All about IP & Price Discrimination

It's a grading/break week, so just a short post. A recent article that I enjoyed a lot, but that hasn't found much love on SSRN is Price Discrimination & Intellectual Property, by Ben Depoorter (Hastings) and Mike Meurer (Boston University). The paper has the following abstract:

This chapter reviews the law and economics literature on intellectual property law and price discrimination. We introduce legal scholars to the wide range of techniques used by intellectual property owners to practice price discrimination; in many cases the link between commercial practice and price discrimination may not be apparent to non-economists. We introduce economists to the many facets of intellectual property law that influence the profitability and practice of price discrimination. The law in this area has complex effects on customer sorting and arbitrage. Intellectual property law offers fertile ground for analysis of policies that facilitate or discourage price discrimination. We conjecture that new technologies are expanding the range of techniques used for price discrimination while inducing new wrinkles in intellectual property law regimes. We anticipate growing commentary on copyright and trademark liability of e-commerce platforms and how that connects to arbitrage and price discrimination. Further, we expect to see increasing discussion of the connection between intellectual property, privacy, and antitrust laws and the incentives to build and use databases and algorithms in support of price discrimination.

They call it a chapter, but they don't identify the book that the chapter will appear in. It's probably an interesting book.

In any event, the chapter is a really interesting, thorough look at price discrimination generally, in addition to price discrimination as it relates to IP. It discusses the pros and cons as well as the assumptions that underlie each. If you are interested in a better understanding of the economics of IP (and secondarily, the internet), this is a good read.

The Value of Patent Applications in Valuing Firms

It's an age-old question that we've blogged about here before - what role do patents have on firm value? And is any effect due to signaling or exclusivity? Does the disclosure in the patent have any value? Does anybody read patents?

These are all good questions that are difficult to measure, and so scholars try to use natural experiments or other empirical methods to divine the answer. In a recent draft, Deepak Hegde, Baruch Lev, and Chenqi Zhu (all NYU Stern Business) use the AIPA to provide some useful answers. For those unaware, the AIPA mandated that patent applications be published after 18 months by default, rather than held secretly until patent grant. The AIPA is the law that keeps on giving; there have been several studies that use the "shock" of the AIPA to measure what effect patent publications had on a variety of dependent variables.

So, too, in Patent Disclosure and Price Discovery. A draft is available on SSRN, and the abstract is here:

We focus in this study on the exogenous event of the enactment of American Inventor’s Protection Act of 1999 (AIPA), which disseminates timely, detailed, and credible public information on R&D activities through pre-grant patent disclosures. Exploiting the staggered timing of patent disclosures, we identify a significant improvement in the efficiency of stock price discovery. This improvement is stronger when patent disclosures reveal firms’ successful, new, or technologically valuable inventions. This improvement is more pronounced for firms in high-tech or fast-moving industries, or with a large institutional ownership or analyst coverage. We also find stock liquidity rises and investors’ risk perception of R&D drops after the enactment of AIPA. Our results highlight the importance of timely, detailed, and credible disclosures of R&D activities in alleviating the information problems faced by R&D-intensive firms.

This is a short abstract, so I'll fill in a few details. The authors measure the effect on  intra-period timeliness, a standard measure used to proxy for "price discovery," or how quickly information enters the market and settle the price of a stock. There are a lot of articles on this, but here's one for those interested (paywall, sorry).

In short, the authors look at how quickly price discovery occurred before and after the AIPA, correcting for firm fixed effects and other variables. One of the nice features of their model is that patent applications occurred over a period of years, and so the "shock" of patent publication was not distributed only in one year (which could have been affected by something other than the AIPA that happened in that same year).

They find that price discovery is faster after the AIPA. Interestingly, they also find that the effect is more pronounced in high-tech and fast moving fields -- that is, industries where new R&D information is critically important.

Finally, their results say something about the nature of the patent disclosure itself - the effects come from disclosure of the information, and not necessarily the patent grant. Thus, the signaling effect may really relate to information, and (some) people may well read patents after all.

Adam Mossoff: Are Property Rights Created By Statute "Public Rights"?

I greatly enjoyed Professor Adam Mossoff's new article, Statutes, Common-Law Rights, and the Mistaken Classification of Patents as Public Rights, forthcoming in the Iowa Law Review.  Mossoff's article is written in the wake of Oil States Energy Services v. Green's Energy Group, where the Supreme Court held it is not unconstitutional for the Patent Trial & Appeals Board (PTAB), an agency in the Department of Commerce, to hear post-issuance challenges to patents, without the process and protections of an Article III court. Justice Thomas' opinion concluded that patents are "public rights" for purposes of Article III; therefore, unlike, say, property rights in land, patents can be retracted without going through an Article III court.

