Early warning or too late? A (pseudo-)real-time identification of leading indicators of financial stress

Abstract

We use logit and Markov switching models to assess, in (pseudo-)real-time, the ability of 27 indicators to predict systemic financial crises in the European Union. Before the global financial crisis (GFC), some models provided early warning signals, but it is unclear whether a specific model would have been favored over other candidate models providing contradictory evidence. Only after the GFC do debt service ratios, credit-to-GDP gaps as well as house price-to-income and house price-to-rent ratios appear as robust early warning indicators. Our results highlight that the predictive ability of indicators may change due to new risk factors or policy actions.