Econometric methods for financial crises

Known as Early Warning Systems (EWS), financial crises forecasting models play a key role in defining economic policies at microeconomic, macroeconomic and international level. This PhD thesis focuses on their improvement and the evaluation of their abilities to correctly forecast future turmoils. We show that models taking into account crisis dynamics (persistence) and the possibility that crises spill-over to other markets (currency; banking; sovereign debt; etc.) perform best in forecasting crises. Besides, we argue that adequate EWS evaluation significantly contributes to the improvement of financial crises forecast. The methodology proposed in this research should thus provide useful information to regulatory authorities for implementing policies to prevent or at least attenuate the impact of a financial turmoil.