EDITORIAL NOTE: The Author will present these findings on February 28, 2014 at the 58th Annual IP Conference at the John Marshall Law School in Chicago, Illinois.
Anti-patent groups who seek to diminish patent rights have turned the public’s imagination against licensing entities. Sometimes called NPEs or PAEs, sometimes called an epithet that need not be repeated here, such licensing entities do not practice the patent rights that they seek to license out. For now, let’s call them monetizing companies. I have previously written about the economic utility that monetizing companies bring to the patent system, and to the overall economy. As a general matter, they ensure that patents achieve their uses to promote technology transfer, promote commercialization of marketable ideas, and enable securitization of intangible assets to facilitate startup financing.
The anti-patent forces have convinced many members of the public – and their representatives in Congress – that such patent monetizing companies “abuse” the court system. But what do the data say?
If monetizing companies as a class “abuse” litigation more often than plaintiff operating companies, we would expect to see data that show that they bring less meritorious patent cases. Some have sounded the alarm that such data already exists. They cite to the annual PricewaterhouseCoopers report on patent litigation. But is this a reliable source for this conclusion?
The chart to the left, which is Chart 5b from page 12 of the latest PwC report released in 2013, appears to show comparative “success rates.” The chart appears to answer the question negatively for monetizing companies. As depicted in the chart, operating companies seem always to have had a higher “success rate” in court compared to monetizing companies, most recently 38% compared with 26%. That is, in fact, how alarmists (including those at the highest levels of government who advise the present Administration) have used these statistics. Examples are everywhere. Here are four notable ones:
“8. PAEs are Less Successful than Practicing Entities in Litigation
PwC’s excellent annual litigation report is chock full of statistics about patent litigation and in particular, with respect to NPEs, that: they look to . . . a lower than practicing company success rate (34% practicing co v. 23% NPE) that is declining (Chart 5B).”
Colleen Chien, White House’s Office of Science and Technology Policy (OSTP), Patent Trolls by the Numbers (March 14, 2013).
“For instance, the PricewaterhouseCoopers, 2011 Patent Litigation Study shows that PAEs are successful in just 23% of litigation, and even less successful in certain industries including business/consumer services, software, and telecommunications.”
David A. Balto, Comments of Food Marketing Institute and the National Restaurant Association to the Federal Trade Commission and U.S. Department of Justice on Patent Assertion Entities, at 2 (Apr. 3, 2013)
“While non-practicing and practicing entity success rates were very close to each other in the 2001-2006 time period (28% vs. 33%), they diverged in the 2007-2012 time period (25% for NPE’s vs. 38% for non-NPE’s) due largely to an increase in the number of NPE cases disposed of by summary judgment.”
Prepared Judiciary Committee Statement of Philip S. Johnson, Chief IP Counsel, Johnson & Johnson, at 12 (Dec. 17, 2013) (quoting PwC 2013 Study).
“Trolls Wither on the Papers. Trolls and practicing plaintiffs are equally successful at trial, but practicing entities are much more successful winning on summary judgment.”
R. David Donaghue, Partner at Holland & Knight, Patent Litigation Strategies for Retailers Based Upon the PwC Patent Litigation Study, Patent Law Practice Center Website (Nov. 7, 2011).
Unfortunately for the cause of rational discourse, commentators and government advisors have overlooked that these statistics do not, in fact, report the “success rates” that most people think they do. For example, here is the relevant “methodology,” taken from the PwC report itself (page 34):
To study the trends related to patent decisions, PwC identified final decisions at summary judgment and trial recorded in two WestLaw databases, U.S. District Court Cases (DCT) and Combined Jury Verdicts and Settlements (JC-ALL), as well as in corresponding Public Access to Court Electronic Records (PACER) system records. The study focuses on 1,856 district court patent decisions issued since 1995. Definitions for important terms used throughout the study are listed here:
A success included instances where a liability and damages/permanent injunction (if included) decision was made in favor of the patent holder.
