Does the U.S. Patent and Trademark Office grant too many bad patents? Evidence from a quasi-experiment.

Author:Frakes, Michael D.
Position:Introduction through IV. Data and Methodology, p. 613-644

INTRODUCTION I. HARMS FROM PTO OVERGRANTING AND INSUFFICIENCY OF EVIDENCE A. Overgranting Explanations and Harms B. In Search of Empirical Evidence II. A NOVEL THEORY FOR WHY THE PTO IS BIASED TOWARD ALLOWING PATENTS A. Never-Ending Stream of Repeat Filings B. PTO Objectives III. THE DIFFERENTIAL IMPACT OF THE PTO'S INCENTIVES TO ALLOW PATENTS A. Varying Propensity in the Filing of Continuation Applications B. Per-Claim Allowances Versus Application Allowances IV. DATA AND METHODOLOGY A. Data B. Methodology 1. Basic framework 2. Expansions beyond basic framework 3. Causal assumptions 4. Resource sustainability metrics V. RESULTS A. Primary Results: Graphical Analysis 1. Description of graphical approach 2. Description of findings 3. Interpreting the magnitude of the findings 4. Related support 5. Inclusion of control variables 6. Number of technology groups B. Additional Robustness Checks 1. Overview 2. Substantive patent law changes 3. Selection-effects concerns 4. Difference-in-difference-in-difference 5. Alternative resource-health metric 6. Other C. Results: Concluding Remarks VI. THE MECHANISM BY WHICH THE PTO PREFERENCES GRANTING CERTAIN TYPES OF PATENTS OVER OTHERS VII. IMPLICATIONS AND WAYS TO REDUCE THE PTO'S INCENTIVE TO GRANT PATENTS A. Implications for Patent Policy B. Reducing the PTO's Incentives to Grant Patents in High-Repeat-Filing Technologies CONCLUSION INTRODUCTION

In recent years the patent system has come under trenchant criticism. There is a general consensus that the system is both broken and a failure. (1) Many believe the root cause of the patent system's dysfunction is that the U.S. Patent and Trademark Office (PTO or Agency) is issuing too many invalid patents that unnecessarily drain consumer welfare, stunt productive research, and unreasonably extract rents from innovators. (2) Such sentiments have even been the subject of multiple reports by the National Academies and the Federal Trade Commission. (3) Concerns regarding the Agency's overgranting tendencies have been so pressing that they have spurred the Supreme Court to take a renewed interest in substantive patent law, (4) while likewise driving Congress to enact the first major patent reform act in nearly sixty years. (5)

Despite the centrality of patent quality to the health of the patent system and the plethora of reasons put forward as to why the PTO may be biased toward allowing patents, there exists little to no compelling empirical evidence that the Agency is actually overgranting patents. (6) To date, the debate regarding patent policy has been driven by anecdotes of a few infamously issued patents or data that alone are insufficient to deduce that the PTO is overly permissive in its granting decisions. (7) As a result, policymakers have been modifying the system in an effort to increase patent quality in the dark. Without guidance as to which features of the system may actually be causing the PTO to overgrant patents or even compelling evidence that the Agency is biased toward allowing patents, policymakers are left trying to fix the patent system without even understanding the root causes of the system's shortcomings.

This Article begins to rectify this deficiency and advances the conversation along two dimensions. First, it describes a novel theoretical source of bias that may cause the Agency to grant too many invalid patents. This newly addressed source of bias stems from an oddity of the U.S. patent system--the inability of the PTO to reject a patent application with finality. The capacity of aggrieved patent applicants to continuously restart the examination process upon rejection by filing repeat applications can potentially overwhelm the existing examination infrastructure. The fact that the Agency currently faces a crushing backlog of over 600,000 patent applications, of which close to forty percent constitute repeat filings, suggests this has already occurred. The PTO can effectively turn off the spigot of repeat filings by allowing--i.e., granting--patents earlier in the examination process. (8) That is, the PTO could attempt to decrease the incentives of applicants to file repeat applications (and hence concomitantly decrease its backlog of applications) by biasing its grant rate upward.

