Abstract:This paper addresses the critical challenge of detecting financial crises in their early stages given their profound economic and societal consequences. It investigates the efficacy of machine learning models by comparing the standard econometric model of Logistic Regression with k-nearest Neighbours, Random Forest, Extremely Randomised Trees, Support Vector Machine, and artificial Neural Network models in a cross-validation experiment. The study utilises financial crisis observations between 1870 and 2020 from the MacroHistory database and up to 14 early warning indicators from different theoretical backgrounds. The results demonstrate that advanced machine learning models, particularly Random Forest and Extremely Randomised Trees, outperform Logistic Regression, as measured by the area under the receiver operating characteristic curve (AUROC). This finding holds across different crisis data sources, early warning indicator sets, and model specifications. Furthermore, this study emphasises the importance of model explainability for both researchers and policymakers by employing Accumulated Local Effect (ALE) plots to unravel the complex predictive processes of advanced machine learning models and provide deeper insight into the underlying crisis dynamics. Nonlinear connections to financial crises were identified for several early warning indicators, which are difficult to capture by linear models. Several U-shaped relationships were observed, highlighting the disruptive role of strong economic changes for financial stability. The results emphasise the potential of advanced machine learning methods to limit the societal impact of financial crises by enabling earlier intervention by policymakers and public institutions.

Early warning models for systemic banking crises: Can political indicators improve prediction?

Does ESG Predict Systemic Banking Crises? A Computational Economics Model of Early Warning Systems with Interpretable Multi-Variable LSTM based on Mixture Attention

Do Changes in Risk Perception Predict Systemic Banking Crises?

Multinomial logit as an early warning model for predicting banking crises

A systematic review of early warning systems in finance

Can Text-Based Statistical Models Reveal Impending Banking Crises?

Risk-taking and systemic banking crisis in Africa: do regulatory policy framework provide new insight in threshold models?

Cross-Country Linkages and Spill-Overs in Early Warning Models for Financial Crises

Predicting financial crises: an evaluation of machine learning algorithms and model explainability for early warning systems

Determinants of financial crises—An early warning system based on panel logit regression

Unveiling Early Warning Signals of Systemic Risks in Banks: A Recurrence Network-Based Approach

Probabilistic multivariate early warning signals

Determinant of Sovereign Default — Does Political Factors Matter ?

Are Crises Predictable? A Review of the Early Warning Systems in Currency and Stock Markets

Crisis Early Warning Study Based on Greek Sovereign Debt Crisis——And the Quantitative Analysis of the Proportion of Social Security Expenditure

Financial measures and banking crisis: New evidence

Predicting systemic financial risk with interpretable machine learning

The Research on China Macro-financial Early-Warning System Establishment

Political risk modelling and measurement from stochastic volatility models

Assessing the systemic risk impact of bank bail-ins

The effect of political factors on sovereign default