German Federal Patent Court (partially) invalidates 80% of litigated patents

Patent grave yard?
A study by Hess, Müller-Stoy and Wintermeier provides an interesting empirical look at the invalidation rate before the German Federal Patent Court and German Federal Court of Justice. Based on data from 2010 to 2013, the authors report that the Federal Patent Court invalidates 44% of all litigated patents, partially invalidates 36% and maintains (unamended) just 20%. An earlier study by Henkel and Zischka came to essentially the same conclusion (78% partial or total invalidity for decisions from 2010 to 2012), but draws different policy recommendations.

For software and telecommunications patents, the invalidation rate is even higher at 88% (58% totally invalid, 30% partially invalid, 12% maintained). Patentees fare slightly, but only slightly, better on appeal, as the win rate for appeals against decisions invalidating a patent is higher (60%) than the win rate for appeals against decisions maintaining a patent (40%).

These results are surprising as economic (i.e. rational choice) theory predicts that - if the stakes for both parties are the same (an assumption that does not necessarily hold for patent invalidity cases) - plaintiff and defendant win with equal probability (the Priest/Klein selection hypothesis, but see Gliksberg for empirical support for the "legal model" to which most lawyers would subscribe). The high invalidation rate would indicate that the stakes for the nullity plaintiff are much higher than the stakes for the patentee, which is possible but mere speculation.

Hess et al propose three tentative explanations for the high invalidation rate: (i) errors by the examiners, (ii) the introduction of new prior art before the court and (ii) different standards between the granting office and the court. New prior art could be found because potential infringers invest much more resources in prior art search than patent examiners. The authors consider this unlikely, because - according to their experience - "unsearchable" prior art is rarely dispositive; invalidations are rather based on (searchable) patent documents. The authors tend to think that the higher standard for inventive step by the court is the most important reason for the high invalidation rate.

If I may add my two cents, I would think that part of the puzzle lies in the categorization of "partially invalidated" patents as "invalidated", or "patentee loss". As everyone involved in patent litigation knows, there are claim amendments that really hurt (the patentee), and others that are mere formalities. Without knowing whether an amendment leads to a claim that is no longer infringed or is easily circumvented, it is difficult to say whether a partial invalidation is a win for the patentee or the nullity plaintiff. For an outsider, it is almost impossible to judge the importance of an amendment - this is not evident from the file wrapper, and certainly not from the file wrapper in a nullity proceeding, where the allegedly infringing embodiment is not at issue. If we were to classify the partial invalidations as 50% wins for patentee, we would arrive at an overall win rate for patentees in invalidity proceedings before the German Federal Patent Court of 38% (20% fully maintained plus half of 36% partially maintained). Still not great, but closing in on 50%.

A look at US data shows that patentees have a hard time there, too. While US Courts find for patentees in 58% of invalidity challenges, overall patentees only win 26% of cases (decisions from 2009 to 2013). Allison, Lemley & Schwartz summarize the reasons thus:
Why do patentees lose nearly three-quarters of the time when the court definitively resolves the merits? The answer is twofold. First, while courts turn away most validity challenges, patentees do not fare as well when it comes to infringement. Accused infringers won 54% (256 of 473) of their summary judgment motions alleging noninfringement of individual patents. That number rises to 57% (292 of 509) when we include stipulated judgments of noninfringement after claim construction, which are functionally equivalent to summary judgments of noninfringement; the patentee concedes that it cannot win under a particular claim construction in order to tee the case up for appeal. Second, the nature of patent litigation requires patentees to win every issue before the court. A patentee who defeats five of six invalidity challenges, only to lose the sixth, loses the case. So does a patentee who wins on validity and inequitable conduct but loses on infringement. One of us has referred to this as the “fractioning” of patent law. Our data suggest that it has a significant effect on patent cases overall because many of our cases had motions on multiple issues, and those motions were not always decided in favor of the same party. In patent law, a split decision is almost always a decision for the accused infringer, not the patentee.