To be clear, repeat filings have been the subject of a growing academic discourse. (9) These debates, however, have mostly focused on applicant behavior rather than Agency incentives. This Article aims to build on and fill various gaps in the literature by refocusing the debate on the PTO's inducements.

Second, this Article attempts to explore, through a sophisticated methodological design, whether the Agency is in fact operating on this incentive and actually overgranting patents. We begin this exercise by setting forth a theoretical model that predicts that the PTO will only inflate its granting tendencies in order to diminish repeat filings during times in which the PTO faces binding resource constraints--that is, during times in which it cannot meet its expected examination demand with available resources. To empirically test whether a resource-constrained PTO will grant additional patents, we amassed a rich database of previously unavailable patent data through the filing of Freedom of Information Act (FOIA) requests to the PTO. This novel database includes patent processing statistics on all 4,733,263 patent applications filed with the PTO over an approximately twenty-year period. (10) We then employ a natural experiment framework, wherein we compare the Agency's grant rate before and after a period of time in which the PTO experienced deterioration in its net resources. Of course, grant rates may change over time for a number of reasons unrelated to the story at hand; for example, the quality of the underlying applications may change from year to year. As such, we do not rely solely on this before-after comparison as a mechanism to isolate the effect of repeat filings on the Agency's grant rates. Rather, we explore how a negative resource "shock" to the Agency differentially affects a treatment group and a control group, much like a randomized medical experiment would. For the purposes of this design, our treatment group is represented by those technologies that have historically exhibited a stronger disposition to partake in the costly activity of filing repeat applications. It is with respect to these applications that the PTO would be especially inclined to act on the above-specified incentive. Likewise, our control group constitutes those technologies that have not historically exhibited such a tendency and with respect to which the PTO may be less inclined to inflate its grant rate upon experiencing a downturn in its resources that leaves the Agency unable to process all of the patent applications that are awaiting review.

By subtracting the granting experiences of the control group from the treatment group over this relevant period of time, this methodology allows us to hone in on the effects on the Agency's grant rate stemming from the theorized incentive itself. In other words, this quasi-experimental framework (supplemented by a number of robustness exercises) enables us to better make causal inferences--that is, to more convincingly trace the Agency's observed increase in grant rates to the potentially biasing feature itself and thus to rule out potentially confounding explanations for the documented grant-rate patterns.

We find evidence that the PTO is indeed overgranting patents during times in which the Agency lacks sufficient resources to meet its expected demand for examination. Moreover, our findings suggest that the PTO is preferentially granting those patents it stands to benefit the most from--those in high-continuation-filing-rate technologies, such as information and communication technologies, which include software, business methods, and information storage, and health-related technologies, which include surgical and medical instruments and genetics.

What are we to make of the evidence that the PTO is granting patents in an effort to diminish repeat filings? On the one hand it suggests that policymakers are generally correct in suggesting that the Agency allows too many bad patents. Importantly, however, it also suggests that policymakers have been barking up the wrong tree. By ignoring the important role that repeat filings play in influencing PTO decisionmaking, legal scholarship has overlooked a substantial source of Agency bias. As a result, the recent "fixes" to the patent system to improve patent quality fail to address the substantial overgranting bias identified in this Article. (11) Additionally, and equally as worrisome, policymakers have also failed to appreciate that the PTO may not just be biased toward allowing patents in general but also be biased toward allowing particular types of patents. Accordingly, recognition of the harms from the granting of bad patents has been underinclusive. To the extent that the PTO is extending preferential treatment to technologies based on backlog incentives and not based on legitimate social interests in intervening in certain industries, the Agency may also be undesirably distorting the allocation of resources across different sectors of the economy. That is, because the PTO is preferentially granting patents in high-continuation-rate technologies, such as information and communication and health-related technologies, society may be overinvesting resources in these technological fields.

The rest of the Article is organized as follows. Part I briefly describes the primary complaint that scholars and stakeholders have registered against the patent system: the PTO issues too many bad patents. This Part also describes how the literature to date has failed to produce any persuasive empirical evidence demonstrating that the Agency is in fact biased toward allowing patents. Part II theorizes the PTO's possible incentive to grant...

To continue reading