The words “patent troll” brings a shudder to the ranks of Silicon Valley executives. These are non-practicing entities (NPEs) whose main business is to own patents — enforcing their rights by charging fees or suing. Critics believe that their behavior impedes innovation because it will make others hesitant about developing new technology in the same area. Legislation is underway on the federal and state level to rein them in. But the paper, “The NPE: Benevolent Middleman or Stick-Up Artist?” points to evidence that patent trolls can actually help innovation, as well as potentially impede it, according to authors David S. Abrams, a professor of business economics and public policy at Wharton and the University of Pennsylvania Law School, Ufuk Akcigit of the University of Chicago and Gokhan Oz, Penn economics graduate student.
Knowledge at Wharton recently spoke to Abrams to discuss his research findings. The research is supported by Wharton’s Mack Institute for Innovation Management.
An edited transcript of the conversation follows.
Defining Patent Trolls
You think that’s an easy first question, but it turns out even the definition of a patent troll is a bit controversial. So they go by … NPE for non-practicing entity, and PAE has become popular recently for patent assertion entity. But essentially, what all these terms are trying to [describe] are companies that hold patents, that make their money by using patents, but that don’t make products that involve those patents.
Focus of the Research
I have a working paper called “The NPE: Benevolent Middleman or Stick-Up Artist?” The title tries to capture what I think are the two main competing theories at the moment about the NPEs’ — or patent trolls’ — impact on innovation. The idea behind the benevolent middleman story, which is one that is often told by economists, says “Look, NPEs are simply middlemen like in all kinds of markets and we don’t think middlemen are necessarily bad. In fact, they can improve efficiency by helping match up buyers and sellers.” And of course, they are taking a cut for themselves while they are doing so. But they’re not necessarily harmful.
“There’s some evidence for the middle-man theory — they seem to help reallocate patents to companies where they are most useful.”
Contrast that with the stick-up artist idea — which is very popular now, especially in Silicon Valley — that patent trolls are inherently harmful, that they are impeding innovation by either suing or extracting large license fees and making it very difficult for companies to innovate. And those are the two dominant perspectives on NPEs that you hear nowadays.
In this paper, which I co-authored with a University of Chicago professor, Ufuk Akcigit and a Penn Econ grad student, Gokhan Oz, we use a really large empirical data set that we got from NPEs to try to inform this question and look at it empirically. Basically, we try to let the data help us understand something about this big argument.
Things in reality are a bit more complicated than they are often portrayed, with people often arguing out of self-interest or a kind of ideological perspective. We come at it from a pretty neutral perspective — just trying to understand what is going on. We did find a few things; this work is still preliminary and we’re still trying to make sure that we understand everything right. But one takeaway is that it does look like there is some evidence for the middleman theory — [NPEs] seem to help reallocate patents to companies where they are most useful.
But we also have some other evidence that contradicts that a little, or at least supports the other theory in that it looks like patents get cited a little bit less after they get acquired by NPEs. What that might mean is that other companies may be getting scared off these kinds of patents. They stop innovating, stop working in that area, and therefore, stop citing the patents that are acquired by NPEs. So that gives a little bit of support for the stick-up artist theory. And we have a few other pieces of evidence as well that are suggestive of one or the other.
The bottom line is we do a few different types of analyses to try to test and differentiate between these two different theories. We find some evidence that supports one and some evidence that supports the other, which in my mind, tells us that these two theories are both too simplistic.
Defining Patent Citations
Every time someone files a patent application, they cite the previous literature that they’re building upon and that’s often used. This actually relates to a previous paper that I’ve worked on that more directly examined citations. It’s important because economists and business school professors and people who are interested in innovation and growth use this as their main measure of the value of innovation. The notion is that if a lot of patents are citing a single previous patent — so the previous patent gets a lot of what is called forward citations — that means it’s very valuable because a lot of people are building on it. And so the fact that the number of citations might drop could indicate that you’re inhibiting people from doing work in this area.
“The biggest implication is for legislators who have been debating several bills aimed at reforming patent law and particularly, reigning in NPEs.”
Citations have been used across a variety of fields to proxy for the value of innovation based on this notion that more citations mean that the underlying technology is more valuable. We’re actually able to test that hypothesis and find a relationship that is more complicated than has been widely believed — that simply, more citations equal more value. And in fact, we found that highly valuable patents actually don’t receive that many citations. That was kind of a surprising finding in this other work using patent data.
[In this paper looking at competing views of NPEs,] the fact that there is support for … the middleman role has got to be surprising to a lot of people who are dead certain that they are bad for the world. This doesn’t mean that some of them or many of them might not be doing harmful things for the world, but at least in this kind of middleman function, it seems like they are helping patents to get to firms where they are more likely to be useful, that seems more productive. And then, at the same time, [evidence to] support that they may also be inhibiting innovation has got to be surprising to the people who think, “well, let the market function and all will be well.”
The biggest implication is for legislators who have been debating several bills aimed at reforming patent law and particularly reigning in NPEs. So far, I think major legislation hasn’t passed, but this is something that’s a hot topic in Congress nowadays.
What this says to me is that before you pass a law and before you make any major changes to patent law, we really need to learn more about the impact of NPEs in markets and we need to think very hard about what the potential reforms may do — not just to the existence of NPEs, but more importantly, to the innovating companies and for the ability of maybe smaller companies to sell their patents and potentially capitalize on them especially if they’re up against much larger competitors.
That’s where this has immediate impact. It’s going to matter to legislators, it’s also going to matter, obviously, to inventors both in small entities and in large ones. I think the bottom line is we need to know a bit more. … But I can say this — I don’t think NPEs are all good or all bad and any legislation that is premised on either of those views is misguided.
“I don’t think NPEs are all good or all bad and any legislation that is premised on either of those views is misguided.”
What Sets the Research Apart
The biggest thing that sets us apart is that I was very lucky to be able to get some large data sets from large NPEs that gave me access to individual-level patent data and licensing revenues, and a lot of other information. That’s just something that nobody else has had access to before. That’s the only reason we’re able to get this kind of insight into these questions due to the unique nature of the data. So my co-authors and I are doing everything we can to try to learn as much as possible from these data sets, because these kinds of companies tend to be very secretive.
There are a few other topics of interest. Probably, the next one will be to try to use machine learning and maybe other approaches to find out what makes for more valuable patents. Are there observable correlates that we can find either from patent data or industry data? Or otherwise that tends to make patents more valuable? This is something that would be of interest to innovators and investors and then, of course, researchers, as well. So that’s next on the list.