Yearly Archives: 2020

Opting Out of Liability Rules

May 14, 2020 Dotan OliarAdd a Comment

Economic activities often conflict: a rancher’s stray cattle may reduce the value of a neighboring farmer’s crops, or a tech company’s file-sharing app may reduce the value of music labels’ records. When conflicts arise, society needs to decide which party’s interest to protect, and whether to do so with a property right or a liability rule. The law and economics literature teaches that lawmakers should devise entitlements while taking into account post-allocation transaction costs, with the goal of ensuring that resources end up in the hands of parties who put them to their most productive use.

Scholars have accordingly debated the relative performance of property rights and liability rules. In a seminal article, Rob Merges famously argued that one should not worry too much about transaction costs accompanying property rights. Should these prove prohibitive (as in the case of radio stations who need to license rights to many musical compositions that they wish to play over the air), and the property right choice inefficient, IP owners are likely to “contract into liability rules”;¹ that is, they will privately arrange liability-rule-based licensing schemes (such as ASCAP) to lower licensees’ costs of access.

In a wonderful new article, Kristelia García reviews recent market dynamics that lend support to the mirror-image argument; namely, that one should not overstate the arguable inefficiencies of liability rules.

Through a series of case studies of privately ordered deals, she shows how time and again IP owners have opted out of their statutorily-dictated liability rule protections and have in effect transformed them contractually into either property rights or differently delineated liability rules. Importantly, García argues, this form of private ordering has resulted in a stronger form of protection than that allocated to IP owners statutorily. Examples include YouTube’s Content ID agreement (allowing content owners to prevent infringing uploads ex-ante rather than only remove them ex-post under section 512 of the Copyright Act); or the 2012 Big Machine-Clear Channel Communications deal (creating a new terrestrial public performance right protected by a negotiated-rate liability rule where none exists statutorily). Analyzing these deals, García concludes that a new and additional consideration – right holders’ “perceived control”, i.e. their ability to grant or withhold permission to use their work and to set terms for its use—should guide lawmakers’ choice between property rights and liability rules.

Where does García’s article leave us? It seems that when transaction costs become substantial and may hinder owners’ ability to profit from their entitlements, they would rationally engage in market and institutional innovation in order to reduce transaction costs and get licensing deals done. They would do so regardless of whether the initial entitlements allocated were property rights or liability rules. Either way, when the initial entitlement proves inefficient, IP owners are expected to contract around the statutory default. This tends to suggest that the choice between property rights and liability rule protection, while important, should not be overstated (especially in an age where parties’ ability to contract around statutory defaults is aided by technological advancements in detecting, measuring and charging for use). This realization should not come as a surprise, but merely as a reminder of the Coase Theorem that whenever transaction costs are zero, bargaining will lead to the efficient result regardless of the initial allocation of entitlements.

As we approach a zero transaction costs world (or some attainable lower bound), should we care about the property right vs. liability rule debate? Even if transaction costs are zero, the legal rule still affects the division of surplus between the transacting parties. As I have shown elsewhere,² in cases where parties make investments ex-ante (i.e. at a time prior to transacting, so the investments cannot be subject to negotiation), the ex-post division of surplus affects their ex-ante incentive to invest. Thus, the choice between property rights and liability rules in IP should be made primarily in light of its effect on parties’ ex-ante incentives to invest rather than in view of ex-post transaction costs.

Be that as it may, García’s article eloquently reviews recent deals in which IP owners engaged in contractual and institutional innovation to contract around inefficient default liability rules and at the same time managed to assert greater control over their content. It is an enjoyable and worthwhile read that makes an important contribution to a longstanding and venerable debate.

The Price of Novelty and the Novelty of Price

Apr 16, 2020 Christopher J. BuccafuscoAdd a Comment

Patents exist to promote the progress of innovation, but a wealth of recent scholarship has demonstrated the ways in which patents influence the pace and direction of innovation in potentially problematic ways. The prospect of patent protection may cause innovators to focus on particular kinds of solutions to problems over others (i.e., those that can be patented), and it can cause researchers to focus on solving certain kinds of problems over others (e.g., those that offer the greatest opportunities for financial returns). In his new essay, Nicholson Price II describes another way in which patent law doctrines encourage certain kinds of innovations—“differentiating” innovations—over others—“deepening” or “exploring” innovations.

