On Monday, October 29th, Stanford University’s Hoover Institution Working Group on Intellectual Property, Innovation, and Prosperity (Hoover IP2) issued a revised working paper looking at the effects of patent assertion entities (PAEs) on the innovation economy in the United States. The paper, co-authored by Noel Maurer of George Washington University and Stephen Haber of the Hoover Institution, found that a collection of public companies identified by patent risk management firm RPX Corporation as PAEs don’t fit the hypothetical model of “patent trolls,” the moniker with which these companies are often maligned. Further, rather than frustrate innovation, Maurer and Haber found that these companies have research and development expenditures which, on average, are twice that of U.S. high tech firms.
The working paper starts off with a simple question: “Do firms that earn revenues from licensing patent portfolios, rather than producing physical products—often called patent assertion entities (PAEs)—frustrate or facilitate innovation?” To answer this, Maurer and Haber examined a sample of 17 years’ worth of U.S. Securities and Exchange Commission (SEC) filings made by 26 publicly-traded PAEs to estimate spending on litigation and patent acquisition. The authors relied on RPX’s definition of PAEs so that they didn’t have to create their own definition which might create selection bias. The authors’ methodology also included aggregation of firm-level data for 153 large high tech firms and an exploration of harmful aspects of PAEs as defined by government agencies during the Obama Administration and academic literature.
In the first stage of the authors’ analysis, they find that the public PAEs do not appear to operate in a manner consistent with the hypothesis on patent trolls, which includes the view that PAEs own patents which have no value and that they file frivolous lawsuits that amounts to a tax on innovation. Far from spending only a negligible amount on R&D expenditures, a hallmark of the patent troll hypothesis, these identified PAEs spent twice as much on R&D than 153 firms identified in PricewaterhouseCoopers’ 2017 Global Innovation 1000 study during the period between 2011 and 2016. 17 of the 26 identified PAEs spent either the same share or more of their revenues on R&D as major tech firms Apple or Hewlett-Packard.
The PAEs in the study also didn’t appear to be successful with a business model of obtaining money through nuisance lawsuits, another supposed characteristic of patent trolls. “If they were filing nuisance lawsuits using valueless intellectual property, then they would be highly profitable,” the authors note. To the contrary, as a group these 26 firms lost $3.1 billion between 2000 and 2016, with only six firms yielding positive returns for shareholders during that time. Further, these firms are modest in size, most having revenues “lower than a typical Safeway supermarket” and thus don’t pose the systemic risk to which many suppose that PAEs contribute.
In the second stage of the analysis, Maurer and Haber attempt to estimate the “innovation tax” potentially created by the group of PAEs. The findings in this stage indicate that the group of PAEs identified by RPX is too small to have much effect at all on the U.S. high tech sector. Revenues and litigation costs transferred from the high tech sector to the PAE group averaged 0.28 percent of revenues for the high tech sector between 2011 and 2016.
The authors also conducted a third stage of analysis in which they assessed publicly available information for three major private PAEs: Intellectual Ventures, Rockstar Consortium and Conversant IP. Not only do these firms behave and perform similarly to the identified public PAEs, the addition of the revenues and litigation costs stemming from these firms do not substantially change the magnitude of any potential innovation tax that PAEs create.
While the authors point out that they’re not claiming that patent trolls don’t exist, nor is it clear that the RPX-identified PAEs are helpful intermediaries in the market. However, by “operationaliz[ing] the characteristics of harmful PAEs” based on claims made by the government and academic literature to yield testable predictions and building and analyzing a dataset regarding identified PAEs, the authors found that “the testable predictions of the patent troll hypothesis are inconsistent with the data.” The authors also specify evidence that would render their analysis invalid as a guide for future research.