It is plain to see that the PwC methodology is full of holes. PwC’s pool of cases is limited to actual trials or patentee-initiated summary judgments. In other words, the “denominator” of their rate calculation is (a) all trials plus (b) all summary judgments sought by the rights holder. The numerator is the quantity of these where the rights holder won.
What is missing here? Settlements. PwC does not collect or code any statistics to determine the quantity, quality or rate of settlements. In many cases, rights holders view settlements as touchstones of success. So why does this matter?
I assert that, using currently-available data, it is not possible to proclaim with any credibility that monetizing companies “succeed” at litigation less often than operating companies. I even assert that it is not possible to assert the contrary proposition, that they succeed more often. We simply do not know. The absence of data that characterize settlements makes this impossible. Policy makers should be skeptical of anyone who claims that current data (especially PwC’s data) reveal the answer.
I do propose, however, that there is a better metric. This is based on my experience as a licensing and trial attorney. It looks at the problem from the other direction. I propose that if any comparison is made at all, we should look at patentee loss statistics. Patentee loss statistics are much more likely to allow a comparison between monetizing companies and operating companies, and the cases they bring.
Why is this? Two reasons. First, imminent patentee merits victories will get vacuumed into the settlement category. The old litigator’s aphorism comes to mind – good cases settle, bad cases get tried (at least from the patentee’s perspective). There is no way to tell if that happens more for one class over the other. And second, trial and patentee-initiated summary judgment proceedings are a tiny statistical blip. It turns out that in terms of quantity, there are about ten times more defense merits wins than patentee merits wins among all cases that get litigated and do not settle. The explanation for this is simple – a patentee does not have to “win” to succeed – it only has to settle on monetary terms that it can convince an opponent to give.
One special case deserves mention: when the defense seeks but fails to win on summary judgment. The PwC data are blind to this phase of a typical patent case. This time is often when monetizers finally convince a defendant to pay. This phase of a case presents a powerful negotiating position for a monetizer-patentee, compared with an operating company-patentee. Monetizers have no competitive need to proceed to trial, but operating companies might (perhaps to get an injunction against a competitor).
A recent scholarly work reveals the truly revealing statistics: Robin Feldman, Tom Ewing & Sara Jeruss, The AIA 500 Expanded: The Effects of Patent Monetization Entities, UC Hastings Research Paper No. 45, (April 9, 2013). Feldman, Ewing & Jeruss performed high granularity coding of patent litigation outcomes based on cases filed in two periods: 2007-2008 and 2011-2012. Outcomes fit into one of the following “buckets” per case (p.85):
Of the potential outcomes, settlement dwarfs all others. Indeed, the researchers note that given how many lawsuits settle, the remaining numbers are too small to reach definitive conclusions and are only useful as observations. Of the rest, only the following “buckets” indicate a defense victory on the merits:
Procedural – Dismissal
Claim Defendant Favored in Consent Judgment
Claim Defendant Win – Judgment on the Pleadings
Claim Defendant Win on Summary Judgment
Claim Defendant Win at Trial
Claim Defendant Win on JMOL
It is eye opening to add up the percentages for the respective entries. Doing so allows some stunning observations:
|Procedural – Dismissal||4.29%||3.20%|
|Claim Defendant Favored in Consent Judgment||0.32||0.46|
|Claim Defendant Win on Pleadings||0.02||0|
|Claim Defendant Win on Summary Judgment||1.59||1.89|
|Claim Defendant Win at Trial||0.57||0.29|
|Claim Defendant Win on JMOL||0.04||0.04|
|Compare with “Likely Settlement” Rates||71.94%||74.44%|
These data allow observations completely at odds with the PwC reports. These data show that, if anything, monetizers lose less (5.88% compared to 6.83%, a difference of 14%) and settle more (a difference of 3.5%) compared to operating companies. PwC’s results ignore that 70-75% of patent cases settle, and do not account for the opposite comparison-conclusion shown by the “defense merits wins” metric (i.e., that monetizers perform 14% better in court than operating companies).
