Abstract: This paper presents the first systematic theoretical and empirical study of high-low
agreements in civil litigation. A high-low agreement is a private contract that, if signed by
litigants before the conclusion of a trial, constrains the future damages payment to lie
between a minimum and a maximum amount. Whereas the existing literature describes
litigation as a choice between trial and settlement, our examination of high-low
agreements—a relatively new phenomenon in civil litigation—introduces partial or
incomplete settlements. In our theoretical model, trial is both costly and risky. When
litigants have divergent subjective beliefs and are mutually optimistic about their trial
prospects, cases may fail to settle. In these cases, high-low agreements can be in the
litigants’ mutual interest because they limit the risk of outlier damages awards while still
allowing for an optimal degree of speculation. Using unique insurance claims data from a
national insurance company, we describe the features of these agreements and empirically
investigate the factors that may influence whether litigants discuss or enter into high-low
agreements. Our empirical findings are consistent with the predictions of the theoretical
model. We also explore extensions and alternative explanations for high-low agreements,
including their use to mitigate excessive, offsetting trial expenditures.