Mardi | 2009-12-08

Bertrand Candelon – Elena-Ivona DUMITRESCU – Christophe HURLIN – Jaouad MADKOUR

This paper proposes a new statistical framework inherited from the traditional credit-scoringliterature, to evaluate currency crises Early Warning System (EWS). Applied so as to assessthe predictive power of panel logit and Markov frameworks, it results that the panel logitis outperforming the Markov switching ones. Furthermore, the introduction of forward loo-king variables clearly improves the forecasting properties of the EWS. It thus con rms theadequacy of the second generation crisis models in explaining the occurrence of crises.