An early warning system for predicting systemic banking crises in the Eurozone: A logit regression approach

Eurosystem macroprudential policies require shared action between national authorities and the European Central Bank (ECB). This has created the need for a common basis for macroprudential analysis and as result the Macroprudential Database (MPDB) was created by the ECB and the European Systemic Risk Board (ESRB) in 2015 in order to support the central bank's functions and ESRB's needs. This paper examines a multivariate binary logit Early Warning Model (EWM) for systemic banking crises with the aim to evaluate the predictive validity of the risk indicators included in the MPDB, as well as further variables not employed in previous relevant studies. The main finding is that most of the risk indicators employed from MPDB are important for forecasting from 4 to 1 years before the onset of a systemic banking crisis. Specific banking variables that capture industry concentration, assets, funding, and liquidity, are more important, on average, than macroeconomic variables. Important financial stress indicators such as CLIFS and SovCISS and economic expectations are also significant. The model is robust to various specifications and has a better performance when post-crisis observations are not included.