Forecasting bank leverage

Abstract

Standard early warning models to predict bank failures cannot be estimated during periods of few or zero failures, precluding any updating of such models during times of good performance. Here we address this problem using an alternative approach, forecasting the simple leverage ratio (equity/assets) as a continuous variable that does not suffer from the small sample problem. Out-of-sample performance shows some promise as a supplement to the standard approach, despite measurable deterioration in prediction accuracy during the crisis years.