An Evaluation of Early Warning Models for Systemic Banking Crises: Does Machine Learning Improve Predictions?
Johannes Beutel, Sophia List, Gregor von Schweinitz
Abstract
This paper compares the out-of-sample predictive performance of different early warning models for systemic banking crises using a sample of advanced economies covering the past 45 years. We compare a benchmark logit approach to several machine learning approaches recently proposed in the literature. We find that while machine learning methods often attain a very high in-sample fit, they are outperformed by the logit approach in recursive out-of-sample evaluations. This result is robust to the choice of performance measure, crisis definition, preference parameter, and sample length, as well as to using different sets of variables and data transformations. Thus, our paper suggests that further enhancements to machine learning early warning models are needed before they are able to offer a substantial value-added for predicting systemic banking crises. Conventional logit models appear to use the available information already fairly effciently, and would for instance have been able to predict the 2007/2008 financial crisis out-of-sample for many countries. In line with economic intuition, these models identify credit expansions, asset price booms and external imbalances as key predictors of systemic banking crises.
Read article
Reports of the European Forecasting Network (EFN)
Reports of the European Forecasting Network (EFN) The European Forecasting...
See page
Ludwig (Interview)
About the CIA and a glass of red wine ... Professor Dr Udo Ludwig on the...
See page
Loose (Interview)
When there were almost no flats in Halle yet ... Brigitte Loose about IWH's...
See page
IWH-CompNet Discussion Papers
IWH-CompNet Discussion Papers The IWH-CompNet Discussion Paper series presents research...
See page
Working Papers
Fiscal Policy and Fiscal Fragility: Empirical Evidence from the OECD ...
See page
Publications
Does Machine Learning Help us Predict Banking Crises? ...
See page
Avoiding the Fall into the Loop: Isolating the Transmission of Bank-to-Sovereign Distress in the Euro Area and its Drivers
Hannes Böhm, Stefan Eichler
Abstract
We isolate the direct bank-to-sovereign distress channel within the eurozone’s sovereign-bank-loop by exploiting the global, non-eurozone related variation in stock prices. We instrument banking sector stock returns in the eurozone with exposure-weighted stock market returns from non-eurozone countries and take further precautions to remove any eurozone crisis-related variation. We find that the transmission of instrumented bank distress, while economically relevant, is significantly smaller than the corresponding coefficient in the unadjusted OLS framework, confirming concerns on reverse causality and omitted variables in previous studies. Furthermore, we show that the spillover of bank distress is significantly stronger for countries with poorer macroeconomic performances, weaker financial sectors and financial regulation and during times of elevated political uncertainty.
Read article
27.09.2018 • 18/2018
Joint Economic Forecast Autumn 2018: Upturn Loses Momentum
Berlin, 27 September – Germany’s leading economics research institutes have downwardly revised their forecasts for 2018 and 2019. They now expect economic output to increase by 1.7 percent in 2018, and not 2.2 percent as forecast in spring. They also scaled back their 2019 forecast slightly from 2.0 to 1.9 percent. These are the results of the Joint Economic Forecast for autumn 2018 that will be presented in Berlin on Thursday.
Oliver Holtemöller
Read press release