In the run-up to Oil States, a frequent criticism by opponents of patent reform was that the PTAB was “unfair” and that it invalidated patents at a rate far higher than the district courts would if they had the chance.
Unfortunately this is contradicted by the actual facts, which is something we’ve seen before from opponents of patent reform. (See, e.g., the Chamber of Commerce rankings, the Alice ‘storm’, and misleading testimony to Congress regarding conflicts between PTAB decisions and district court decisions.)
The truth is that the PTAB, when faced with a patent, has outcomes roughly similar to comparable procedures both in the U.S. and overseas—when they make a decision, they find that a little more than forty percent of patents are invalid. The PTAB just makes it significantly cheaper to get to that point, allowing small and medium businesses to defend themselves when patent trolls come calling, instead of being forced to pay because they couldn’t afford to defend themselves.)
We’ll start our comparative tour of patent invalidity in Europe, before swinging through the U.S. district courts and ending up back at the PTAB.
EPO Oppositions and Nullity Proceedings
In Europe, rather than inter partes review and district court litigation on validity, there are opposition proceedings at the European Patent Office (EPO) and nullity proceedings in European national courts.
Oppositions are the most similar1 European procedure to U.S. post-grant proceedings like IPR. So it’s instructive to examine how many European patents are opposed and the outcomes of those oppositions.
Fortunately, the EPO makes available detailed statistics on exactly these questions. Approximately 4% of European patents are opposed, and of those, approximately 32% of oppositions are not instituted.2 27% of oppositions result in a final decision invalidating all claims of the patent, while the remaining 40% of patents are modified in some form and maintained as patents.3 (EPO opposition statistics appear to omit patents where the parties settled mid-proceeding.) Each year, the EPO handles approximately 4,000 oppositions, meaning that approximately 1,100 European patents are invalidated in their entirety on a yearly basis.4
Currently, nullity proceedings occur in national courts (most frequently the German courts). These somewhat resemble an invalidity claim in a U.S. court, though with significant differences regarding court procedures and the composition of the relevant judicial body. Data on European court proceedings can be difficult to obtain; as a result, it can be difficult to estimate outcome rates for nullity actions. One study, examining German nullity data,5 found that—excluding cases that settled or were otherwise withdrawn—24% of challenged patents were entirely valid, 37% were partly valid and partly invalid, and 39% were completely invalid.
While inexact because of these two separate systems, a split of 33%/33%/33% is a good rule of thumb for outcomes in European patent challenges. One third of the time the patent is entirely valid, one third of the time it’s entirely invalid, and one third of the time you get a mixed result.
U.S. District Courts
While Oil States has made clear that neither an Article III judge nor a jury is required when invalidating a U.S. patent, district court judges still decide validity issues on a regular basis. The most comprehensive study of outcomes I am aware of, authored by Professors John Allison, Mark Lemley, and David Schwartz, examined every merits decision on every patent case filed in the years 2008 and 2009. This data set included decisions at all stages, from summary judgment to jury decisions. And the result?
When a patent is challenged as invalid, district courts agree that it was in fact invalid 42.6% of the time.
So when courts in the U.S. look at patents, just like European courts, it appears that a little more than 40% of the time, the court finds the patent should not have been granted.6
That brings us to the PTAB.
The PTAB provides regularly updated, highly granular statistics on case outcomes. Their waterfall chart gives the total breakdown of all filed petitions. The below chart, taken from the USPTO’s February 2018 statistics, shows the per-petition breakdown.
Removing PTAB Data That Isn’t About What The PTAB Does
Obviously, petitions that are still open aren’t relevant to determining PTAB outcomes on a percentage basis. Similarly, petitions which have been joined to other petitions will never receive a separate judgment; if we care about what the PTAB chooses to do, it’s irrelevant whether they’re deciding on one petition or three joined petitions, it remains a single decision.7
As a result, those categories should be removed from the data. But there are two other categories that need to be removed because they relate to the behavior of the parties, not of the PTAB. Settlements may occur because the patent is clearly valid, or clearly invalid, or because the parties reach an amicable agreement; it’s essentially impossible to determine what the outcome of a settled petition would have been if the PTAB had examined the case. As a result, settlements—and, for the same reasons, requests for adverse judgment—aren’t relevant to statistics on the decisions made by the PTAB.8
Understanding Decisions Reached By The PTAB
After removing these categories, we receive a better picture of the PTAB’s actions.
There are 2,016 final decisions, 123 dismissals, and 2,000 denials of institution. That’s 4,139 total decisions.