Borrowing insights from research on cumulative innovation and product differentiation, Price develops a taxonomy of different innovation strategies that researchers might adopt. They might focus on developing richer knowledge about existing technologies. Price calls this “deepening” innovation. Or, researchers might seek to take a large step beyond the existing field of knowledge. This is “exploring” innovation. Finally, researchers might opt for a middle strategy that does not produce substantial differences from existing approaches. These “differentiating” innovations do not take the great leaps that exploring innovations do, and nor are they intended to enrich our knowledge of existing solutions.

As Price explains, however, none of these strategies is necessarily better from the perspective of social welfare. Sometimes exploring innovation will prove more valuable, but in other cases deepening innovation can offer more utility, for example by developing richer knowledge about available technologies and their uses. But, he argues, patent law exhibits a strong preference for differentiating and exploring innovations over deepening innovations. Most obviously, the novelty and obviousness doctrines demand differentiation from the prior art in order to obtain patent protection. While some opportunities exist for patenting new uses of existing technologies, the incentive for this sort of R&D is much weaker than it is for differentiating innovations. Similarly, developing variations from prior art shields firms from the threat of infringement for their efforts.

Yet while patent law encourages variation, Price argues, it does not systematically favor exploring innovations over more modest differentiating ones. Patent law doctrine, including the mostly toothless utility requirement, does not specifically encourage large creative leaps forward (although they may receive broader scope). Instead, patent law relies on the noisy and biased signals of market participants to reward inventions. Patents provide value only to the extent that consumers purchase the products that innovators create. And very often, innovators can reap substantial returns by simply mimicking others’ products or tweaking their own. To demonstrate these effects, Price explores how patent law and the market affect the incentives of pharmaceutical and biotech firms, often resulting in “me too” drugs and “evergreen” patent portfolios that generate little social welfare improvement.

Price’s main points in the Essay are descriptive, rather than normative. He wants to show us that patent law’s novelty and obviousness doctrines have significant innovation costs, especially in their current formulations. Of course, Price acknowledges, differentiation has significant benefits. Having multiple options increases consumer choice, which is especially valuable if consumers have heterogeneous needs. And having multiple drugs treat the same condition can create some downward pressure on prices, although the evidence for this is decidedly mixed. But differentiation’s costs are also severe. Price focuses on three principal costs:

1. The costs of inventing around upstream products are high.
2. When products diverge from each other, they are less interoperable. This is costly for consumers who face higher switching costs (e.g. between different medical devices) and for innovators who are trying to work across incompatible products.
3. Differentiating innovation reduces society’s depth of knowledge about existing products, in favor of shallow knowledge about a broader range of products.

Price helpfully illustrates each of these three costs with examples from pharmaceutical and biotech innovation that appear to be decidedly suboptimal. He also adds a lot of nuance derived from his extensive knowledge of FDA regulation and insurance reimbursement.

The essay finishes with some ideas for how, if policymakers decide that innovation incentives are improperly skewed, they might intervene either in patent law (by increasing the obviousness threshold) or via the regulatory system (by giving FDA the power to limit approval for drugs that do not demonstrate meaningful improvements over the status quo). There might also be opportunities for insurance companies and government payers to play a role in directing innovation towards more socially beneficial outcomes.

While I understand Price’s choice to frame this essay modestly and descriptively, I believe that the available evidence points towards clear innovation failures in the pharmaceutical context. Recent empirical research indicates that as many as half of FDA approved drugs are no better than previous treatments, and as many of a quarter of them are actually worse.¹ Moreover, while I’m nervous about giving the FDA more power to engage in ex ante cost-effectiveness analysis as some European agencies do, I think there may be opportunities for patent law to affect firms’ incentives with ex post adjustments to patent duration and strength. In a new paper, Jonathan Masur and I lay out a few of these options. Neel Sukhatme and Gregg Bloche have recently explored similar ideas. Ultimately, however, Price’s essay is another characteristically thoughtful contribution to a hugely important field.

