Financial Stability Index for SADC Banks: Estimation, Property, and Inference

Monitoring the financial stability of commercial banks is crucial for assessing their long‐term viability. Traditional stability indexes are typically constructed by identifying key financial and macroeconomic determinants and applying regression‐based methods. However, these approaches lack flexibility, as they require recalibrating stability measures when market conditions evolve. Additionally, existing methods struggle to capture complex interdependencies between bank‐specific financial indicators and external macroeconomic factors. To address these challenges, this study proposes a novel framework for constructing financial stability indexes using a Kullback–Leibler divergence (KLD)–inspired dynamic model. Additionally, the study derives theoretical properties of the index, including sensitivity measures that quantify how stability responds to variations in financial and macroeconomic indicators. This flexible approach allows for incorporating diverse financial stability determinants without redefining existing stability proxies while also capturing entity‐level characteristics and pairwise dependencies through multivariate dynamic models and Frank copulas. The paper evaluates the framework using data from five Southern African Development Community (SADC) countries and compares three alternative modeling strategies: a mixture copula‐based framework, a hybrid framework integrating copulas with autoregressive distributed lag (ARDL) models, and an extended hybrid framework incorporating both ARDL and a neural network. The results demonstrate that the hybrid KL‐ARDL and KL‐NN models effectively differentiate between performance of banking industries in weak and stable economies, with the KL‐ARDL model performing best.