Bank Failure Prediction : A Two-Step Survival Time Approach

In this paper we develop a novel bank failure prediction approach that uses the output of a multiperiod logit model to assess banks' risk situations and then estimates a survival time model for the subset of at-risk (quot;illquot;) banks. Our empirical analysis reveals that this two-step approach significantly outperforms benchmark logit models with respect to out-of-sample prediction performance. Furthermore, we identify important differences between the subset of at-risk banks and the entire population of banks regarding the set of significant predictive variables. Management efficiency and size relative to geographically close competitors, for example, are important default predictors for at-risk banks but not for the entire banking population. Consequently, these results have notable policy implications for the design of off-site banking supervision