The Productivity Gains of Innovation Clusters

Mar 13, 2020 Lisa Larrimore Ouellette1 comment

Why do inventors increasingly locate near each other in metropolitan technology hubs like the Bay Area, Seattle, or Boston, despite the high costs of living in these areas? Just ten cities accounted for a remarkable 70% of computer science inventors in 2007. A leading view has been that “agglomeration economies” make researchers inside these innovation clusters more productive, although measuring this effect is difficult. In his new working paper, economist Enrico Moretti examines the location of U.S. patent inventors over time to estimate just how large these productivity gains are: If inventors were distributed uniformly across the United States, Moretti estimates that their overall patenting rate would decline by 11%.

I like this paper (lots) not because the result is counterintuitive—quite the opposite. Rather, in a field with so many barriers to real empirical progress, it is worth celebrating work that attempts to rigorously understand what factors actually affect innovation. And Moretti’s work on the geography of innovation has important lessons for law and policy scholars, including about the importance of looking outside IP for evidence-based innovation policies and the complex connection between innovation and growing wage inequality.

Moretti motivates the productivity effect of innovation clusters by examining the rapid collapse of Kodak in 1996, which led to a nearly 50% decline in the Rochester technology cluster. By 2007, patenting by the average non-Kodak and non-photography inventor in Rochester had declined by 20% relative to the typical inventor in other cities. In his main analysis, Moretti studies this effect for all U.S. patents filed between 1971 and 2007, classified into five technology areas—semiconductors, computer science, biology and chemistry, other engineering, and other science—and assigned by the inventors’ addresses into 179 “economic areas” covering the United States. As he notes, there are limitations to this data; for example, citation-weighted patent counts are imperfect proxies for innovation (particularly in computer science), and he only has observations for inventors in years when they patent. But having an enormous dataset with worker-level measures of both productivity and location allows Moretti to go beyond prior work in studying the productivity effect of agglomeration.

Using different empirical methodologies and robustness checks (such as separately examining effects on inventors who stay in a city while its cluster size changes, and using models based on instrumental variables), Moretti estimates how patenting productivity varies with cluster size. For example, the average computer scientist moving from the median cluster (Gainesville, FL) to 75th percentile cluster (Richmond, VA) has a 12% increase in patenting, with no evidence of a “pre-trend” that would predict this productivity increase before the move. And a Bay Area computer scientist patents 23% more than they would in an average cluster. Furthermore, Moretti finds that the patenting gains for large clusters outweigh the losses for small clusters, which is why smoothing out the distribution of inventors would lead to an overall decline in U.S. patenting by 11%.

This work has at least three lessons for legal scholars. First, scholars who care about innovation should pay attention to geography. Some do, of course; Camilla Hrdy has excellent articles on incentives for local innovation clusters, for example. But we still generally teach our students that IP law is the most important tool in the innovation policy toolkit, despite continued empirical uncertainty about whether stronger patent laws even increase aggregate research investments. Policymakers may have a more demonstrable effect on innovation by focusing on legal institutions affecting where people live—including laws affecting land use, taxation, and immigration.

Second, Moretti’s work highlights how innovation policy can’t escape discussions of inequality. This is not just an issue of IP reflecting a choice in how to allocate access to knowledge goods; IP and other innovation incentives also reflect transfers to innovative firms and individuals, and the returns to patenting are distributed unequally. In addition to persistent disparities by race, gender, and socioeconomic class, the geographic distribution of patents is increasingly concentrated in metropolitan areas and near research universities and is correlated with a region’s economic health. As Moretti explains, clustering is “important for overall production of innovation in the US” but also “may exacerbate earning inequality across US communities.” Colleen Chien tackles this issue in a terrific new working paper, in which she argues that the history of U.S. patenting supports both an optimistic view of increasing innovation and a pessimistic view “that the innovation pie has become increasingly unevenly distributed and centered on immigrants and coastal elites.” There is no easy solution, but more widespread acknowledgement of the problem is a first step.

