Manasseh Zechariah, economist and assistant professor at Johnson & Wales University College of Business, decided to take groups of decisions from the CAFC, and run variables related to those decisions through different theoretical models to see what came out. Here are some highlights from the study:
• Cases that gets litigated at the trial court which were brought by the patentee are more likely to be decided in favor of the infringer on appeal, as opposed to cases brought to trial by the infringer (i.e., declaratory judgments). Also, a case is likely to be lost at the appeals court by the litigant who brought the case to trial in the first place. However, Zechariah notes that the study does not take into account cases that have not gone through trial (e.g., summary judgment).
• A pro-patent decision by a lower court is more likely to be decided pro-patent by the CAFC if it is appealed. Similarly, an anti-patent decision by a lower court is more likely to be decided anti-patent by the CAFC.
• Overall, the CAFC is likely to rule anti-patent relative to the trial courts on average, but different technology classes of patents are treated differently. While chemical, mechanical and computer patents are more likely to receive an anti-patent ruling, the CAFC has been mostly pro-patent for drug patents.
• Republican appointed judges are more pro-patent than democratic judges.
• In general, patent cases are not likely to be appealed once they have gone through the trial stage. Chemical, electrical and drug patents have higher rates of appeal, but computer patents are less likely to be appealed.
From Zechariah's abstract:
The enforcement of patent rights affects the scope of the monopoly rights provided by them and hence the incentives they provide. Greater likelihood of a pro-patent decision by the enforcement mechanism provides a larger scope of monopoly rights. I investigate the likelihood of a pro-patent decision by the Court of Appeals for the Federal Circuit in patent cases to see how it varies with characteristics of the patent, the litigants and the litigation itself. This model corrects for the fact that not all patent cases get appealed after trial by incorporating a selection model for the appeal of cases. I also incorporate the litigants’ uncertainty about appealing a case using a heteroscedastic selection model. I estimate the model using simulated maximum likelihood and use the GHK simulation method. Results show that overall the Court of Appeals for the Federal Circuit is an anti-patent court. However, significant differences exist between different classes of patents with drug patents being more likely to be decided pro-patent than software patents. The Court of Appeals for the Federal Circuit is found to defer to the decisions of the lower court. We also see an increased likelihood of a pro-patent decision by the appeals court with an increase in the proportion of republican judges on the bench.
Read "An Empirical Analysis of the Adjudication of Patent Cases by the Court of Appeals for the Federal Circuit" (link)