Mossoff's article objecting to this conclusion is a logical follow on to his prior work, while also providing new insights about the nature of patents, property, and the public rights doctrine. He does so quite concisely too, with the article coming in at only 21 pages.

Helsinn Argument Recap: Did the AIA Change the Meaning of Patent Law's "On Sale" Bar?

As Michael previewed this morning, the Supreme Court heard argument today in Helsinn v. Teva, which is focused on the post-America Invents Act § 102(a)(1) bar on patents if "the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public" before the relevant critical date. The Federal Circuit held that Helsinn's patents were invalid because Helsinn had sold the claimed invention to a distributor more than one year before filing for a patent, but Helsinn (supported by the United States as amicus) argues that the "on sale" bar is triggered only by sales that make the invention "available to the public" under a broad reading of "public."

During argument, none of the Justices seemed inclined to favor Helsinn's attempt to argue that "on sale" clearly means on sale to everybody—Justice Kavanaugh said "it's pretty hard to say something that has been sold was not on sale," and Chief Justice Robert's noted that Helsinn's interpretation "might not be consistent with the actual meaning of the world 'sale'" because "if something's on sale, it doesn't have to be on sale to everybody." Nor did they jump at the government's argument that "on sale" means a product can be purchased by its ultimate consumers—Justice Sotomayor said: "This definition of 'on sale,' to be frank with you, I've looked at the history cited in the briefs, I looked at the cases, I don't find it anywhere."

Helsinn's better statutory argument is that the meaning of "on sale" is modified by "or otherwise available to the public" to require that the sale be publicly available. Indeed, for a reader with no background in patent law, this might seem like the most natural reading of the statute. Justice Alito said that "the most serious argument" against the Federal Circuit's position is "the fairly plain meaning of the new statutory language," and that he "find[s] it very difficult to get over the idea that this means that all of the things that went before are public." And Justice Gorsuch suggested, at least for hypothetical purposes, that "the introduction of the 'otherwise' clause introduced some ambiguity about what 'on sale' means now." But if there was more support to reverse the Federal Circuit, it was not apparent from the argument.

Much of the statutory language used in the Patent Act—including "on sale"—has developed a technical legal meaning over time, generally due to courts' attention to the law's utilitarian focus. For example, patentable subject matter caselaw is "implicit" in § 101, courts have put a highly specialized gloss on the word "obvious" in § 103, and—relevant here—the § 102 categories of prior art have long been interpreted to include relatively obscure and private uses. Although this expansive definition of prior art might seem unfair to patentees, there are also strong policy arguments in its favor, including (1) encouraging patentees to get to the patent office early (leading to earlier disclosure and patent expiration) and (2) avoiding patents when their costs (including higher prices for consumers and subsequent innovators) aren't likely to be outweighed by their innovation-incentivizing benefits, such as when there is independent invention—even when evidence of that invention is relatively obscure.

As Justice Kavanaugh noted at argument today, Mark Lemley's amicus brief on behalf of forty-five IP professors describes the long history of treating relatively non-public disclosures as prior art, including (1) "noninforming public use" cases, (2) "output of a patented machine or process" cases, and (3) cases involving secret, confidential, and nonpublic sales transactions. Justice Breyer also mentioned the Lemley brief, and he said it "seems right" to have the on-sale bar include private sales "to prevent people from benefitting from their invention prior to and beyond the 20 years that they're allowed." The legislative history of the AIA does not suggest that Congress intended to do sweep away all of these cases—Justice Kavanaugh said that he thinks "the legislative history, read as a whole, goes exactly contrary" to Helsinn's contention because "there were a lot of efforts … to actually change the 'on sale' language, and those all failed," leaving the losers "trying to snatch victory from defeat" with "a couple statements said on the floor."

It is perhaps because of this history that Helsinn and the government seemed more focused on the argument that "on sale" has always excluded nonpublic sales than on the argument that the AIA changed the law. Justice Ginsburg's only comment during argument was to ask Helsinn to clarify this: "I thought that one argument was that the AIA changed the way it was. But … you seem to say there was no change; 'on sale' never included the secret sale." Arguing for the government, Malcolm Stewart even conceded—in response to questioning from Justice Kagan—that if the law was settled pre-AIA such that "on sale" included nonpublic sales, then the new AIA language ("or otherwise available to the public") "would be a fairly oblique way of attempting to overturn" the law. But based on my reading of the transcript, it doesn't seem likely that the argument that "on sale" has always meant "on sale publicly" will get five votes.

