Assessing the nonlinear effects of climate shocks on macroeconomic stability in ASEAN : an interpretable machine learning approach for sustainable finance

Abstract

This study analyzes the effects of climate shocks (Temperature, Floods, Storms, Droughts) on the macroeconomic stability (GDP growth, inflation and industrial growth) of eight ASEAN economies for the period 2000–2024. It seeks to estimate the impact of climate shocks, uncovering nonlinear and threshold relationships, and to measure the climate and macroeconomic variables' relative importance. The methodology used is a hybrid approach that involves both panel econometric analysis (Ordinary Least Squares (OLS)) and machine learning (Random Forest (RF)), which includes both linear and nonlinear dynamics such as PDPs and SHAP-Like analysis by using data based on World Bank, EM-DAT, and Berkeley Earth data sources. According to the results, climate shocks have significant impact on macroeconomic stability, having higher impact on growth of GDP and industrial growth than on inflation. Temperature is the most important climate variable and floods, droughts and storms have negative impacts on economic activity by disrupting infrastructure and production. The impacts are also nonlinear and threshold, meaning that economic damages increase beyond a certain threshold of climate change. These impacts are muted by macro-economic variables like government consumption, exchange rate and capital formation. The study emphasizes the need for putting climate risk on the agenda of the macroeconomic policy and the contribution of sustainable finance measures to sustainable finance approaches, including green bonds and Environmental, Social and Governance (ESG) investment strategies.