What question did this study set out to answer?

The study aims to explore how global economic policy uncertainty affects banking systems and the role of institutional and financial buffers in enhancing stability.

May 26, 2026Discover Sustainability0 citationsOpen Access

Integrating structural and machine learning approaches to assess bank stability through the enabling buffer channel

TPThuy Tu PhamHo Chi Minh University of Banking

Key Points

The study aims to explore how global economic policy uncertainty affects banking systems and the role of institutional and financial buffers in enhancing stability.
Two-stage structural–machine learning framework combining Panel VAR and Double Machine Learning
In the first stage, XGBoost counterfactual simulations estimate the impact of shocks on non-performing loans and return on assets
In the second stage, Causal Forests measure how institutional quality and financial metrics influence shock absorption
GEPU shocks negatively impact banking performance, particularly in low- and middle-income economies
Both institutional and financial buffers help mitigate these adverse effects; capitalization and liquidity are significant only in strong institutional settings
Findings support the EBC hypothesis: effective financial buffers depend on the strength of institutional conditions

Abstract

This study examines how global economic policy uncertainty (GEPU) transmits through banking systems and how institutional and financial buffers jointly determine their resilience. It introduces the “Enabling Buffer Channel” (EBC) framework, which highlights the complementarity between institutional quality and prudential buffers in shaping systemic stability. By improving understanding of how institutions and capital buffers interact to sustain financial resilience, the study contributes to the United Nations Sustainable Development Goals (SDGs), particularly SDG 8 (Decent Work and Economic Growth), SDG 9 (Industry, Innovation and Infrastructure), and SDG 16 (Peace, Justice and Strong Institutions), thereby enhancing sustainable financial systems. A two-stage structural–machine learning framework integrating Panel VAR (PVAR) and Double Machine Learning (DML) is employed on a dataset of 91 countries from 2008 to 2023. In the first stage, XGBoost-based counterfactual simulations estimate the causal “damage” of GEPU shocks on non-performing loans (NPL) and return on assets (ROA). In the second stage, Causal Forests within the DML framework quantify heterogeneity in shock absorption as a function of institutional quality (EQ), capitalization (CAP), and liquidity (LIQ). Results reveal that GEPU shocks adversely affect banking performance, with effects concentrated in low- and middle-income economies. Both institutional and financial buffers mitigate these effects, but the stabilizing role of CAP and LIQ becomes statistically significant only under strong institutional conditions, providing empirical evidence consistent with the EBC hypothesis. Institutional strength is associated with stronger effectiveness of financial buffers in stabilizing banking outcomes, thereby fostering more resilient and sustainable financial systems aligned with the SDGs. The findings underscore that prudential policies are not universally effective; their success depends on the institutional environment. Context-sensitive regulatory frameworks are essential for enhancing resilience under global uncertainty. This study is among the first to integrate structural modeling and causal machine learning to empirically identify institutional–financial complementarity, advancing a new paradigm for understanding global banking stability.

问 AI

Bookmark

View Full Paper