The number of banks downgraded to problem status in recent years has been
substantially larger than the number of bank failures. During a period of few
bank failures, the relevance of this bank failure model for surveillance depends to
some extent on the accuracy of the model in predicting which banks will have
their supervisory ratings downgraded to problem status in future periods. This
paper compares the ability of two models to predict downgrades of supervisory
ratings to problem status: the Board staff model, which was estimated to predict
bank failures, and a model estimated to predict downgrades of supervisory
ratings. We find that both models do about as well in predicting downgrades of
supervisory ratings for the early 1990.