IPR Statistics Revisited: Yep, It’s A Patent Killing Field

Some view Inter Partes Review (IPR) as a patent killing field. Yet, the statistics published by the PTO seem to suggest that the patent kill rate in IPR is modest, and other commentators suggest that it’s roughly comparable to the kill rate in district court. The truth is closer to the killing field. Whether this will change under Trump remains to be seen, but there are reasons to believe it might.

There are four reasons why the killing field view is more accurate. First, the PTO’s reported numbers are somewhat misleading. When reporting IPR numbers, the PTO reports on petitions “completed to date.” This little fact, as others have noticed, makes the kill rate seem lower than it is.

Second, not only are the kill rates under 102 and 103 higher in IPR than in district court, but also those who suggest the kill rates are roughly comparable compare apples to oranges. In IPR, only printed publications can be used to attack patents. In district court, challengers can also attack with the other categories of prior art: prior public use or sale, public knowledge, derivation or incorrect inventorship, and invention by another. In district court, these other categories account for about 40% of kills under 102.

Third, although the institution rate has dropped since the early days of IPR, some of the drop is attributable not to a decrease in the Board’s willingness to institute any given claim but to an increase in the willingness of patentees to settle IPRs before the institution decision.

The fourth reason that the killing field view is more accurate concerns the rate at which parallel litigation is stayed, which does not speak to the kill rate but does speak to the overall power that IPR confers on patent challengers. Some commentators report a grant rate for stays that is misleadingly low because it’s based on all motions to stay filed on the ground of a co-pending IPR. It’s somewhat common for an accused infringer to file one motion before institution and one after and have the first fail and the second succeed. When calculated per motion filed, the grant rate is only around 50% to 60%. But when calculated per case, which is more meaningful, the grant rate is closer to 82%.

PTO Numbers Are Misleading

Below is a diagram from the PTO report on AIA Trial Statistics, current through 2016. It shows the disposition of IPR petitions completed to date. Take a quick look at the light blue hexagon at the top and the red and yellow hexagons at the bottom.

The light blue hexagon shows the total number of petitions filed from the inception of IPR through 2016. The red hexagon shows the number of trials in which all instituted claims were held unpatentable and also states it as percentage of total petitions filed. The yellow hexagon shows the number of trials in which some instituted claims were held unpatentable and likewise states it as a percentage of total petitions filed.

Add the 23% in the red hexagon to the 5% in the yellow hexagon and we get 28% as the percentage of petitions in which the Board invalidated at least one claim. In addition, as shown in two of the gray hexagons, patentees requested adverse judgment after institution in 191 IPRs and before institution in 26 IPRs, resulting in another 5% of kills, for a total kill rate of 33% (28% + 5%).

The real kill rate, however, exceeds 33%. The reason is that after a petition is filed it takes about 18 months to get to trial and decision. Accordingly, the trial numbers in the red and yellow hexagons relate to petitions filed 18 or more months prior to the last day of 2016. Yet, in calculating the percentages in the red and yellow hexagons, the PTO divided by the number of all petitions filed as of the last day of 2016, instead of by the number of petitions filed as of 18 or more months prior to the last day of 2016.1 Similarly, the requests for adverse judgment in the gray hexagons relate to petitions filed at various times prior to the end of 2016.

One study that accounts for the delay until final decision is the Report on Patent Trial and Appeal Board Contested Proceedings (2016), prepared by Fitzpatrick, Cella, Harper & Scinto. This study looked at a set of IPRs consisting of those denied institution for substantive reasons through 2014 plus those that reached final written decision or request for adverse judgment through 2015. Among this set of IPRs, 60% of claims challenged in petitions were killed either directly by the Board or by way of request for adverse judgment by the patentee.

The Fitzpatrick study states the kill rate as a percentage of claims rather than as a percentage of IPR petitions. Arguably, the latter is more meaningful because the broader and more valuable claims in a challenged patent are more likely to be killed. For example, suppose all 20 claims in a patent are challenged. The two broadest claims cover a swath of infringers, while the remaining 18 claims cover few or no infringers. The two broadest claims are killed; the other 18 survive. In substance, this largely or entirely amounts to a loss for the patentee, but if we look only at the percentage of claims killed, it appears to be a 90% victory for the patentee.

