5.2 Theoretical Considerations
5.2.2 Previous Empirical Work
The analysis of the determinants of settlement failure after detection of infringement has been undertaken extensively for the US (Lanjouw and Lerner, 1996 and 1998, Lanjouw, 1998; Lanjouw and Schankerman, 2001 and 2003; Ziedonis, 2003; Llobet, 2003; Somaya, 2003). Due to data constraints there is little published empirical research for Europe available (Stauder, 1989; Cremers, 2004.) The three mechanisms for settlement failure sketched above were partly validated in empirical studies.
Studies for the U.S. distinguish between suits dealing with attacks on validity and suits claiming an infringed patent. In Germany the District Courts presume validity of the infringed patent as long as no annulment suit has been filed at the Federal Patent Court (“Bundespatentgericht”).
Seminal empirical investigations of determinants of patent suits have been com-piled by Lanjouw and Schankerman(2001 and 2003) who show a strong positive impact of the expected value of a patent on the probability of litigation. Further-more, they report a higher settlement probability in cases where the patentee’s portfolio is very large. A specific investigation of the semiconductor industry has been undertaken by Ziedonis (2003). Lanjouw and Schankerman (2003) focus on the determinants of patent litigation and they investigate outcomes of cases and the timing of eventual settlement. They find, that the probability of being involved
in patent suit in the U.S. is very heterogenous. For individuals or companies with small patent portfolios the probability of litigation suits is much higher than for patentees with large portfolios. Lanjouw and Schankerman interpreted their results in the way that cooperative interaction induced by patents to trade induce more pre-trial settlements. The analysis of these authors reveals a detailed picture of the U.S. patent enforcement system. The heterogeneity of the European patent enforce-ment system contrary to a harmonized European patent granting system and raises the question whether this peculiarity causes a different pattern of the probability of patent litigation for certain patentees.
Trajtenberg (1990) and Harhoff et al. (2003a) stress the need to take account of the strategic aspects of patenting when assessing the value of patents, and not only the rents achieved by exploiting the patented innovation exclusively. This includes signaling technological capacities, blocking competitor’s development of new prod-ucts and building portfolios of patents as bargaining chips (Hall and Ziedonis, 2001), and as patent thickets protecting major inventions (Shapiro, 2001a; Hall, 2003).
The study of Harhoff and Reitzig (2004) focuses on the analysis of the determi-nants of opposition procedures against EPO patent grants. Opposition is an early form of patent dispute. Contrary to litigation suits they directly attack the validity of the opposed patent. Additionally, unlike litigation suits there is no direct dispute between the opposing and opposed party but a negotiation between the patent office and the opposed party. However, this study and an investigation of patent litiga-tion in Germany (Cremers, 2004) demonstrate that the value of the stakes included in the suit, namely the patent value and the expected payoffs of the suit, have a positive impact on the probability of non-settlement - i.e.that cases will be filed and brought to trial. The differences in the characteristics of the parties lead to diverging expectations about the value and the market payoffs.
Perloff et al. (1996) performed the first study of trial outcomes and settlement not involving individuals but firms. However, their focus is on testing whether risk aversion plays a role in determining the litigation rates (self selection) and settlement outcomes. For the outcomes of patent trials Lanjouw and Schankerman (2001) and (2003) and Somaya (2003) find that most of the factors correlated with the probability of settlement before the trial do not have any significant effect on the settlement probability once the trial has started. The choice of the court and
the correspondence among the parties and the court provide credible knowledge about the merits of the case and the stakes involved. Rational behavior will likely promote realistic expectations about the outcome of the trial. However, Lanjouw and Schankerman (2003) analyze the winning probability in certain technical areas and estimate settlement probability during trial. They find that more than 80% of all settlements occur before the pre-trial hearing and that the post-trial settlement does not significantly vary with portfolio size or ownership of the patents.
A more detailed analysis of different outcomes (settlement or court decisions at certain stages of a patent suit) is conducted by Somaya (2003). He investigates the impact of the strategic stakes the parties have in a litigation suit. These strategic stakes are measured for the patentee as the number of self citations of the litigated patent. For the non-patentee, the number of his citations to all citations is used as a measure of the strategic stakes. Somaya finds that the respective stakes matter in varying degrees for the computer and research medicine industries as far as the decision to proceed the trial or to settle is concerned.5
In the previous studies the reported empirical results with respect to the strategic stakes are ambiguous. Somaya (2003) reports a positive impact of strategic stakes on the probability to proceed to trial until a court judgement. The same interpretation is given by Fournier and Zuehlke (1989) who report a negative effect of costs on the settlement probability. Lanjouw and Schankerman (2003) find that the main charac-teristics of the patents and the patentees affect neither the probability of settlement nor the plaintiff’s win rate in trials. The study by Perloff et al. (1996) reports that risk aversion plays an important role in explaining settlement in antitrust suits.6 They also argue that the outcomes of antitrust trials vary between certain jurisdic-tions owing to the existence of different legal rules in different jurisdicjurisdic-tions, different enforcement of similar rules, or different attitudes toward plaintiffs on the part of judges. For the U.S. patent system, Farrell and Merges (2004) show that litigation is not a reliable tool to verify patent validity. Asymmetric stakes of the parties imply
5In the sample of Somaya (2003) research medicine includes patents from biotechnology, drug delivery systems, assays and dental innovation. The sample of computer patents contains semi-conductors, data storage, computer systems, I/O devices, computer applications, and networking technologies.
6They find that a change in the winning probability at trial towards 0.5 increases the variance and this in turn leads to an increase in pre-trial settlements.
skewed incentives to litigate and lead to high expenditures on litigation processes which influence the winning probabilities. Farrell and Merges (2004) argue that the efforts depend highly on the stakes involved. If there are asymmetric stakes, the ef-forts will be asymmetric and this shifts the probability of winning towards the party with the relatively high stakes involved. A study predicting winning probabilities in selected samples of litigation is presented by Waldfogel (1995). In detail, Marco (2004) refers empirically to the selection problem and he finds considerable differ-ences in the winning probabilities. I do not estimate the probability of plaintiff’s win at trial since I do not have data which cover a time period sufficiently long to estimate this precisely.
The empirical analysis of this Chapter is related to Somaya (2003). I test whether the strategic aspects of patent right enforcement play a role in the determination of settlement during trial or not. Furthermore I investigate the impact of early stage success of failure of the plaintiff on later stage settlement rates.