The early warning system (EWS) literature on banking crises usually relies on linear
classifiers, estimated with international datasets. We construct an EWS based on an artificial
neural network (ANN) model, and also account for regional heterogeneity in order to improve
the generalization ability of EWS models. All of the banking crises in our test set are then
predictable at a 24 month horizon, using information from earlier crises. For some countries,
estimation with a regional dataset significantly improves the predictions. The ANN
outperforms the usual logit regression, assessed by the area under the receiver operating
characteristics curve.