I waited until after writing the above to get Ronald Mann's take at SCOTUSblog, but I think I very much agree on his bottom line conclusion: while this isn't "a case in which the argument clearly presages the result," the overall transcript "suggests that the most likely outcome will be an affirmance."

How Important is Helsinn?

In honor of the oral argument in Helsinn today, I thought I would blog about a study that questions its importance. For those unaware, the question the Supreme Court is considering is whether the AIA's new listing of prior art in 35 U.S.C. §102(a)(1): "the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public..." changed the law.

Since forever, on sale meant any offer or actual sale, regardless of who knew about it. Some have argued that the addition of "or otherwise available to the public" means that only offers that are publicly accessible count as prior art. I think this is wrong, and signed on to an amicus brief saying so. We'll see what the Court says. Note that non-public does not mean "secret." True secret activity is often considered non-prior art, but the courts have defined "public" to mean "not-secret." The question is whether that should change to be "publicly accessible."

But how big a deal is this case? How many offers for sale would be affected? Steve Yelderman (Notre Dame, and soon to be Gorsuch clerk) wanted to know as well, so he did the hard work of finding out. In a draft paper on SSRN that he blogged about at Patently-O, he looked at all invalidity decisions to see exactly where the prior art was coming from. Here is the abstract for Prior Art in the District Court:

This article is an empirical study of the evidence district courts rely upon when invalidating patents. To construct our dataset, we collected every district court ruling, verdict form, and opinion (whether reported or unreported) invalidating a patent claim over a six-and-a-half-year period. We then coded individual invalidation events based on the prior art supporting the court’s analysis. In the end, we observed 3,320 invalidation events based on 817 distinct prior art references.

The nature of the prior art relied upon to invalidate patents informs the value of district court litigation as an error correction tool. The public interest in revoking erroneous patent grants depends significantly on the reason those grants were undeserved. Distinguishing between revocations that incentivize future inventors and those that do not requires understanding the reason individual patents are invalidated. While prior studies have explored patent invalidity in general, no study has reported data at the level of detail necessary to address these questions.

The conclusions here are mixed. On one hand, invalidations for lack of novelty bear many indicia of publicly beneficial error correction. Anticipation based on obscure prior art appears to be quite rare. When it comes to obviousness, however, a significant number of invalidations rely on prior art that would have been difficult or impossible to find at the time of invention. This complicates — though does not necessarily refute — the traditional view that obviousness challenges ought to be proactively encouraged.

So, let's get right to the point. The data seem to show that "activity" type prior art (that is sale or use) is much more prevalent in anticipation than in obviousness. This is not surprising, given that this category is often the patentee's own activities.

With respect to non-public sales, they estimate that a maximum of 14% of anticipation and 2% of obviousness invalidations based on activity were based on plausibly non-public sales. This translates to about 8% of all anticipation invalidations and 1% of all obviousness invalidations. Because there are about as many obviousness cases as anticipation cases, this averages to 4.25% of all invalidations. They note that with a different rule, some of these might have been converted to "public" sales upon more attention paid to providing such evidence.

A related question is whether the inventor's actions can invalidate, or whether the AIA overruled Metallizing Engineering, which held that an inventor's secret use can invalidate, even if a third-party's secret use does not. The study found that the plaintiff's actions were relevant in 27% of anticipation invalidations and 13% of obviousness invalidations.  Furthermore, they found that most of the secret activity was associated with either the plaintiff or defendant--this makes sense, as they have access to such secret information.

So, what's the takeaway from this? I suppose where you stand depends on where you sit. I think that wiping out 4% of the invalidations, especially when they are based on the actions of one of the two parties, is not a good thing. It's bad to allow the patentee to non-publicly sell and have the patent, and it's bad to hold the defendant liable even if it has been selling the patent in a non-public (though non-secret) way. We're talking about 20 claims per year that go the other way - too high for my taste, especially when it means we have to start defining new ways to determine whether something is truly public.

Furthermore, the stakes of reversing Metallizing are much higher. I freely admit that the "plaintiff's secret actions only" rule has a tenuous basis in the text of the statute, but it has been the law for a long time without being expressly overruled by two subsequent revisions. Given that more than 25% of the invalidations were based on the plaintiffs actions, I think it would be difficult to reverse course.