Dole has made their operations the more difficult–delayed access, raised the cost, even denied them the control they would otherwise have had over inventions if assigned directly to the startup by inventors. A mere count of startup companies does not begin to address whether Bayh-Dole has produced more such companies, or less, or has been a net benefit to these companies or a net drag. Again, what do you think the likelihood is that institutionally controlled patents are seen as a boon to entrepreneurs seeking to start companies?

The other figures cited by Kirby appear to come from a student paper at MIT, which itself cites only a few sources, mostly dated, that themselves lack credible data (if one cares to chase these things down, turtle to turtle). To her credit, Kirby acknowledges that Bayh-Dole might not be the only driver:

The Bayh-Dole Act may not have been the only contributor, but these large numbers show the importance of university innovation to the economy and make it clear that innovation spurring legislature can have enormous positive effects on economic growth.

As we’ve seen, the “large number” of startups is actually an insignificant number in the context of total startups. Same for new products, same for economic activity. Tiny numbers, even when spun, skewed, and slid. Next to no impact, in context. While there may be important products arising from university patent licenses, that importance is not established by counting. Even one really important product might make the case for Bayh-Dole and the value of trying to create ever more monopoly positions from federally funded research inventions. But if numbers are not the thing, then it is worth looking squarely at the activity of making patent monopolies and asking if this is how we wish to live our lives and communities. If we take joy in the products of monopoly investment and pricing, perhaps. But if we think there are other ways to do things, then those ways, too, may be worth developing as a matter of federal invention policy.

It is clear, in context, that Bayh-Dole has had inconsequential effects on economic growth. Not even a drop in the bucket, really. Sure, some people have gotten wealthy from their monopoly positions. For them, to the extent they thank Bayh-Dole for their wealth and don’t curse Bayh-Dole because they may have been even wealthier, Bayh-Dole has been a great thing. What’s not at all clear, however, is that Bayh-Dole has been “innovation spurring legislation.” It has led to more patent monopolies in the hands of university administrators. That appears to be true. Innovation–introduced change to an established order?  Not so much. And as for practical application–established use with public benefits on reasonable terms–that’s still a government secret.

As for the criticism Kirby cites, these are odd ones. Data sharing doesn’t generally have anything to do with patent rights. A patent cannot exclude sharing data. Why would anyone bother with worrying that? Oh, yes, data that haven’t been analyzed might contain the roots of discovery. That’s why clinical trial data are often closely held until the data have been worked over. But to get at data sharing before analysis requires a different sort of study. In any event, the paper Kirby cites by Mowery, Thompson, et al. finds that academic patents may suppress research findings as inputs for further research, at least in the University of California. They don’t “leave open” this point–they publish a draft paper in progress and indicate the direction of their finding. Further, their paper is not about “data” but scholarly communication. Some difference.

As for the biotech industry a patent royalty rate of 1% is not going to drive up the price of drugs. A bigger driver is the monopoly position that companies have as a result of an exclusive=assignment license. They can charge monopoly prices, not the “reasonable” prices that Bayh-Dole implies. Why mention “reasonable terms” in Bayh-Dole if any terms a monopolist chooses are reasonable? Competition can bring down prices, where there’s not collusion and the markup over costs is already high.

Another bigger driver appears to be the cost associated with developing a biomedical product within a monopoly. The cost of drug development within a monopoly appears to be much greater than developing the same drug in a collaborative setting, where organizations can share test results, data, and models rather than having to do everything for themselves. It would be interesting to run the experiment–whether licensing non-exclusively a compound with attractive therapeutic effects might lead to lower cost development and reasonable pricing–and still provide a nice profit for those involved. But that experiment is unlikely any time soon, so long as the public policy idea that inventions made with federal support at institutions dedicated to a public mission ought to be available to the public is readily circumvented using Bayh-Dole.

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

1 Response to Getting at the truth about Bayh-Dole’s impact, Part 5

  1. Pingback: Getting at the truth about Bayh-Dole’s impact, Part 4 | Research Enterprise

Comments are closed.