Law and economics theorists have long advanced theories of litigation and settlement, including the canonical Landes-Posner-Gould (LPG) and Priest and Klein (PK) models. Famously, PK predict that, as settlement rates rise, plaintiff win rates approach 50%. Empiricists have tested this and other predictions from the theoretical literature, finding qualified support for the PK model. So far, though, empirical testing of these models has been hampered by two major limitations: First, these models make clear predictions about the effect of case stakes on settlement rates and plaintiff win rates, but lack of reliable data on stakes means these predictions have gone untested. Second, most of the studies have used data from the U.S., a high-settlement, high-litigation-costs setting, and the generalizability of these models to other institutional settings has been less explored. In this paper, we use a novel dataset of Taiwanese court data to test previously untested predictions of the LPG and PK models and explore the extent to which these models apply to a low-settlement, low-litigation-cost setting. We find strong support for the predictions of the LPG model that we test. We find at best weak support for the 50% hypothesis of the PK model, consistent with recent research suggesting that the hypothesis will have limited applicability in a low-settlement, low-litigation-cost environment.
Chang, Yun-chien and Hubbard, William, "New Empirical Tests for Classic Litigation Selection Models" (2018). Coase-Sandor Working Paper Series in Law and Economics. 80.