There are other reasons to question why policymakers rely on the PwC results. PwC reports do not explain their coding methodology (whereas Feldman, Ewing & Jeruss do). PwC also shows some surprising and unexplained year-over-year changes to their reported data. For example, PwC’s charts from 2010 to 2011 ought to show identical prior-year data. But they do not. They instead show sharp changes. Take note of Year 2004 patent holder success rates:
2010 Study (page 15)
2011 Study (page 17)
As can be seen, according to the 2010 study, monetizing companies appear to have a higher “success rate” for the year 2004 than operating companies. But in the 2011 study, the charts show a reversal for 2004. Why do the data change so much, for a year that was long past? The reports contain no explanation.
There is a second strange leap. From 2011 to 2012, PwC reports the overall “success rate” in a shifting way. In 2011, the study reports it on an individual year-by-year basis (see chart mentioned immediately above). But starting in 2012, the study reports it in arbitrary 6-year groupings (see left).
Is it possible that someone did not like these charts being used to show how many years monetizing companies had a higher “success rate” than operating companies (i.e., 2002, 2003, 2008 and 2009, even ignoring the 2004 turnabout)?
In short, no one can definitively say whether monetizers bring better cases (or worse cases) than operating companies. But if the data support any conclusion, they support that monetizers bring more meritorious cases than operating companies (i.e., they lose on the merits less often), and act more successfully to unclog them from court dockets (i.e., they settle more often).
Be skeptical, and think critically, when you see biased advocates quoting the PwC Reports.
Join the Discussion
18 comments so far.
Paul F. MorganMarch 9, 2014 04:07 pm
Yes Gene, this is a litigation problem. But it is peculiarly so for patent litigation, because the defendants have the vast majority of the discovery and other pre-trial cost burdens, and an additional burden of “clear and convincing evidence” on every possible issue other than infringement,. Yet defendants all to often do not even get a Markman hearing to define the claim scope, and thus be able to get a summary judgement for non-infringement, until the eve of trial. Because there is no requirement for judges to conduct a more timely Markman or restrict any irrelevant discovery before that. So, not surprisingly, almost all defendants settle earlier. This would be such a simple thing to change that it is amazing there has not been more support for it in the pending legislation?
Instead we just get more arguments about dubious studies with illogical conclusions from interested parties on both sides, and expensive internet and media advertising from organizations allegedly representing “small inventors” presumably funded by much deeper-pockets interested parties.
Gene QuinnMarch 9, 2014 01:47 pm
You say: “Looking at cases is not the right metric to measure any potential harm from pursuit of cases with little or no merit. Measuring the number of defendants is more relevant. That measurement also avoids the artificial change in number of cases that occurred due to AIA.”
That is fine, but the AIA has made is far more difficult, if not impossible, to bring lawsuits with numerous defendants.
You say: “settlements for weak cases may be occurring.”
That is true, but it could also be that settlements are for strong cases as well. It is pure speculation unless you look at the patent claims and analyze the cases, which is not being done by those who claim there is a patent problem.
You say: “A survey of more than 3800 members of the American College of Trial Lawyers (ACTL) in 2008 had 83% of respondents saying that litigation costs drive some cases to settle that should not settle on the merits…”
Interesting, but mainly because it undercuts the position of those who erroneously claim we have a patent problem. Let’s accept this at face value for a moment. Across the board 83% of cases settle when they should not if merits mattered. That conclusively proves that the problem is a litigation problem and NOT a patent problem.
Paul F. MorganMarch 9, 2014 11:39 am
Richard, I cannot agree [without proof] that the majority of PAE suits [or even others] do not settle until the very eve of trial. Many patent suits are settled early-on by many defendants simply to avoid discovery costs and employee time burdens, even if they realistically think the patent claims are invalid, unless the settlement demand greatly exceeds total litigation costs. Pre-trial discovery costs are almost entirely a cost burden just on on defendants, especially in the case of PAE plaintiffs since they did not even file or prosecute the subject application and are often just a shell corporation with nothing in it to discover, and defendants have the burden on all issues other than infringement in patent suits.
