Popular insurance models of judicial independence contend that electoral competition induces executives to establish or maintain independence as insurance against the risks associated with losing office. Existing accounts, however, focus only on variation in the likelihood of losing office, treating risks associated with losing as constant. This inattention to the model’s causal logic limits theoretical development and empirical conclusions. We model the demand for insurance rather than simply the likelihood of losing office, with empirical implications tested via instrumental variables. This paper offers a major development of the insurance account, with important implications for the study of judicial independence.