The third lesson is the flip side of the second: just as innovation scholars shouldn’t ignore inequality, scholars and policymakers concerned about inequality should recognize that trying to lure tech firms to distressed cities with economic incentives likely comes with a real efficiency tradeoff. Greater improvements for both innovation and equality might be made, for example, by seeding promising tech hubs with federal funding, or by focusing on the land use policies that are currently limiting growth in places like the Bay Area. These policies wouldn’t just help tech workers: in earlier work, Moretti has concluded that “for each new high-tech job in a city, five additional jobs are ultimately created outside of the high-tech sector in that city,” and has argued that the federal government should provide greater relocation assistance to unemployed workers. Of course, that assistance is only helpful if there is actually affordable housing to relocate to. In January, for the third year in a row, the California Senate rejected a bill meant to stimulate housing production. California’s inability to increase the supply of housing is not just a failure for Californians facing homelessness and housing instability—it is also a failure for U.S. innovation. And based on Moretti’s work, policymakers have at least a rough quantitative measure of how much more productive researchers can be when they are able to locate in Silicon Valley and other technology clusters.

Intellectual Property Rights: A Destroyer as Well as a Creator of Jobs?

Feb 17, 2020 Pamela SamuelsonAdd a Comment

Do intellectual property (IP) rights create or destroy jobs (or both)? Industry associations and governmental agencies, such as the Patent & Trademark Office (PTO), frequently tout IP as a major force in creating (good) jobs as well as significantly contributing to economic growth. In 2016, the PTO, for instance, claimed that IP-intensive industries were directly or indirectly responsible for 45.5 million jobs, said to represent 30 percent of all jobs in the US. Without questioning this statistic, Professor Hrdy’s article explains that this is at best only one side of the story.

The main insight of the article is this: “Intellectual property may be partly responsible for job creation for people who work within IP-intensive industries . . . But a significant subset of innovations protected by IP, from self-service kiosks to self-driving cars, are labor-saving, and in many cases also labor-displacing” (emphasis in the original). The development and deployment of automated systems for performing a wide variety of tasks in a wide array of industries is “drastically reduc[ing] the amount of paid human labor required to complete a task.” Job losses resulting from technological change give rise to what economists call “technological unemployment.”

While some studies have concluded that the displacement of labor due to technological innovations has resulted in more job creation than job destruction, Hrdy questions whether this result will hold true in the near future owing to several factors. These include the increasing quality and pace of automation in various sectors, a decrease in quality of the work that remains unautomated, a rising inequality in who has what kinds of jobs, and the inability of education to keep pace with the needs of displaced workers.

Hrdy offers self-driving trucks as a case study. According to the American Trucking Association, there are currently 3.5 million professional truck-drivers in the US. The median salary of these truck drivers is about $40,000 per year. If the huge investments now being made in the development of IPR-laden self-driving trucks pay off, truck driving may no longer be a viable source of employment for most, if not all, of these people. The highly skilled engineers who are developing the software and hardware for self-driving vehicles generally earn more than $200,000 per year. But truck drivers cannot easily or quickly become engineers. Their jobs are going to be substantially displaced by automation. What is to become of these workers?

What should the US do about the labor-displacing impacts of technological innovation? The right answer is not the one Queen Elizabeth I chose when asked to grant exclusive rights to a knitting machine in 1589. She denied the inventor’s request for a patent because it would bring ruin to the many workers who made their living by hand-knitting clothing and other products. Elon Musk has suggested a universal basic income initiative as a solution. Bill Gates has proposed a robot tax. These and other possible solutions to the end-of-work problem brought about by advances in technology and IPRs on which these innovations are built are discussed in Hrdy’s fascinating paper. IP may not be wholly responsible for the end of work, but Hrdy says that it “magnifies the division of rewards between generators of IP and the workers whom their innovations replace.” Without destroying the incentive effects of IP, Hrdy would have us consider and address the distributive effects.