German Federal Patent Court (partially) invalidates 80% of litigated patents German Federal Patent Court (partially) invalidates 80% of litigated patents Reviewed by Mark Schweizer on Tuesday, February 02, 2016 Rating: 5


  1. given that Hess/Müller-Stoy/Wintermeier's study is cited by Henkel and Zischka, I doubt that the latter is the earlier (as stated in the article)

  2. Before anyone starts using this as a stick with which to beat patents in general, or the quality of drafting, or the quality of examination, it should be noted that patents which are the subject of nullity proceedings are *not* necessarily a representative sample of patents overall. Patents which are the subject of litigation are necessarily a biased sample.

    If you will permit a generalisation, is not one possible explanation that an invalidity challenge will normally only be issued when there is a reasonably strong case for invalidation and/or sufficient economic/strategic advantage to make it worth the gamble (e.g. as a counterclaim in an infringement suit)? Thus, weak patents are overrepresented, and strong patents underrepresented, in litigation compared to their relative proportions among granted patents as a whole.

  3. New prior art could be found because potential infringers invest much more resources in prior art search than patent examiners. The authors consider this unlikely, because - according to their experience - "unsearchable" prior art is rarely dispositive; invalidations are rather based on (searchable) patent documents.

    I've not read the paper, but if this is an accurate reflection of the authors' views it seems to be total hogwash.

    Any patent attorney will be able to point you to an application which has been searched by multiple different patent offices with wildly differing results. The authors here seem to assume that all applications are examined by some sort of Platonic ideal of the Examiner with all the resources of Borges' Library of Babel.

    No database is complete. No two Examiners will look at the same invention in the same way, nor enter exactly the same keywords into their search. Compared to parties to litigation, Examiners have neither the time, nor the resources, nor the same level of sheer determination to find the crucial piece of invalidating prior art.

    Why do the authors assume that any new prior art brought in proceedings can only be "unsearchable"? To the contrary, it is likely to have been searchable, but simply wasn't found; or was only found in another country's search; or its importance was overlooked; or...

  4. the Henkel and Zischka paper is available in draft form since at least August 2014 ( and predates the Hess et al study. I cited the latest draft of the Henkel/Zischka paper, which does indeed postdate the Hess et al study.

  5. Having now read through the paper, I can see that the quote about "unsearchable" prior art in the original post is actually slightly misleading.

    The authors in fact state that cases are rare where a *public prior use* is asserted and is dispositive. This is subtly different than the impression given above.

    Nevertheless, the authors are curiously dismissive of the notion that new prior art (whether documentary, e.g. patent documents, or non-documentary, e.g. prior use) is relevant. This is despite the fact that they explicitly acknowledge that "In almost every case, nullity plaintiffs introduce new prior art [...] playing often the central role". This is simply hand-waved away with the observation that "very few nullity plaintiffs have large patent departments" and therefore (in the authors' opinion) few plaintiffs are able to locate more relevant prior art.

    A large patent department is not a prerequisite for conducting your own prior art search which finds prior art missed by the Examiner. You can do a basic search on Espacenet, or ask your attorney to do it for you, or engage a search agency. If (as the authors observe) a majority of nullity plaintiffs introduce new, highly-relevant prior art, yet few plaintiffs have their own large patent department, I do not understand their dismissive treatment here.

    The authors are also rather dismissive of the concept of "errors" by the patent examiners, but their justification rests on the comparative rarity of invalidation due to added subject matter or lack of enablement. They use this to dismiss the notion of examiners overlooking relevant passages in the prior art which has been found. This seems a flawed reasoning. Just because Examiners are apparently normally correct in their assessment of added matter or enablement, does not mean that they are somehow immune to misreading the prior art or interpreting ambiguous passages differently from the Court. Assessment of added subject matter and enablement is based on the patent application alone whereas assessment of the content of the prior art is a different skill...

  6. Let me set out the incidents of a correctly operating patent system. They include:

    i) litigation as a last resort
    ii) when both sides think they can win, and
    iii) both infringement and validity are settled on a "more likely than not" basis.

    In the 2:2 Matrix of possible outcomes, ranging from "not invalid and infringed" (ie a win for the patent owner) to "not infringed but invalid anyway"(a total blamage for the owner) patent owner wins on only one of the four domains of the matrix ie loses in 75% of cases.

