Abstract
This study investigated the causes, patterns, and predictive indicators of bank failures in Nigeria from 1930 to 2023, integrating historical analysis with machine learning (ML) to develop an early warning system for banking distress. Despite successive reforms, Nigeria’s banking sector continued to experience cyclical instability, marked by recurrent failures driven by weak corporate governance, inadequate capital structures, liquidity crises, and macroeconomic shocks. The research adopted a mixed-method, longitudinal approach grounded in realism and positivism, drawing on secondary data from the Central Bank of Nigeria (CBN), the Nigerian Deposit Insurance Corporation (NDIC), and published financial statements.
Quantitative models were developed using Python and Statistical Packages for the Social Sciences (SPSS), employing Ridge Regression, Random Forest, and Logistic Regression algorithms to analyse both financial and non-financial predictors of bank failure. Key financial indicators examined included Capital Adequacy Ratio (CAR), Non-Performing Loans (NPLs), Liquidity Ratios, and profitability metrics were examined alongside governance quality, risk management, and regulatory compliance as non-financial dimensions. Model validation through K-fold cross-validation confirmed the robustness and predictive accuracy of the ML framework.
Findings revealed that while financial metrics remained strong predictors of distress, governance and risk management practices are the most decisive differentiators between successful and failed banks. The study also demonstrated that machine learning offered superior accuracy and scalability compared to traditional statistical methods, effectively identifying early warning signals that could inform regulatory oversight and proactive intervention.
The research contributed theoretically by extending the application of machine learning in financial risk prediction and practically by proposing an empirically tested early warning framework adaptable for use by regulators and stakeholders. It underscored the necessity of coupling technological innovation with ethical leadership and robust supervision to achieve a stable, trustworthy banking system. Ultimately, the study bridged historical insights and modern computational analytics to support predictive governance and strengthen financial system resilience in Nigeria.