Whether you agree with Hrdy’s conclusions or not, this provocative article is well worth a read.

Plagiarize This Jot

Jan 13, 2020 Christopher J. Sprigman3 comments

Oscar Wilde: “That was an awfully good joke you made last night. I wish I could say it was mine.

James Whistler: “You will my boy. You will.”

Melvin Helitzer: One day Milton Berle and Henny Youngman were listening to Joey Bishop tell a particularly funny gag. “Gee, I wish I said that,” Berle whispered. “Don’t worry, Milton, [said Henny,] you will.”

Plagiarism is not a crime, or even a cause of action. But it is the “academic equivalent of the mark of Cain,” a curse that cannot be undone. Even an unsubstantiated accusation leaves an indelible stain, and a credible complaint cannot be countered. A plagiarist is an academic pariah, a transgressor of the highest law of the profession, the embodiment of the “great deceiver,” who leads everyone astray. Anything else can be forgiven, for the sake of the scholarship. Plagiarism tarnishes the scholarship itself, and leaves it forever suspect. If the purpose of scholarship is dowsing for truth, then the plagiarist is a liar who poisons the well from which everyone draws.

This is a jot recommending Brian Frye’s short, lively, and incisive article about plagiarism, Plagiarize This Paper. And, fittingly, everything you’ve read before this paragraph I’ve plagiarized from Brian’s work.¹

Or have I?

Brian wants others to copy his words, and even his ideas, and he doesn’t care whether we attribute them to him. He tells us that very clearly in his paper:

I explicitly authorize plagiarism of this article. I permit and encourage people to copy this article and republish it under their own name. I permit and encourage people to copy expressions from this article and use them without attributing them to me. And I permit and encourage people to use the ideas expressed in this article without attributing them to me.

Brian’s attempt to license plagiarism raises a fascinating question – what are norms against plagiarism really about? There are two, often entwined, interests that are usually identified as being protected by norms against plagiarism. First, it is often said that norms prohibiting plagiarism are there to protect authors’ interest in attribution. So, for example, anti-plagiarism norms would protect a junior academic against unacknowledged taking of her ideas by a senior academic who might otherwise have the power to get away with it. That seems like an interest worth protecting, though, of course, figuring out the origin of ideas with any precision is often difficult in reality. And second, anti-plagiarism norms are often characterized as protecting readers’ interest in not being defrauded – i.e., preventing a writer from fooling readers into thinking that a penetrating idea or a felicitous sentence is his when in fact it was invented by another. As a law professor who grades papers, this justification rings true to me – I want to know that the brilliant things that my students say in their papers are actually the product of their minds.

With respect to authors’ attribution interest, Brian has made clear that he doesn’t care about attribution, and that raises the question whether it’s fair to condemn as a “plagiarist” someone who denies to Brian what he doesn’t want. Brian is engaging in a provocation here: He thinks that people are likely to resist the notion of licensed plagiarism, which suggests that anti-plagiarism norms aren’t actually about protecting authors or readers, but about something else:

Ultimately, plagiarism norms are just cartel rules dressed up as moral obligations. Different discursive communities have adopted different plagiarism norms because they have different economic interests. And the plagiarism norms adopted by a community reflect the economic interests of its members. As those economic interests are contested and shift, the community’s plagiarism norms also are contested and shift. Accordingly, the plagiarism norms of any particular discursive community typically reflect the consensus interests of that community at that point in time.

I think Brian is on to something: If you look closely at institutional anti-plagiarism policies in the real world, it’s difficult to explain their actual content purely by reference to the interest of authors in attribution, or of readers in not being lied to. As an example of how anti-plagiarism norms tend to over-run both of these justifications, take my own institution, NYU. Its standard of “academic integrity” begins with a general definition of plagiarism as “presenting others’ work without adequate acknowledgement of its source, as though it were one’s own.” “Plagiarism,” the general definition continues, “is a form of fraud. We all stand on the shoulders of others, and we must give credit to the creators of the works that we incorporate into products that we call our own.”

