The Occurrence and Timing of Events: The Application of Event History Models in Accounting and Finance Research

Abstract

Event history analysis is a statistical method concerned with the study of events and the substantive processes that govern the occurrence and timing of events. Many research studies focusing on events use qualitative response models such as probit or logistic regression and typically examine the relation between the probability of an event and a set of covariates for a single year. Event history models, on the other hand, use longitudinal rather than cross-sectional data and focuses on the process that governs the occurrence and timing of events rather than the observation of a specific state. In this paper I review event history models and their applications in accounting and finance research. The paper discusses the issues of censored observations and time-varying covariates and reviews the more commonly used survival distributions. I provide a review of the trade-offs involved in choosing among non-parametric, parametric, and semi-parametric models as well as competing risk, multi-episode, and multi-episode/multi-state models.