New developments and perspectives on financial crisis indicators: How can signal processing enrich existing financial crises indicators?

This survey provides a critical overview of crisis indicators, synthesizing established contributions and identifying unresolved debates. It encompasses expert-based and model-based approaches and textual analysis methods that lie at their intersection. The survey offers a global comparison of the key aspects of the different methods, and introduces a new class of multivariate nonlinear segmentation model of indicators, grounded in signal processing. This approach, which combines historical insights with advanced modeling techniques, enhances our understanding of crises. • We propose an exhaustive survey of expert-based and model-based crises dating. • We offer a perspective on textual analysis methods dating. • We provide a synthetic comparison of the key aspects of methods for financial crisis indicators. • We introduce a novel method for crisis detection indicator based on signal processing. • The novel method detects all benchmark crises and provides a new typology of crises.