The Unexpected Value of Litigation

Litigation is a sequential process. Litigants learn new information. They adjust their strategies in response to that information. The dominant economic models of litigation, however, assume that litigation is a single stage process, and that adjustments to the mean and variance of a lawsuit's payoffs are adequate to capture the implications of the learning, adaptation, and uncertainty that characterize a lawsuit. We suggest that this reductionist assumption oversimplifies the litigation process to a degree that no longer accurately describes its reality and that causes single-stage models to generate predictions that diverge dramatically from the behavior of rational litigants who understand litigation's procedural realities. To demonstrate this point we present a two-stage model in which risk neutral litigants have no private information and in which the claim's expected value is held constant throughout the analysis. Information is revealed to the litigants at the beginning of the second stage. The plaintiff then has the option of abandoning the lawsuit contingent on the information that is disclosed. The model thus constitutes an application of real option theory to the litigation process. We demonstrate that the equilibrium settlement value of this lawsuit can diverge dramatically from the same lawsuit's settlement value when modeled as a single stage process. It follows that the traditional expected value model contains simplifying assumptions that obscure important implications of the learning process that commonly characterizes litigation. The model also demonstrates that settlement value is a function of the variance of the unknown information, even though the litigants are risk neutral. Variance, in this model, acts as a proxy for the value of the option to react to information. The larger the variance the more valuable the option and the larger the settlement that a rational defendant will pay. The model also demonstrates that settlement values can be a discontinuous function of the information's variance; that negative expected value lawsuits are credible in a broad range of circumstances; and that changes in the allocation of litigation costs over time can significantly affect settlement values even when total expenditures are held constant. The model also provides a technique for the evaluation of the implications of procedure qua procedure in the litigation process.