However, there are some incentives to some for not settling patent litigation earlier. The eve of trial is the point at which both previously-decision-avoiding corporate management and their trial counsel must make a realistic decision to risk (by proceeding) an adverse judgment which could hurt their future job opportunities, and defendant trial counsel [typically being paid by the hour rather than by contingent fees for winning] have already had many good months of high billings.
Note that I am not even mentioning defendants winning early and cheaply by a successful summary judgment motion, because that only occurs in a very small percentage of patent suits. Many judges will not even decide such motions until the eve of trial. S.J. is almost never for granted for invalidity [notwithstanding KSR support for that] rather than clear non-infringement, or, [recently] alleged 101. [The obvious answer to that is for defendants to make faster and better use of AIA PTO IPR and CBM proceedings whenever possible. ]
Richard FalkMarch 8, 2014 11:56 pm
Looking at cases is not the right metric to measure any potential harm from pursuit of cases with little or no merit. Measuring the number of defendants is more relevant. That measurement also avoids the artificial change in number of cases that occurred due to AIA. Most manufacturer cases are not against dozens to hundreds of defendants — most are against one or a small number of competitors. In comment #36 in the following article on this site:
I wrote about the results of non-proprietary information (the study was done by academics who published the detailed raw data and methodology) and they show an increase from 31.59% in 2010 to 36.75% in 2012 in number of defendants (technically “Number of Parties Excluding the Patentee”) sued by plaintiffs in the PAE categories of Large Aggregators and Patent Holding Companies. For Operating Companies, it fell slightly from 48.91% to 48.35%. Also, even if one does look at cases, for Operating Companies that fell from 69% in 2010 to 45% in 2012. It’s not 80%.
As for why this non-proprietary study shows lower absolute numbers for PAE categories, a clue is given in footnote 54 on page 29 that explains why Uniloc was coded as an Operating Company. “Uniloc USA Inc. provides a good example of how our coding may differ from others. A Google search for Uniloc USA Inc. returns a link for http://www.uniloc.com. From here, if the link to NetAuthority is clicked, the website for http://netauthority.com is provided, where products related to the patents in suit are clearly being sold by Uniloc USA Inc. Accordingly, Uniloc USA Inc was coded as an operating company.”
The bottom line is that most cases settle (around 92%) and these settlements are generally hidden behind non-disclosure agreements so anyone can claim anything about these hidden cases. If one believes that companies will fight cases with little or no merit and not settle, then one can claim that settlements are strong cases. Otherwise, if one believes companies will settle when the costs of doing so are less than the costs of proving their case in court, then settlements for weak cases may be occurring.
A survey of more than 3800 members of the American College of Trial Lawyers (ACTL) in 2008 had 83% of respondents saying that “litigation costs drive some cases to settle that should not settle on the merits” or as the task force chairman wrote ““the costs and burdens of discovery are driving litigation away from the court system and forcing settlements based on the costs, as opposed to the merits.” Of course, that was in the day when discovery was often done early in cases before their merit from preliminary claim construction was determined and the discovery used to include e-discovery.
Gene QuinnMarch 8, 2014 02:44 pm
You say: “have not stemmed the tide of patent assertion entity lawsuits against”
That may be true, but the publicly available data states that 80% of patent infringement cases are brought by manufacturers. Only those who use proprietary information come up with different numbers. Of course, proprietary numbers that cannot be challenged can’t be accepted as true when the repeated studies using publicly available and verifiable data reach the same 4 to 1 conclusion.
Richard FalkMarch 8, 2014 11:57 am
In the Kaspersky cases and similar cases, the plaintiff doesn’t drop out just after filing a complaint, but just before trial. That is, they force maximum costs for defendants by going through pretrial, claims construction, discovery, summary judgment motions (and leave to file for EDTX), etc. This is to maximize the likelihood of settlement because there isn’t just the specter of costs from the district court case but also from an appeal which is almost always done. They figure that at some point the millions of dollars spent by the defendant will have them back down to avoid spending millions more.
