| After recent crises occurred in South East Asia, Russia, and Brazil, balance of payments crisis prediction and understanding have become increasingly important. Chapter one of this dissertation models how problems arising in the banking sector are transmitted to the rest of the economy. Under the assumption that firms must borrow from banks in order to finance their operations, we show how a run on banks may prove harmful for the economy at large. The second chapter compares the crises that occurred in Argentina and Mexico in 1995, and concludes that former predicting models failed to foresee them. Finally, a new crisis-forecasting model is constructed in chapter three, which is successfully tested within sample and out of sample. The model is based on a set of monthly indicators that produce warning signals once they surpass certain thresholds. An accumulation of signals produces a monthly alarm and three monthly alarms in a 12-month period is consider to flag an upcoming crisis. The model successfully anticipated 50% of in sample crises, and five out of six out of sample crises. |