Hovenkamp and Masur on the Unreliability of Comparable Licenses

Erik Hovenkamp and Jonathan Masur have posted a paper on ssrn titled Reliable Problems from Unreliable Patent Damages.  (This is an updated version of the paper they presented at the University of Texas Patent Damages Conference back in June.)  Here's a link to the paper, and here is the abstract:

If a litigated patent has previously been licensed to a third party, the courts generally adopt the terms of the prior agreement as the best measure of damages. However, while administratively convenient, this “licensing-based damages” standard creates problematic incentives and undermines the efficient commercialization of patented inventions. It rests on the trivialized (and generally false) presumption that a patent license is like a commodity, with the patentee charging a common price to all comers. As a consequence, patentees distort their future recovery prospects – and by extension the outcomes of future licensing negotiations – whenever they license their patents, whether or not today’s agreement will be a good proxy for tomorrow’s dealings or disputes. Knowing this, patentees are discouraged from licensing at anything less than a high royalty rate, even if they could reach many additional mutually-beneficial agreements on more modest terms. The result is that patent holders rationally cut off the bottom segment of the licensing market, creating substantial deadweight loss. This injures not only patentees, but also prospective licensees and their consumers. The standard creates additional problems by encouraging secrecy and “gamesmanship” in patent licensing. 

We propose that the licensing-based damages standard be abandoned, and that damages should generally be awarded ad hoc. This does not mean that private parties should ignore comparable licenses in their private dealings; it simply means that courts should not use them as a measure of damages. That this necessitates some speculation does not suggest it is the less desirable approach, for it is better that damages be somewhat random than systematically harmful. Further, while the licensing-based damages standard is clearly easy to apply, there is little reason to believe it is accurate in a typical case. As such, its apparent lack of randomness does not suggest that it is producing good results.

The authors make a number of important points about why comparables often aren't a very realistic indicium of the reasonable royalties the court should award in a given case--and how their use may in turn distort the market for patent licensing.  In my paper Patent Damages Heuristics I cite Masur's and Hovenkamp's previous solely-authored work highlighting some of the flaws in using comparable licenses, as well as an earlier version of their coauthored paper--though I'm still inclined to believe the appropriate use of comparables can sometimes be a justifiable time- and cost-economizing means for approximating the correct royalty (and thus I wouldn't advocate discarding the use of comparables altogether).  Another possible approach (which is not exclusive to the authors' proposal, and which they appear to support at p.29) would be to make more information on actual licenses publicly available (as this proposal recommends).  Whether private parties will be motivated to make substantially more of these available, however, remains to be seen.