This was made clear in the links in the comment I linked to. I didn’t want to copy and paste what I had written before, but since it apparently isn’t read I’ll copy and paste anyway.
In the above first lawsuit filed on 12/30/08, Kaspersky Labs was the only one of 35 companies who refused to settle and the plaintiff, Information Protection and Authentication of Texas (IPAT) blinked and dropped the case with prejudice before trial (in part due to a partially accepted summary judgment decision). However, it cost Kaspersky Labs 3-1/2 years of distraction and $2.5 million in legal fees that was not recovered (i.e. no fee shifting and no sanctions).
Then, as described in the above link, came Lodsys who filed on 2/11/11, where again Kaspersky Labs was the only one of 55 defendants who refused to settle and the plaintiff dropped the case just before trial. However, this cost Kaspersky Labs 2-1/2 years of distraction and an undisclosed cost in legal fees (probably at least $2 million).
You might then think that a rational plaintiff would put Kaspersky Labs on the “do not sue unless you have a strong case” list, but that is not how this works. On 10/18/13, Kaspersky Labs filed a lawsuit for declaratory judgment against Device Security who had sent them a demand letter dated 9/18/13. Also on 10/18/13, Kaspersky Labs was sued by Uniloc. So now they have two active lawsuits from patent assertion entities.
Other companies who have a policy of always fighting back, such as Twitter and Newegg have not stemmed the tide of patent assertion entity lawsuits against them (do a Pacer search and correlate cases against known patent assertion entities if you don’t believe me).
Paul F. MorganMarch 8, 2014 09:08 am
Richard, your comment is confusing merely “going to court” [filing a complaint costs almost nothing] to get pre-trial settlements, with going all the way in the litigation through an an expensive trial to get an actual decision, which is what my note was about and is the primary subject of the study being reported on this blog ,
Richard FalkMarch 7, 2014 07:42 pm
Regarding Paul Morgan’s statement in comment #6: “They have no incentive [rather, a strong disincentive] to risk their own money to go to trial unless they have to and they realistically think that they will win.”
I’ve posted before about why this is simply not true, at least regarding incentive to go to Court (i.e. before trial). See my post #52 in the comments to the following article on this site where the plaintiffs dropped the case before trial (two separate times with two different plaintiffs) and where all but one company settled:
One cannot assume that settlements mean that cases have any merit. Weak cases are pursued until they result in settlements or are stopped by the courts with summary judgment or are dropped before trial by the plaintiff if the defendant has still not settled. So long as there is any chance of getting a settlement that recoups at least some plaintiff’s costs, there is incentive to go forward — prior costs are sunk costs. Since nearly all defendants settle even for weak cases because it’s less costly than fighting, the plaintiff’s presume that they will. I’ve been on the receiving end twice (at the company I work for) with cases totally without merit and the plaintiffs would not drop the case under any circumstances except settlement — the first time the company settled presumably because it was cheaper than spending a million or more dollars to go forward and they had already spent nearly that much; the second time the company fought back and won though the case is (of course) appealed by the plaintiff, but there is no cost recovery whatsoever (no sanctions, no fee shifting).
The easy solution to this is to penalize plaintiffs with weak cases and in particular plaintiff’s attorneys especially when the behavior is repetitive. Unfortunately there is no doctrine of cumulative repetitive harm. If such a doctrine existed, then Courts would not view behavior as being isolated to a single case, but would consider the effect on the Courts and on society if every case had similar conduct because if the behavior is not punished it will usually become more prevalent.
Michael RischMarch 2, 2014 07:53 am
I know of at least one study that will consider summary judgment DENIED – we’ll see what the results are.