On the other hand, the numbers in the red and yellow hexagons above tell us that, when the Board holds claims unpatentable, it holds all of the instituted claims unpatentable 82% of the time (23% / 28%). Similarly, a study by Brian Love and Shawn Ambwani found, that among IPRs reaching final decision, 77% of all instituted claims are either held unpatentable or disclaimed.2 If the norm when killing instituted claims is to kill them all, then stating the kill rate as a percentage of claims falls only a bit short of stating it as a percentage of petitions.

In any event, other data in the PTO report on AIA Trial Statistics suggests that the kill rate for petitions is about four percentage points higher than the kill rate for claims. If we adjust by adding 4% to the 60% from the Fitzpatrick study, we get an estimated kill rate per petition of about 64%. That is, in about 64% of petitions, one or more of the claims will not survive.

Kill Rates in IPR Greatly Kill Rates in District Court

The Fitzpatrick study finds that, of challenged claims that reach a decision on the merits, 49.6% are invalidated under 103 and 41.1% under 102. (Note that some claims are invalidated under both.) The study notes further that, in district court cases in which validity is adjudicated, 27.8% of claims are invalidated under 103 and 31.1% under 102. (Again, some claims are invalidated under both.) According to these numbers, the potency of 103 is much greater in IPR than in district court, while the potency of 102 is modestly greater.

In IPR, however, the only category of prior art available is printed publications. In district court, available prior art also includes prior public use or sale under 102(b), public knowledge under 102(a), derivation or incorrect inventorship under 102(f), and (for pre-AIA patents) invention by another under 102(g).

It is remarkable that, although printed publications are the only type of art available in IPR, the kill rate under 102 is nonetheless about a third higher than in district court. In district court, printed publications account for about 60% of invalidations under 102; the other art categories account for the remaining 40%.3

If we take 60% of the 31.1% kill rate under 102 in district court, we get a kill rate under 102 printed publications in district court of only 18.7%. Now, we can compare apples to apples and see that the 18.7% kill rate for printed publications in district court is dwarfed by the 41.1% kill rate for printed publications in IPR. (Note that in IPR attack under 103 is likewise limited to printed publications, but this limitation is far less significant for 103 because the vast majority of 103 attacks have always been based on printed publications.)

The Drop in the Institution Rate Is Due in Part to Rise in Settlement

A patentee may request adverse judgment or settle by dismissing a co-pending infringement suit against the petitioner or by entering into a covenant not to sue. These measures occur in the shadow of what the patentee believes the Board would likely hold were it to rule on the challenged claims. When IPR first started, its institution and kill rates were very high and IPR quickly developed a reputation as a killing field.4

In what appears to be a reaction, patentees became more willing to settle. In 2014 there were more than five times as many settlements before institution as there were in 2013. In 2015 there were, compared to 2014, more than two and half times as many settlements before institution and 82% more after.5 Accordingly, some of the decrease in the institution rate is due not to a decrease in the Board’s propensity to institute any given claim but to an increase in the settlement rate.

The institution rate dropped in part because some of the IPRs the Board would have instituted settled before they could be instituted. According to the PTO’s AIA Trial Statistics, among IPRs that reached the institution decision, the Board instituted 86.5% of challenged claims in FY2013, 74.3% in FY2014, and 65.3% in FY2015. We can estimate the upper limit of the extent to which pre-institution settlement accounts for the decreases in institution across those years by (1) seeing what the institution rates would have been had all of the IPRs that settled pre-institution instead been instituted and by (2) comparing the changes in the counterfactual institution rates from year to year to the changes in the reported institution rates from year to year.

The counterfactual institution rates come out to 87.8% in FY2013, 77.5% in FY2014, and 71.6% in FY2015. When we compare these to the reported rates, it appears that pre-institution settlement accounts for only as much as one-sixth of the decrease in institution from FY2013 to FY2014. Pre-institution settlement, however, accounts for as much as two-thirds of the decrease in institution from FY2014 to FY2015. That is, the difference between the counterfactual rates for FY2014 and FY2015 is only one third as large as the difference between the reported rates for FY2014 and FY2015.

Patentees may settle to preserve the validity of the claims challenged in the IPR. In addition, when the IPR challenges fewer than all of the claims in the patent, the patentee may settle to minimize the effect on the unchallenged claims. The patentee may also settle to minimize the effect on related patents.

