By Janet Freilich, Michael Meurer, Mark Schankerman, and Florian Schuett
Forthcoming U.C. Irvine Law Review
Patents are supposed to cover new and innovative inventions, so why are there patents on old or obvious creations such as a stick, a method of swinging on a swing, and bread with the crust cut off? The Patent Office regularly grants patents on inventions that should not have been patented, with detrimental and widespread consequences for social welfare. This problem has spawned an extensive scholarly literature and dozens of Congressional bills proposing policies to improve patent quality. The content of these proposed policies varies significantly—from strategies to make patent applications more expensive to file, to increasing the intensity of examination at the Patent Office, changing substantive legal doctrines of patentability, reducing patent term, increasing the cost of maintaining granted patents, easing the process of reviewing patentability after a patent has been granted, and altering procedural and remedial aspects of litigation. Even with creative thinking on this topic, scores of wide-ranging policy proposals, and appetite for political reform, patent quality remains a problem and there is no consensus on the best solution(s).
Despite its insights, patent scholarship may be partially at fault for this morass of failed reform efforts. Reform advocates systematically make several crucial errors that render existing policy predictions unreliable, misleading, or outright wrong. First, overly simplistic analyses of policies focus on the impact of reform on just one part of the patent system—but reforms generally have consequences for multiple parts of the system and affected parties will react along multiple margins, not merely in the specific area of change. To illustrate, imagine a policy proposal to improve litigation procedures that accurately predicts that the policy will ameliorate litigation outcomes but fails to consider detrimental changes to a second area of the patent system—for instance, decisions to file patents. If the unaccounted-for negative effects outweigh the positive, the proposal will be counterproductive. A second critical error in existing scholarship is that it does not seek to quantify the magnitude of a policy’s impact. Reforms generally have multiple effects with both positive and negative effects on social welfare—without quantification, policy makers cannot know whether overall the reform will be helpful or harmful.
Our recent paper provides not only new solutions, but a new approach for designing and assessing policies both in patent law and legal systems more generally. Using a perfect Bayesian equilibrium model and calibration methods developed by Schankerman and Schuett we move beyond existing scholarship to (1) determine how a policy change to one part of the patent system affects the system as a whole; and (2) quantify the impact of policy changes.
We apply our model to several of the most prominent policy debates in patent law. We conclude that certain reforms such as regulation of settlement licenses and increased examination intensity yield large gains in social welfare and should be prioritized. Other reforms that are popular with scholars, including decreasing the availability of injunctions and reducing litigation costs produce surprisingly small gains in social welfare. Often existing scholarship operates too much on intuition, which, we show, can be wrong. Our new approach to patent reform provides an approach that offers deeper understanding and a more effective evaluation framework.
As an example of the value of an integrated, equilibrium framework with quantification, we assess the impact of the Supreme Court’s Actavis decision which regulates the terms of patent litigation settlement agreements, making it more difficult for litigants to agree to halt or circumvent court proceedings. The Actavis reforms are a popular topic of scholarly analysis, much of which focuses on the first-order conclusion that when parties have unrestricted ability to settle, owners of low-quality patents can settle to stop litigation, which ends the case and prevents courts from reaching a decision that would invalidate patents. Consequently, patent quality would improve if settlements by the owners of low-quality patents decline, and courts invalidate more of these patents. This analysis is clearly important, but only a beginning—we must trace other equilibrium responses at different stages of the patent system. Using our model, several other responses of interest to policy makers become clear. In particular, the cost of patent enforcement via litigation increases, which may decrease incentives to invent.
When these direct and indirect effects of settlement restraints are considered, the direction of the policy’s effect on patent quality becomes ambiguous. In some respects—such as the ability of courts to invalidate bad patents, for instance—restraints on settlement should improve patent quality. In other respects—such as decreased incentives to invent—restraints on settlement might reduce patent quality by disproportionately deterring high quality inventions. These countervailing effects reveal the importance of quantification. Our calibrated model can be used to estimate the magnitude of the policy’s effects. As we show, despite the presence of both positive and negative effects of restraints on settlement on quality and social welfare, the ultimate effect of the policy is large and positive: a 3.6% gain in welfare.
The generalizable point—and a key contribution of our paper —is that the patent system (indeed any legal system) is complex, and good policy analysis should recognize that affected parties will adjust their behavior, potentially in disparate parts of the system. Thus, effective analysis requires an understanding of reform consequences across the patent system coupled with quantification of the direct and indirect effects of reform.
In addition to insights into specific policy reforms, we also offer a significant reformulation to how scholars approach the theory of patent reform. Much patent scholarship is bifurcated into those favoring “early” reforms—fixes at the Patent Office to screen out low-quality applications—and those favoring “late” reforms—changes to litigation so that courts could invalidate more low-quality patents. Our method of assessing patent reform gives us theoretical and empirical reasons to reject this polarized “before-or-after” approach. First, we show that reforms implemented after patent grant can influence behavior before patent grant; “after” reforms therefore are also “before” reforms, and vice-versa. Second, our model allows us to predict the effect of varying multiple policies at once; doing so shows that “before” and “after” policies can be complementary. Finally, our quantified model indicates that increased examination intensity, a pre-grant reform, and regulation of settlement licenses, a post-grant reform, both promise to increase social welfare significantly, while other pre- and post-grant reforms offer little or no gain in terms of social welfare. It is not, therefore, that reforms before grant are categorically better (or worse) than reforms after grant, but rather that both types of reform can be helpful, unproductive, or worse, counterproductive.