This general definition that seems to focus precisely on the two justifications discussed earlier: (1) the interest of readers in not being lied to and (2) the interest of writers in receiving attribution when others are “standing on their shoulders.” These are valid justifications for anti-plagiarism norms, so all seems fair enough so far. But then the NYU plagiarism statement gets down to some examples of use by one writer of the work of another that would count as plagiarism:

• a sequence of words incorporated without quotation marks
• an unacknowledged passage paraphrased from another’s work
• the use of ideas, sound recordings, computer data or images created by others as though it were one’s own

These examples don’t mesh well with the two justifications. Take the first, “a sequence of words incorporated without quotation marks,” or the second, which bars unacknowledged paraphrasing. If you think about whether taking a “sequence of words” or paraphrasing without acknowledgment should count as either fraud on readers or unacknowledged shoulder-standing that hurts writers, you will realize rather quickly that the answer depends entirely on the particular words involved.

Let’s first imagine a student who takes these words: And then the day came, when the risk to remain tight in a bud was more painful than the risk it took to blossom.

Some people might recognize that these words comprise the entirety of Anais Nin’s short poem “Risk.” But some people won’t. If the student uses the words without acknowledgment, she will have defrauded readers who don’t recognize the source. We recognize the unacknowledged taking here as fraud on readers because of the effect of the specific words. They are lovely, insightful, richly metaphoric. Someone who doesn’t have Nin’s poetic oeuvre committed to memory might think that the genius of the words is the student’s genius, which it is not. And that gets to the other justification – the attribution interest of writers. Nin has a legitimate attribution interest here, precisely because this particular sequence of words is so highly creative The student is standing on a giant’s shoulders, but unless she says so, not everyone is going to see that. If anyone qualifies as a plagiarist, this student qualifies.

Now let’s imagine a second student who takes these words: When he lived in Chicago, Mojica sang in punk bands, ran a record label, and owned the Jinx Café and a video rental shop called Big Brother.

I doubt you will recognize these words, and there’s no reason anyone should. They comprise a sentence of workmanlike journalistic prose, reporting basic facts about an actual person. These words, and others like them, were the subject of a recent public plagiarism scandal involving former New York Times editor Jill Abramson, who was accused of copying (actually, mostly paraphrasing) in a book she wrote short passages from articles written by a number of less-well-known journalists. In fact, Abramson cited (in the book’s endnotes) virtually all of the articles she used. But in several places she failed to put language that she copied or paraphrased in quotations. For that, she was publicly flayed.

Did Abramson deserve to be labeled as a plagiarist? She did not commit fraud in the same sense as did the student who took Anais Nin’s words without acknowledgment. The words Abramson took are certainly useful to her narrative, but no one is going to give Abramson any credit tied to the particular words. As literary composition they are utterly banal. The words do report facts, but the facts are of little or no value in themselves. Abramson is not making off with someone else’s diamonds, nor is she standing on another writer’s shoulders – the words comprise, at most, a very small foot-stool. At bottom, Abramson is simply reproducing a sequence of words without telling you where they originated.

In the end, that’s what many plagiarism disputes are largely about. Not fraud on readers, or failing to acknowledge that one is “standing on the shoulders” of another writer, as the NYU general definition of “plagiarism” claims, but the mere reproduction of words. That is also what copyright law is about, and one might ask why we extend plagiarism norms beyond their core when we have copyright to police mere word-taking. Brian has an explanation:

Essentially, academic plagiarism norms are the equivalent of a tax imposed on junior scholars, for the benefit of senior scholars. Junior scholars must err on the side of attributing ideas to senior scholars, whether or not attribution is accurate or helpful, on pain of suffering a plagiarism accusation. As a consequence, senior scholars collect “interest” on the intellectual capital of junior scholars.

That’s a big idea, one that extends beyond ideas to works, and also beyond academia to journalism and other places where enforcement of anti-plagiarism norms is particularly fierce. And Brian deserves full credit for this idea. Even if he doesn’t want it.