By settling before the patentee’s preliminary response, which comes due about three months before the institution decision, the patentee can avoid taking a position on the prior art or on claim construction. By settling before the institution decision, the patentee can avoid a decision by the Board to institute that suggests in and of itself, or that includes language that suggests, that the claims are invalid. The patentee can also avoid claim construction by the Board that could conflict with the construction the patentee may want to advance in a future infringement suit.

To be sure, however, settlement in view of the threat posed by IPR amounts to some degree of loss for the patentee. Although settling officially preserves the validity of the patent, the fact that the patentee chose to settle puts other infringers and licensees on notice that the patent is vulnerable. Furthermore, the IPR file, a public record, sets forth the prior art to attack it with. And of course when a patentee settles an IPR by dismissing a co-pending suit against the petitioner, the patentee foregoes some or all of what it would have gained had it maintained the suit.

Co-Pending Suits Are Usually Stayed

Some commentators report a grant rate for stays that is based on all stay motions filed before and after institution of a co-pending IPR. When based on all filed stay motions, the grant rate appears to be around 60% or so. Motions filed after institution, however, are more likely to succeed. When the grant rate is based on stay granted per case, which is more meaningful, the grant rate is higher.

According to Love and Ambwani, in 80% of instituted IPRs the challenged patent was being asserted in litigation between the patentee and the petitioner. Among the co-pending suits in which a motion(s) to stay was filed and ruled on, stay was granted 82% of the time. (About 10% of the motions were never ruled on because either the case first settled or it was stayed for some reason other than the IPR.)

Note, however, that there was variation across the district courts. Love and Ambwani found that the Eastern District of Texas granted stay in view of IPR in only 56% of cases. Another source reports that the E.D.Tex. grants stay in view of IPR in only 50% of cases.6


In the 13 years that inter partes reexamination was available, only 1,919 were ever filed. In contrast, in the four and a half years that IPR has been available, over 5,600 have been filed. The popularity of IPR with patent challengers is no surprise given the statistics above.

Can we expect changes under the Trump administration that tilt the scales back to patentees? After all, it can be argued that investment in technology may be diverted to countries more likely to uphold patents. One could point out, for example, that China recently embraced software patents and that in 2016 the Chinese patent office received nearly twice as many patent applications as the USPTO.7

Although Trump’s views on patents are unknown as of yet, a new report from Shearman & Sterling suggests that experiences with his own IP (at least in the context of branding) and the views of those who have his ear, many of whom are staunch supporters of patents, would tend to make him pro-patent.8 There is, however, no reason to believe that patents are a priority for Trump at this point in time.

  1. See also Michael E. Sander, By The Numbers: Is the PTO Underreporting the Rate They Institute IPRs and CBMs? Patentlyo (May 16, 2016).
  2. See Brian J. Love and Shawn Ambwani, Essay: Inter Partes Review: An Early Look at the Numbers, 81 University of Chicago Law Review Dialogue 93 (2014).This study was limited to an early subset of IPRs – those terminated by March 31, 2014.
  3. This 40% is derived from data on 3316 cases decided from 2005 to 2013, which were coded according to rulings on 40 legal issues. For the raw data, see U.S. Patent Litigation Statistics, Institute for Intellectual Property and Information Law, University of Houston Law Center, at www.patstats.org (last visited Jan. 2017).
  4. See Michelle Carniaux and Michael E. Sander, Riding the PTAB Wave: An Empirical Study of the Surge of Petitions Filed in the Patent Trial and Appeal Board, available at SSRN (April 1, 2015) (note Fig. 5 and Fig. 7, showing high institution rates and kill rates early on).
  5. See USPTO AIA Trial Statistics, as of December 31, 2014 and as of December 31, 2015.
  6. See Statistics for Orders for Motions to Stay in View of Pending IPR or CBM as of April 10, 2014, available at Kenyon.com (Jan. 2017).
  7. See Glyn Moody, China Files A Million Patents In A Year, As Government Plans To Increase Patentability Of Software, TechDirt.com (Dec. 1, 2016).
  8. See Jeff John Roberts, What Trump Means for the U.S. Patent System, Fortune.com (Jan 24, 2017).
0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply