The SHIELD Act: When Bad Economic Studies Make Bad Laws

Adam Mossoff —  15 March 2013

Earlier this month, Representatives Peter DeFazio and Jason Chaffetz picked up the gauntlet from President Obama’s comments on February 14 at a Google-sponsored Internet Q&A on Google+ that “our efforts at patent reform only went about halfway to where we need to go” and that he would like “to see if we can build some additional consensus on smarter patent laws.” So, Reps. DeFazio and Chaffetz introduced on March 1 the Saving High-tech Innovators from Egregious Legal Disputes (SHIELD) Act, which creates a “losing plaintiff patent-owner pays” litigation system for a single type of patent owner—patent licensing companies that purchase and license patents in the marketplace (and who sue infringers when infringers refuse their requests to license). To Google, to Representative DeFazio, and to others, these patent licensing companies are “patent trolls” who are destroyers of all things good—and the SHIELD Act will save us all from these dastardly “trolls” (is a troll anything but dastardly?).

As I and other scholars have pointed out, the “patent troll” moniker is really just a rhetorical epithet that lacks even an agreed-upon definition.  The term is used loosely enough that it sometimes covers and sometimes excludes universities, Thomas Edison, Elias Howe (the inventor of the lockstitch in 1843), Charles Goodyear (the inventor of vulcanized rubber in 1839), and even companies like IBM.  How can we be expected to have a reasonable discussion about patent policy when our basic terms of public discourse shift in meaning from blog to blog, article to article, speaker to speaker?  The same is true of the new term, “Patent Assertion Entities,” which sounds more neutral, but has the same problem in that it also lacks any objective definition or usage.

Setting aside this basic problem of terminology for the moment, the SHIELD Act is anything but a “smarter patent law” (to quote President Obama). Some patent scholars, like Michael Risch, have begun to point out some of the serious problems with the SHIELD Act, such as its selectively discriminatory treatment of certain types of patent-owners.  Moreover, as Professor Risch ably identifies, this legislation was so cleverly drafted to cover only a limited set of a specific type of patent-owner that it ended up being too clever. Unlike the previous version introduced last year, the 2013 SHIELD Act does not even apply to the flavor-of-the-day outrage over patent licensing companies—the owner of the podcast patent. (Although you wouldn’t know this if you read the supporters of the SHIELD Act like the EFF who falsely claim that this law will stop patent-owners like the podcast patent-owning company.)

There are many things wrong with the SHIELD Act, but one thing that I want to highlight here is that it based on a falsehood: the oft-repeated claim that two Boston University researchers have proven in a study that “patent troll suits cost American technology companies over $29 billion in 2011 alone.”  This is what Rep. DeFazio said when he introduced the SHIELD Act on March 1. This claim was repeated yesterday by House Members during a hearing on “Abusive Patent Litigation.” The claim that patent licensing companies cost American tech companies $29 billion in a single year (2011) has become gospel since this study, The Direct Costs from NPE Disputes, was released last summer on the Internet. (Another name of patent licensing companies is “Non Practicing Entity” or “NPE.”)  A Google search of “patent troll 29 billion” produces 191,000 hits. A Google search of “NPE 29 billion” produces 605,000 hits. Such is the making of conventional wisdom.

The problem with conventional wisdom is that it is usually incorrect, and the study that produced the claim of “$29 billion imposed by patent trolls” is no different. The $29 billion cost study is deeply and fundamentally flawed, as explained by two noted professors, David Schwartz and Jay Kesan, who are also highly regarded for their empirical and economic work in patent law.  In their essay, Analyzing the Role of Non-Practicing Entities in the Patent System, also released late last summer, they detailed at great length serious methodological and substantive flaws in The Direct Costs from NPE Disputes. Unfortunately, the Schwartz and Kesan essay has gone virtually unnoticed in the patent policy debates, while the $29 billion cost claim has through repetition become truth.

In the hope that at least a few more people might discover the Schwartz and Kesan essay, I will briefly summarize some of their concerns about the study that produced the $29 billion cost figure.  This is not merely an academic exercise.  Since Rep. DeFazio explicitly relied on the $29 billion cost claim to justify the SHIELD Act, and he and others keep repeating it, it’s important to know if it is true, because it’s being used to drive proposed legislation in the real world.  If patent legislation is supposed to secure innovation, then it behooves us to know if this legislation is based on actual facts. Yet, as Schwartz and Kesan explain in their essay, the $29 billion cost claim is based on a study that is fundamentally flawed in both substance and methodology.

In terms of its methodological flaws, the study supporting the $29 billion cost claim employs an incredibly broad definition of “patent troll” that covers almost every person, corporation or university that sues someone for infringing a patent that it is not currently being used to manufacture a product at that moment.  While the meaning of the “patent troll” epithet shifts depending on the commentator, reporter, blogger, or scholar who is using it, one would be extremely hard pressed to find anyone embracing this expansive usage in patent scholarship or similar commentary today.

There are several reasons why the extremely broad definition of “NPE” or “patent troll” in the study is unusual even compared to uses of this term in other commentary or studies. First, and most absurdly, this definition, by necessity, includes every university in the world that sues someone for infringing one of its patents, as universities don’t manufacture goods.  Second, it includes every individual and start-up company who plans to manufacture a patented invention, but is forced to sue an infringer-competitor who thwarted these business plans by its infringing sales in the marketplace.  Third, it includes commercial firms throughout the wide-ranging innovation industries—from high tech to biotech to traditional manufacturing—that have at least one patent among a portfolio of thousands that is not being used at the moment to manufacture a product because it may be “well outside the area in which they make products” and yet they sue infringers of this patent (the quoted language is from the study). So, according to this study, every manufacturer becomes an “NPE” or “patent troll” if it strays too far from what somebody subjectively defines as its rightful “area” of manufacturing. What company is not branded an “NPE” or “patent troll” under this definition, or will necessarily become one in the future given inevitable changes in one’s business plans or commercial activities? This is particularly true for every person or company whose only current opportunity to reap the benefit of their patented invention is to license the technology or to litigate against the infringers who refuse license offers.

So, when almost every possible patent-owning person, university, or corporation is defined as a “NPE” or “patent troll,” why are we surprised that a study that employs this virtually boundless definition concludes that they create $29 billion in litigation costs per year?  The only thing surprising is that the number isn’t even higher!

There are many other methodological flaws in the $29 billion cost study, such as its explicit assumption that patent litigation costs are “too high” without providing any comparative baseline for this conclusion.  What are the costs in other areas of litigation, such as standard commercial litigation, tort claims, or disputes over complex regulations?  We are not told.  What are the historical costs of patent litigation?  We are not told.  On what basis then can we conclude that $29 billion is “too high” or even “too low”?  We’re supposed to be impressed by a number that exists in a vacuum and that lacks any empirical context by which to evaluate it.

The $29 billion cost study also assumes that all litigation transaction costs are deadweight losses, which would mean that the entire U.S. court system is a deadweight loss according to the terms of this study.  Every lawsuit, whether a contract, tort, property, regulatory or constitutional dispute is, according to the assumption of the $29 billion cost study, a deadweight loss.  The entire U.S. court system is an inefficient cost imposed on everyone who uses it.  Really?  That’s an assumption that reduces itself to absurdity—it’s a self-imposed reductio ad absurdum!

In addition to the methodological problems, there are also serious concerns about the trustworthiness and quality of the actual data used to reach the $29 billion claim in the study.  All studies rely on data, and in this case, the $29 billion study used data from a secret survey done by RPX of its customers.  For those who don’t know, RPX’s business model is to defend companies against these so-called “patent trolls.”  So, a company whose business model is predicated on hyping the threat of “patent trolls” does a secret survey of its paying customers, and it is now known that RPX informed its customers in the survey that their answers would be used to lobby for changes in the patent laws.

As every reputable economist or statistician will tell you, such conditions encourage exaggeration and bias in a data sample by motivating participation among those who support changes to the patent law.  Such a problem even has a formal name in economic studies: self-selection bias.  But one doesn’t need to be an economist or statistician to be able to see the problems in relying on the RPX data to conclude that NPEs cost $29 billion per year. As the classic adage goes, “Something is rotten in the state of Denmark.”

Even worse, as I noted above, the RPX survey was confidential.  RPX has continued to invoke “client confidences” in refusing to disclose its actual customer survey or the resulting data, which means that the data underlying the $29 billion claim is completely unknown and unverifiable for anyone who reads the study.  Don’t worry, the researchers have told us in a footnote in the study, they looked at the data and confirmed it is good.  Again, it doesn’t take economic or statistical training to know that something is not right here. Another classic cliché comes to mind at this point: “it’s not the crime, it’s the cover-up.”

In fact, keeping data secret in a published study violates well-established and longstanding norms in all scientific research that data should always be made available for testing and verification by third parties.  No peer-reviewed medical or scientific journal would publish a study based on a secret data set in which the researchers have told us that we should simply trust them that the data is accurate.  Its use of secret data probably explains why the $29 billion study has not yet appeared in a peer-reviewed journal, and, if economics has any claim to being an actual science, this study never will.  If a study does not meet basic scientific standards for verifying data, then why are Reps. DeFazio and Chaffetz relying on it to propose national legislation that directly impacts the patent system and future innovation?  If heads-in-the-clouds academics would know to reject such a study as based on unverifiable, likely biased claptrap, then why are our elected officials embracing it to create real-world legal rules?

And, to continue our running theme of classic clichés, there’s the rub. The more one looks at the actual legal requirements of the SHIELD Act, the more, in the words of Professor Risch, one is left “scratching one’s head” in bewilderment.  The more one looks at the supporting studies and arguments in favor of the SHIELD Act, the more one is left, in the words of Professor Risch, “scratching one’s head.”  The more and more one thinks about the SHIELD Act, the more one realizes what it is—legislation that has been crafted at the behest of the politically powerful (such as an Internet company who can get the President to do a special appearance on its own social media website) to have the government eliminate a smaller, publicly reviled, and less politically-connected group.

In short, people may have legitimate complaints about the ways in which the court system in the U.S. generally has problems.  Commentators and Congresspersons could even consider revising the general legal rules governing patent ligtiation for all plaintiffs and defendants to make the ligitation system work better or more efficiently (by some established metric).   Professor Risch has done exactly this in a recent Wired op-ed.  But it’s time to call a spade a spade: the SHIELD Act is a classic example of rent-seeking, discriminatory legislation.

21 responses to The SHIELD Act: When Bad Economic Studies Make Bad Laws


    While I understand the academic interest in research veracity, focusing on the math ignores the real issue. The problem with the DeFazio-Chaffetz SHEILD bill is that it is overly focused on trolls. The fact is, infringement litigation as a whole (troll and non-troll) is out of control; therefore, the “plaintiff pays if they lose” policy and the escrow account requirement should apply to ALL patent infringement suits, regardless of who the plaintiff is. There are WAY too many frivolous infringement suits. If the bar were raised and a penalty imposed, then those using the court system for blackmail (either to get a ransom or to impede a competitor) will be less inclined to abuse the system unless they think they are likely to win (in which case it isn’t blackmail and it’s a legitimate use of the courts to rectify infringement). As a result, the true trolls — those who are ransom seekers without genuine infringement cases — likely will disappear because the business model no longer works. The strategic NPEs — whether hired by a single innovator or a large company — will survive, allowing *justifiable* infringement suits to be pursued. (Yes, there are some NPEs out there that are not evil and do serve a legitimate purpose.)

    BTW, just as there are too many frivolous patent *infringement* cases, there are too many frivolous *patents* and a similar raising-the-bar approach would help there too. As I blogged about 2 years ago (, USPTO fees should be higher so that innovators/companies pursue a patent only when the amount that the market is likely to value the innovation (and generate revenue in some way) is greater than the cost to process a patent.


    29 billion sounds like allot yet Its a small fraction of what the original parent concievers invention produced for that infringing company. Its sad that business wants to use the invention for free destroying the incentive to create in inventors. They know the little guys inability to fight back in court due to indegency. The fact that inventors have to sell out cheap to trolls is disgusting and Says uspto enforcement of no infringing needs drastic improvement. They are actually are villanized legal enforcers or patrollers as I originally coined the term. They are no wherse than vulture capitalists that swoop in and rob companies of there profits.

    Timothy Platt 20 March 2013 at 1:05 am

    ”Loser pays” does not apply in the US, as it does in courts in the rest of the developed world and IT patents are very easy to get and more and more are granted every year. Suing or threatening many with weak patents is a profitable business in the US and nowhere else.


      This type of spurious “logic” is analogous to the old fable and lesson of the 1,000’s of lemmings running off the cliff:

      “Why; if everyone else is doing it, that must be what I’m supposed to do, too.”

      That other countries don’t do what we do . . . doesn’t make them right and us wrong.

      And have you ever tried to get an “IT” patent? I have, and it was tough as nails.

      And — even if you’re correct — the reason why more and more are granted every year?

      Because there’s more and more innovation.

      Something which is a great thing — including substantial new job creation — indeed.

    Ted Sichelman 19 March 2013 at 9:58 pm

    Very well done, Adam!


    All this talk about “trolls” and I still can’t go to the USPTO website and search only live patents and applications in a single database. Public PAIR has a Status field. So how hard can it be to include it as a search field in the two databases? Just add disclaimers about the uncertainty created by e.g. Term Extension and powers to revive patents. We evaluate risk all day everyday. So why not let us make our own judgement about the risks involved with relying on the search results?

    As for patent validity: applying standards such as obviousness and the presumption of validity that emerged from centuries long mechanical and chemical innovation should never have been applied to new fields technologies. This opened the gates to an inevitable land grab the USPTO (i.e. examiners) were powerless to stop. Where were they going to get their prior art citations from? A slightly different test for obviousness applied to new fields tech like software and biotech could have helped.

    The most basic tools for avoiding patents and avoiding bad patents aren’t readily available.


    Adam and Michael are the experts here, but my impression is that what people are complaining about in complaining about patent “trolls” is not so much non-practicing entities — every garage inventor with only the resources to build a prototype is a NPE — but *bad patents*, patents that should never have been granted. I think that is where our focus should be, not on whether the inventor owns a manufacturing plant.


    As a patent attorney, Ifind myself infuriated every time I hear the arguments targeted at NPEs and their enforcement of their patents – call them trolls or whatever you’d like. The fact is that somewhere along the line, someone created intellectual property protectable under the patent statute. If you’d like to argue that some of those patents are invalid, unenforceable, improperly granted, etc., I’m completely on board with that. However it should make absolutely no difference to the question of enforcement whether the patent owner is practicing the invention or whether the owner invented the invention or simply acquired it from another party (thereby permitting the original inventor to monetize his invention).

    How often do we hear that intellectual property shares many attributes with real property? If I own a piece of land and choose not to build a house on it, that doesn’t allow someone else to build a house on it (or even walk across it) without my permission or without paying me for the privilege of trespassing on my property. There is no logical reason to view patents any differently. The financial effect on the potential infringer/trespasser is not a good enough reason to abrogate the property rights of the patent/land owner.


    As a former patent lawyer and now a cofounder of a self-funded startup company who is currently being threatened by this: for trying to help schools put their lunch menu data online . I have been on both sides of the coin and honestly believe that the patent system hurts innovation more often than it helps it along, at least in the world of web software. The SHIELD act might not be the right solution, but hopefully patent practitioners and academics can put their heads together and help find a fix, or eventually I think that people will throw the baby out with the bath-water.


    Adam – great post! To expand on the point about companies with large portfolios but no current implementation of a particular patent: of course it often happens that companies set out to innovate in certain areas and discover that their intentions have different or broader application. I can think of no reason we would want to deter this. Fundamentally, R&D is risky. Sometimes it works out as hoped, other times it fails, and other times it produces unexpected results. The ability to commercialize even the unexpected results (even if the inventing company has no means of direct implementation) induces risk taking and may even facilitate the very investment that enables the firm to pursue its intended aims. Labeling this “troll” behavior is absurd.


    BadLaw’s comments reminds one of that popular car bumper sticker proudly proclaiming to passersby:

    “I don’t know where I’m going, but I’m making great time.”

    Movement … change … action … for their own sake is of course illogical. Any laws attempting to do such must be carefully examined, vetted, discussed, and considered. Who will they help? Who will they hurt? Who’s supporting them; and why?

    Far from being merely “imperfect,” this act is in fact unconstitutional.

    And for good reason.


    Oh please BadLaws, give us a break. Any movement against “trolls,” even a good try, is good for us per se? Really? There is a saying, garbage in, garbage out, that applies. Professor Mossoff does us all a favor by pointing out that the SHIELD paradigm is fatally flawed. Bad data, bad law? And, despite agreeing with him, Badlaws then makes an ad hominem attack against Professor Mossoff for not coming up with his own plan to “save” the system! Even Aristotle is reeling at that logic!

    We need a healthy debate before rushing even more patent legislation through Congress. Universities and many other patentees are not trolls, and calling them so is Orwellian doublespeak.


    There’s no such thing as a perfect law when man is involved. At least the Act is an action. If it put’s spikes in the road of freeby seeking profit-mongers, then it’s at least a step. Even the bill’s supporters admit it isn’t perfect. As to the flaw of the data, it is possible to poll court rulings for some-odd years back to see what happened. However, by the time that the data is actually recorded, verified, analyzed and presented in bill form, there might be a patent out there that prevents its use and makes it invalid. Then it will be too late to try anything.

    Opinions make people individuals. They cause differences that result in solutions and compromise. Nothing is wrong with the analysis in the article in my opinion. The problem lies in the fact that there are no alternate solutions presented, no alternate data sources or figures and no mention of action to correct the issue at hand. In other words, the legislatures might not be making progress with their law, but they’re trying. Unlike this article.

      Michael Pruden 18 March 2013 at 5:55 am

      The road to hell is paved with good intentions.

      In terms of the possibility of someone patenting the recordation, verification, analysis, and presentation of data, see Bilski v. Kappos.

      The article also does suggest an alternative, e.g. a law which applies to everyone equally and doesn’t single out one group.

Trackbacks and Pingbacks:

  1. Can We Stopping the ‘Trolling’ of the American Innovation System? | Justin Badlam - April 12, 2013

    […] is an admirable step by Congress, although some would argue otherwise. If Congress does indeed move forward with the legislation, there needs to be responsible and […]

  2. The SHIELD Act: When Bad Economic Studies Make Bad Laws | CPIP - April 12, 2013

    […] at Truth on the Market on March 15, […]

  3. How to Neuter Patent Trolls | Web Guru Guide - March 26, 2013

    […] as commentators have pointed out here and here, the Shield Act would create as many problems as it would solve if it became law. The bill […]

  4. How to Neuter Patent Trolls | Build Your Own Social Network Website Tips - March 26, 2013

    […] as commentators have pointed out here and here, the Shield Act would create as many problems as it would solve if it became law. The bill […]

  5. RANDomness | The Essential Patent Blog - March 22, 2013

    […] days ago we linked to a blog post by Prof. Adam Mossoff of George Mason University, in which Prof. Mossoff criticized both the SHIELD […]

  6. If at First You Don’t Succeed: House Reps. Reintroduce the SHIELD Act | JETLaw: Vanderbilt Journal of Entertainment & Technology Law - March 20, 2013

    […] fraction of licensing fees to original inventors.  The term “patent troll” is itself controversial, with most academics now preferring to use “Patent Assertion Entities” (PAEs) to refer […]

  7. Hold the Phone, Someone Put Out Bad Numbers? | IP Troll Tracker - March 18, 2013

    […] was bound to happen, and it finally did.  Someone is disputing the claim that patent trolls cost companies $29 Billion in damages.  I’ve used the claim myself so […]