Systemic Event Prediction by Early Warning System

This work develops an early warning system framework for assessing systemic risks and for predicting systemic events, i.e. periods of extreme financial instability with potential real costs, over the short horizon of six quarters and the long horizon of twelve quarters on the panel of 14 countries, both advanced and developing. First, we build Financial Stress Index to identify starting dates of systemic financial crises for each country in the panel. Second, early warning indicators for assessment and prediction of systemic risks are selected in a two-step approach; relevant prediction horizons for each indicator are found by the univariate logit model followed by the application of Bayesian model averaging method to identify the most useful indicators. Next, we validate early warning model, containing only useful indicators, for both horizons on the panel. Finally, the in-sample performance of the constructed EWS over both horizons is assessed for the Czech Republic. We find that the model over the 3 years’ horizon slightly outperforms the EWS with the horizon of 1.5 years on the Czech data. The long model attains the maximum utility in crises detection as well as it maximizes area under Receiver Operating Characteristics curve which measures the quality of the forecast.