Digitalization and Bank Stability: Challenges and Opportunities for the Banking Industry Across Different Levels of Financial Development

Background and Objectives: Digitalization has fundamentally transformed the global banking sector by reshaping financial intermediation, operational efficiency, and risk management practices. The rapid expansion of digital technologies—such as mobile banking platforms, big data analytics, and automated credit scoring—has created new opportunities for financial inclusion and cost reduction. However, its implications for financial stability remain ambiguous and highly context dependent. While digitalization can improve information processing and enhance credit assessment, it may also introduce new vulnerabilities, including cybersecurity threats, model risk, and operational disruptions, particularly in data-constrained environments. This study aims to examine the complex relationship between digitalization and two critical dimensions of banking performance: credit risk, measured by non-performing loans (NPLs), and overall bank stability, proxied by the Z-score. In addition, the study investigates how these relationships vary across different levels of financial development and assesses the role of bank-specific and macroeconomic factors in shaping these outcomes. By adopting a broad cross-country perspective, the study seeks to provide policy-relevant insights into whether digital transformation strengthens or undermines banking sector resilience. Methodology: The analysis is based on an unbalanced panel dataset covering 143 economies over the period 2010–2021, allowing for comprehensive cross-country comparisons across diverse institutional and financial environments. To address potential endogeneity issues—including reverse causality between digitalization and bank stability—as well as unobserved heterogeneity, the study employs the Dynamic System Generalized Method of Moments (DGMM) estimator. This approach is well-suited for dynamic panel models with persistent dependent variables and endogenous regressors. The empirical specification incorporates lagged dependent variables to capture dynamic adjustments in bank stability and credit risk, alongside a set of control variables including bank size, profitability, capital adequacy, cost efficiency, and key macroeconomic indicators. Furthermore, the analysis conducts sub-sample estimations by grouping countries according to their level of financial development, enabling a more nuanced evaluation of heterogeneous effects. A series of robustness checks is also performed to confirm the consistency and reliability of the empirical findings. Key Findings: The results indicate that digitalization exerts a statistically significant and heterogeneous effects on bank stability and credit risk. In the full sample, digitalization is associated with a decline in bank stability (Z-score) and an increase in non-performing loans, suggesting that the costs and risks linked to digital adoption—such as high initial investment, cybersecurity exposure, and limitations of automated credit models in data-poor settings—may outweigh its benefits in the short to medium term. However, the effects vary considerably across different levels of financial development. In financially advanced economies, digitalization contributes positively to banking sector performance by reducing credit risk and enhancing stability, reflecting stronger institutional frameworks, better data infrastructure, and more effective regulatory oversight. In contrast, in less developed financial systems, digitalization improves borrower screening and loan quality to some extent, but the associated implementation costs and structural constraints tend to weaken banks’ capital buffers and overall stability. The findings also highlight the importance of control variables, with profitability consistently supporting bank stability, while bank size, cost efficiency, and capital investment exhibit varying effects depending on the financial development context. Overall, the evidence supports the view that digitalization functions as