Michael J. Mazzeo, Jonathan Hillel, and Samantha Zyontz have published Explaining the "Unpredictable": An Empirical Analysis of U.S. Patent Infringement Awards, 35 Int'l Rev. L. & Econ. 58 (2013). (I attended a presentation on an earlier version of this paper a couple of years ago, but I thank Professor Siebrasse for bringing the published version to my attention.) People who are interested in U.S. patent damages should definitely read the article. Here's the abstract:
Patent infringement awards are commonly thought to be unpredictable, which raises concerns that patents can lead to unjust enrichment and impede the progress of innovation. We investigate the unpredictability of patent damages by conducting a large-scale econometric analysis of award values. We begin by analyzing the outcomes of 340 cases decided in US federal courts between 1995 and 2008 in which infringement was found and damages were awarded. Our data include the amount awarded, along with information about the litigants, case specifics and economic value of the patents-at-issue. Using these data, we construct an econometric model that explains over 75% of the variation in awards. We further conduct in-depth analysis of the key factors affecting award value, via targeted regressions involving selected variables. We find a high degree of significance between award value and ex ante-identifiable factors collectively, and we also identify significant relationships with accepted indicators of patent value. Our findings demonstrate that infringement awards are not systematically unpredictable and, moreover, highlight the critical elements that can be expected to result in larger or smaller awards.
From just a purely descriptive point of view, the paper is useful because it includes statistics on the number of damages awards per year, the median and mean damages awards by year, and the distribution of damages awards by year, for the years 1995-2008. The factors that appear to have the greatest association with higher damages awards are the number of patents in suit, the average age of the patent, the average number of claims, the average number of forward citations, whether the defendant is a public company, whether the trial was to a jury, and the time to trial. The authors also show that "all else equal, damage awards have been decreasing over time. . . [A]lthough recent awards such as the billion dollar Apple-Samsung verdict used to be unheard of in patent cases, such cases might not reflect a systematic increase in award values across the dataset."
As the authors note, if damages awards were systematically unpredictable/random, that would undermine the "incentive structure underlying the patent grant" because "either the system of granting patents, or the system of litigating them, or both, are fundamentally flawed." I would add a caveat, however: just because damages awards may be more predictable than is sometimes believed doesn't necessarily mean that they are correctly determined. I could imagine a system that predictably either over- or under-rewarded patent owners, in comparison to their actual loss (profit the patent owner would have earned but for the infringement in light of the noninfringing alternatives that were available to the defendant, or the royalty that reasonable parties would have negotiated ex ante). Moreover, I'm not sure I would go quite so far as to state that because "several of the driving factors," such as number of forward citations, "correspond to accepted indicators of patent quality," the authors' findings "bolster the core tenets of the patent system, namely that 'the progress of . . . the useful arts' should receive legal protection and that the exclusive patent rights are an appropriate means for such purpose." It may be that high-quality patents tend to result in higher damages awards; indeed, one would hope that to be true. (On the question of whether forward citations are indeed a good measure of patent value, see this paper by Abrams, Akcigit & Popadak, which posits an inverted-u relationship between patent value and citations; though I would tend to think that the high-value defensive patents that, according to Abrams et al., get cited very little have much higher private than social value.) But I still think we need to proceed with caution: unpredictable may be bad, but predictable doesn't necessarily mean correct. And, while I tend to agree that patents sometimes may be superior, for various reasons, to other policy tools for inducing invention, I'm not convinced that the relative predictability of damages awards proves that hypothesis.
That may be true, but again just because damages are relatively predictable doesn't seem to me to demonstrate that they are, by and large, correct.