    So what's so grim about the loss rate in the Courts of Germany? Is this heavy Attrition for patent owners not the necessary counterpart of an understandable reluctance on the part of Patent Offices to deny the Inventor her "day in court". Imagine the outcry from Inventors and their assignees, if Patent Offices refused all patent applications that strike them as patentability borderline.

    Must I read this Paper? Have I dismissed it in too cavalier a fashion?

  7. The greatest resource in these matters is, of course, time. Something which examiners have progressively less of.

  8. Max Drei you have indeed dismissed it in too cavalier a fashion. The statistics are not as simple as you make out. In any given Patent infringement action each of your four possible outcomes are not equally likely. Grossly simplifying, there are three possible case statuses:

    I) The validity of the patent is not in real doubt but infringement is

    II) Infringement of the patent is not in real doubt but validity is

    III) Infringement and validity of the patent are both in doubt

    With regards to validity only, I would expect the chances of a finding in favour of the patentee in each of these cases of 100%, 50%, and 50%. Thus, if all cases are brought equally the finding of a valid patent should occur in 2/3rds of cases.

    With regards to infringement only, I would expect the chances of a finding of infringement in each of these cases of 50%, 100%, 50%. Thus if all cases are brought equally the finding of infringement should occur in 2/3rds of cases.

    Cases of a valid patent and infringement would occur in 50%, 50%, and 25% of the cases. So, again if all cases are brought equally, the finding of a valid and infringed patent should occur in 5/12ths of cases.

    Of course all of the above is a massive and gross simplification that has no basis in reality: cases are rarely 50/50 and not all cases are brought equally. Hence the need for studies as to the reality of the situation.

  9. How nice of MaxDrei to slip into his "version" that correctness necessarily includes the "more likely than not" yardstick.

    Absolute rubbish.

  10. I think Max Drei has a firmer grasp of statistics than Anonymous does.

    Inclusion of the concept of infringement or validity being “in doubt” is a useful step in that the understanding and intentions of the patentee or the infringer have to be taken into account. However, this does not factor in the “don’t care” aspect that can apply in a bifurcated system.

    If a patentee knows that he can get an injunction (temporary or permanent) before a decision on validity is rendered then the temptation to assert a patent, regardless of validity, is significant.

    So into Anonymous’s “validity in doubt” spectrum of cases should be added “don’t care about validity” cases. This can make the situation rather complex, so let’s go back to Max’s criteria and the 2x2 matrix he puts forward.

    As stated by Max, the patentee loses in 75% of cases. Conversely the patentee wins in 25% of cases, which is pretty close to the 26% of patentee wins reported by the US data referred to. I never thought that the US system would meet Max’s criteria for a correctly operating patent system, but this is an interesting coincidence (I indulge in confirmation bias on every opportunity).
    In contrast, in a bifurcated system, where a decision in infringement can be the outcome before an invalidity case really gets under way, the question posed to the infringer is “Do I feel sure enough of winning, and is this case important enough, that I can stand the delay of waiting for an invalidity decision while being kept out of the market, or is taking a licence a rational approach?”. In such circumstances invalidity actions will be rarer than infringement actions:-

    - None will be filed where the infringer has no doubt that the patent is valid – because there is no point;

    - Few will be filed where the infringer is uncertain as to validity – because it will be a gamble while the infringer is off the market.

    There will thus be a large bias towards cases where the argument against validity is strong.

    So a high invalidity rate before the German courts is to be expected and not a surprise.

    The 2009 Harhoff study “Economic Cost-Benefit Analysis of a Unified and Integrated European Patent Litigation System“ (Tender No. MARKT/2008/06/D) showed the number of validity actions being about 1/3 the number of infringement actions, which accords with what one might expect from the above analysis on the basis of voodoo statistics (confirmation bias again).

    What this means is not that the validity of all German patents is low – it means the validity of German patents that are challenged in invalidity actions is low.

    As a corollary however, it means that many German patents are asserted and found infringed that are of dubious validity – the system just doesn’t encourage challenge of possibly invalid rights.

    These features of German procedures perhaps help to explain why most oppositions are German-on-German affairs. The public will not have validity assessed by the courts when infringement is assessed (except in extreme cases), so had better clear the decks before they get anywhere near a court.

