"Wavelet-based" Early Warning Signals of Financial Stress: an Application to IMF's AE-FSI

The objective of this paper is to propose a new methodology for the construction of early warning composite indicators that is able to exploit all available information conveyed by individual indicators taking advantage of the multiresolution decomposition properties of wavelet analysis. In this paper we construct a “wavelet-based” early 35 warning indicator of financial stress for the IMF financial stress index for advanced economies developed in Cardarelli et al.(2009).Specifically, for each country of the sample we construct a “wavelet-based” early warning indicator of financial stress by aggregating several “scale-based” sub-indices whose components are selected on the basis of their statistical performance at each frequency band. The empirical evidence suggests that: (1) the “general fit” of the early warning indicators in relation to the financial stress index is rather good, with an average lead of about 2 months; and (2) the “wavelet-based” composite indicator largely outperforms any individual financial variable taken in isolation in predicting financial stress at every horizon, with the gain increasing as the time horizon increases.