Ownership Structure, Macroeconomic Uncertainty, and Bank Stability: Analysis Using Threshold Models and Machine Learning

Abstract

Purpose: This study investigates the impact of ownership structure on the financial stability of 28 Vietnamese commercial banks during 2010–2024, while examining the moderating role of macroeconomic uncertainty.
Design/methodology/approach: A panel threshold model is combined with causal machine learning techniques, including Causal Forest and Double Machine Learning, to identify endogenous thresholds and measure conditional effects. Macroeconomic uncertainty is quantified through the Composite Macroeconomic Uncertainty Index (CMUI), constructed via principal component analysis (PCA) from EPU, GPR, WUI, WSI, and ESGUI indicators.
Findings: Results reveal a nonlinear relationship between ownership structure and financial stability. State ownership enhances stability only at moderate levels, but excessive concentration leads to soft budget constraints. Foreign ownership exerts negative effects, with adverse impacts rising at higher shares. Dispersed private ownership undermines stability, whereas concentrated private ownership strengthens monitoring. Moreover, macroeconomic uncertainty exerts direct effects and moderates both the direction and intensity of ownership impacts.
Originality/Value: The study advances theories of agency costs, soft budget constraints, and financial fragility by providing new insights into how ownership interacts with macroeconomic uncertainty. It introduces an integrated framework combining threshold modeling and causal machine learning to assess conditional effects. Empirical evidence from Vietnam enriches the literature on emerging markets and offers policy implications for optimizing bank ownership structures under uncertain macroeconomic environments.