Of those, 2,000 found the patent valid at the first hurdle of institution. Another 123 dismissed the proceeding without affecting the patent. And, of the final decisions, another 389 found all claims patentable.
Out of 4,139 decisions made by the PTAB, 2,512 (61%) didn’t touch a single challenged claim. 9
Petition vs. Patent
These numbers are on a petition-by-petition basis, not a patent-by-patent basis. Since only one petition needs to invalidate a claim for the claim to be invalidated entirely, the patent-by-patent numbers will generally be higher than petition-by-petition; the PTAB’s own statistics illustrate this phenomenon. However, a patent may require more than one petition in order to find all claims invalid (most famously in the case of a single patent with 306 claims).
While this makes it more difficult to estimate a per-patent rate for PTAB decisions, it is possible to make a very rough estimate. The PTAB statistics show that on a per-patent basis, 58% of patents are unchanged, 6% of patents have a request for adverse judgment, 7% of patents have some claims found unpatentable, and 29% of patents have all challenged claims found unpatentable (compared to 66.5%, 4.5%, 5%, and 24% on a per-petition basis). From this, combined with the PTAB’s estimate that per-patent institution rates are around 6% higher than per-petition institution rates, a fair estimate is that of the 58% of unchanged patents, around 25% are due to denials of institution and another 7.5% are due to final written decisions upholding the patent entirely, with the remaining 25.5% due primarily to settlements.10
Again excluding settlements and requests for adverse judgment, 28% of challenged patents are invalidated by a PTAB decision, out of the approximately 60.5% of challenged patents that are disposed of based on PTAB decisions (rather than decisions of the parties.) This results in an estimate that approximately 46.5% of challenged patents are invalidated in whole or in part by a decision of the PTAB. 11
A Little More Than Forty Percent
The end result? Whether you challenge a patent at the PTAB, in district court in the U.S. or in court in Europe, or even at the EPO, around 40-45% of all challenged patents are determined to be invalid in whole or in part. The PTAB is making the same kinds of decisions, in about the same numbers, as every other body charged with reviewing patent validity.
- Though certainly not identical. For example, EPO oppositions can be based on subject-matter eligibility challenges, a ground that is not available in IPR. ↵
- For the sake of coherent terminology throughout, I’ve used U.S. equivalent terms—a proceeding with a threshold requirement is either instituted or not instituted at the threshold, rather than using the European terms such as rejecting an opposition. ↵
- These statistics vary slightly year-to-year, but the roughly 30/30/40 split is consistent. ↵
- For comparison, the PTAB receives approximately 1,500 petitions per year, some of which are directed to the same patent and a number of which are never instituted. The EPO opposition procedure invalidates significantly more patents on a yearly basis than IPR does. ↵
- German data is particularly significant as the German court system is likely the most frequently used European court system for patent disputes. ↵
- This does include district court judgments of invalidity based on § 101 and § 112, which are not available to PTAB petitioners. However, PTAB petitioners choose whether to bring their cases to PTAB—and frequently decline to do so—while district court challengers are not generally given the option not to be in court and are bringing an invalidity defense as a matter of course. The slight reduction to be expected from the unavailability of § 101 and § 112 in the PTAB is balanced by the selection bias in PTAB towards picking weaker patents to challenge. ↵
- Joined proceedings also may have been joined to a proceeding that held the patent valid or invalid, making it difficult to assign a value here. However, even in the worst case, joined petitions only minimally change the outcome. ↵
- In addition, as the statistics for EPO proceedings and district court decisions are measured with settlements excluded, settlements must also be excluded here to ensure commensurability. ↵
- Even if every single joined petition had been joined to one of the 39% of decisions that affected at least one challenged claim—an unlikely circumstance, compared to the likelihood that they are more evenly distributed across decisions—the numbers shift only slightly, becoming 55% and 45%. ↵
- This is consistent with the rough breakdown on a per-petition basis, which is unsurprising given that the vast majority of patents only face one or two petitions. ↵
- This is also consistent with one possible worst-case boundary estimate. Around 79% of petitions are filed either as single petitions or on nearly the same day (meaning that they generally will cover different claims from one another). If we assume that none of the 21% of potentially duplicative petitions are joined, and are all denied institution—an unrealistic worst possible case assumption—the adjusted patent-by-patent invalidity estimate would become 1,627 at least partially invalid patents out of 3,269 adjudicated patents; in other words, a reasonable worst-case bound for the rate at which decisions made by the PTAB invalidate patents would be approximately 50%. ↵