    It will be interesting to see whether this calculation changes when the UPC gets under way (particularly once the transitional period ends).

  11. Think you guys lose perspective. What about an study on judges?
    I believe that you will find that courts have an invalidity rate which is proportional to the number of case they handle.
    Courts with very few cases (i.e. courts not specialised in patents) tend to have very low invalidity rates.
    The reason? Well, a (board of) judges may develop an own criterion only by seeing a number cases.
    Problem in Europe? Well, the EPO does not take into account national criteria on invalidity.
    In any case, MaxDrei has a point. No alleged infringer enters the court room unless it has no other option available, while the opposite is not always true (the patent passed an exam indeed and nobody cares about infringement which are not damaging the commercial results of the company, while dubious cases are usually given a try), so invalidity rates between 50% and 33% are not to be frowned upon "as such".
    What is more worrying is that the invalidity rate of software patents is much higher. There is something wrong with them (Lack of prior art, big companies influencing the examination process???)

  12. Thanks Meldrew. Your last three paragraphs are perceptive. You wonder whether the UPC will change the pattern of patent litigation in Germany. Will it be quite so much German on German as it is now? The writers of the Paper are German litigators, no doubt thinking about how to earn a crust when litigants are choosing the UPC as their forum.

  13. Glad to be out of the madhouseWednesday 3 February 2016 at 08:59:00 GMT

    Anonymous@15:14 : What "national criteria on invalidity" is the EPO supposed to be disregarding? According to Art. 138 of the European Patent Convention, a European patent "may be revoked with effect for a Contracting State only on the grounds that:

    (a)the subject-matter of the European patent is not patentable under Articles 52 to 57;
    (b)the European patent does not disclose the invention in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art;
    (c)the subject-matter of the European patent extends beyond the content of the application as filed or, if the patent was granted on a divisional application or on a new application filed under Article 61, beyond the content of the earlier application as filed;
    (d)the protection conferred by the European patent has been extended; or
    (e)the proprietor of the European patent is not entitled under Article 60, paragraph 1."

    These are the same criteria that the EPO is supposed to apply. If some national courts are applying their own criteria, they are quite simply in breach of law.

    As for the invalidity rates in certain fields, the question to be raised is also whether the technically aware examiners were wrong in granting them, or the not-always-technically-literate judges wrong in revoking them. Nobody who has been exposed to the judiciary will ever expect it to be always right...

  14. An interesting study to be sure. However, it was already published in 2014 - so why the attention now?


    The real relevance of this statistics becomes clear if you take Germany's bifurcated system into account, in which validity is dealt with by a different court than infringement. The Bundespatentgericht usually takes much, much longer to decide on validity than the civil courts do for infringement. On top of that, those civil courts are (were?) very reluctant to stay a case on the basis that the patent might be invalidated by the BPatG. So there is a very real chance that you will be under an injunction for several years (!) before it is decided that the patent was indeed invalid all along. Assuming that the patentee wins on infringement in 50% of the cases, a total of 40% of the infringement cases result in an injunction on the basis of claims that were found invalid (of course, if the patent was upheld in limited form there could still be infringement, but the infringement analysis made by the civil court is no longer valid). That is undesirable and very relevant by any measure, it seems to me.

  15. Litigator is then the bifurcated court system in Germany the reason why 70% of all oppositions at the EPO are filed by Germans? Is it that they fear being enjoined by the owner of a bad patent and so get their retaliation in first?


All comments must be moderated by a member of the IPKat team before they appear on the blog. Comments will not be allowed if the contravene the IPKat policy that readers' comments should not be obscene or defamatory; they should not consist of ad hominem attacks on members of the blog team or other comment-posters and they should make a constructive contribution to the discussion of the post on which they purport to comment.

It is also the IPKat policy that comments should not be made completely anonymously, and users should use a consistent name or pseudonym (which should not itself be defamatory or obscene, or that of another real person), either in the "identity" field, or at the beginning of the comment. Current practice is to, however, allow a limited number of comments that contravene this policy, provided that the comment has a high degree of relevance and the comment chain does not become too difficult to follow.

Learn more here:

Powered by Blogger.