Béatrice Dumont has published a paper titled Does Patent Quality Drive Damages in Patent Lawsuits? Lessons from the French Judicial System in 11 Review of Law & Economics 355-383 (2015), available here behind a paywall. Here is the abstract:
This paper presents an empirical analysis of the link between patent damage awards and patent quality, based on information from suits filed during the period 2008–2013 in France. Our results show that civil damage awards attributed by French judges are significantly linked to patent quality, as assessed from a statistical point of view, but this link is imperfect.
The article reports an empirical analysis "based on a detailed assessment of enforcement suits handled at the Court of first instance of Paris," comprising "483 patent infringement suits encompassing 673 patents" filed in the Tribunal de la grande instance de Paris from 2008 to 2013. From these 483 cases, Professor Dumont eliminated those involving "disputes over license contracts, legal arguments about compensation of employees' inventions, property claims and other cases where no infringement took place," leaving 379 cases. Of these "379 cases decided on the merits, 102 were won by the patent holder, i.e., 26.91% of the cases." Mean damages awarded in these winning cases of €323,270, and median damages were €60,000.
To test the hypothesis that damages "are in line with an objective measure of the quality of the patents involved in litigation cases," Professor Dumont begins by positing various proxies for patent quality (including family size, claims, and backward and forward citations) and uses a factor model "to compute an estimate of the overall quality index of each litigation case." Second, she finds that the outcome of litigation (did the plaintiff win or not?) "relies rather on idiosyncratic unobserved factors than on the few explanatory variables used in the model, and thus remains mainly unpredictable. This may be thought of as a positive result in the sense that it means that the judge is neither influenced by the context (at least by the sector) nor by an a priori quality of patents, but that he rather examines in detail the case and is receptive to the collection of reliable proofs, notably those conferred by the search and seizure proceeding" (p.377). Third, she finds that the damages awarded in the cases that plaintiffs won "are based on objective grounds, related to the quality of the patents involved in the case" (pp. 378-79). She concludes:
"Our results show that the outcome of patent litigation is mainly influenced by idiosyncratic elements that make it unpredictable. However, our results do not support the claim that judges are unable to assess patent quality. Indeed, we demonstrate that damage awards attributed by French judges are positively and significantly linked to patent quality, as assessed from a statistical point of view. . . .
". . . [W]hen patents are valid and declared to have been infringed, their value is correctly assessed by the judges."