Robert GreenspoonFebruary 27, 2014 03:31 pm
Update to the posting (and correction):
The authors of the PwC Patent Litigation Study contacted me today to discuss their methodology. They convinced me that they (like most of us) only want to present (and receive) good and meaningful data. They helpfully clarified several points. I will repeat the information, and leave it to the reader to judge the impact on the conclusions in the posting:
1) The “denominator” for patentee success rates in the PwC Studies includes all dispositive summary judgment motions regardless of which party brought the motion. I had wrongly inferred that the denominator included only patentee-initiated summary judgments. It remains true, it seems, that the “numerator” will only show a patentee summary judgment “success” if the outcome of the motion is a liability judgment. So when an accused infringer’s motion for summary judgment is denied, and a case is cleared for trial, that would not be coded as a patentee “success.” Perhaps the authors can comment if this is so.
2) Prior year shifts in year-over-year reporting are expected. This is for three reasons: (a) WESTLAW provides the data pool and this database has a lag, (b) an appellate result could change the prevailing party for a given case in the data pool, and (c) there is a constant quality control process that every year seeks to correct any unnoticed coding errors.
3) The authors compressed the “success rate” charts into 6-year groups because of a belief that the year-by-year reporting seemed too busy.
Paul F. MorganFebruary 26, 2014 05:30 pm
Anon, I don’t mind that word editing at all.
The absence of any substantive rebuttal of the actual points I made supports a view that the subject study is misleadingly interpreting its data.
AnonFebruary 26, 2014 03:06 pm
Would you mind if I took issue with your use of the word “normal” to describe one form of representation and the word “real” to one type of company? Clearly, both types of representation are “normal” and both types of companies are “real.”
Your use tends to show a subtle, but pernicious bias.
Paul F. MorganFebruary 26, 2014 01:37 pm
This is an interesting statistic to study, but there are very important factors not even discussed here: First, many if not most NPE’s are controlled by, or are represented in court by, contingent fee attorneys. They have no incentive [rather, a strong disincentive] to risk their own money to go to trial unless they have to and they realistically think that they will win. Unlike the attorneys for many normal companies that are getting paid large amounts of money by the hour irrespective of trial outcomes. Secondly, unlike real companies suing competitors, NPE’s have no interest in trying to stop infringements by going to trial to try to get an injunction. Thus they are inherently only interested in pre-trial cash settlements whenever possible. Thus these statistics, even if accurate [which depends on many variables including how NPE is defined, etc.] is not surprising, and does not prove the subject patents involved are necessarily more valid. Especially if Fed. Cir. reversal rates are not also compared.
mikeFebruary 26, 2014 08:25 am
Statistics are like bikinis – what they reveal is interesting, but what they conceal is crucial. Excellent work!
Jeff LindsayFebruary 26, 2014 01:42 am
Excellent investigative journalism! I didn’t know that was still allowed these days.
Has PWC responded? You’ve exposed some very serious weaknesses and major blunders in the study. This should tarnish their credibility on the issue of IP at least, and I would expect a response of some kind, at least “We still stand by our work even if some people don’t fully understand how brilliant it is.”
Gene QuinnFebruary 25, 2014 01:59 pm
I agree with you. The truth is somewhere in between, in a very difficult to quantify space.
Michael RischFebruary 25, 2014 12:56 pm
Interesting stuff. It’s consistent with what I’m finding in my 25 year study of the top 10 most litigious NPEs v. a random control set. If you count all the trial outcomes, a much small percentage of PE patents are invalidated because those patents are uniformly affirmed at trial, while the NPEs have almost no trials. If, however, you remove trials, then NPE patents and PE patents are invalidated at about the same rate (as a percentage of merit rulings) but about twice as often (as a percentage of patents asserted).
A key question I’m going to ask in the new article, is why NPEs don’t go to trial. If you believe that they settle favorably, then it looks like they win. If you believe that they dismiss rather than test their patents, then they are causing harm to the system. The truth is likely somewhere in between.
Gene QuinnFebruary 25, 2014 11:01 am
Excellent work Robert! It seems like the all too familiar tactic is what leads to the erroneous conclusion that operating companies fare better. Just ignore the facts that don’t support your position, focus on a subset of facts that do support your position, and then pretend that the broader facts don’t exist. Sprinkle in a healthy amount of vilification of those who point out the truth and that really explains where we are at in this dialogue.