ESG, Economy, and Fertility: A Machine Learning Analysis of APEC Economy

This paper investigates how environmental, social, and governance (ESG) conditions, together with economic factors, shape fertility dynamics in APEC economies. Using World Development Indicators from 1996 to 2021 and assembling more than 1,400 indicators, we predict annual changes in the crude birth rate. Models are trained on non-APEC economies and tested out of sample on APEC economies. Random Forest achieves the lowest RMSE at 0.397, and models that combine ESG and economic variables outperform those relying on economic indicators alone, with RMSE values of 0.271 and 0.298 respectively. SHapley Additive Explanations (SHAP) reveal that environmental factors are the most influential predictors of fertility in APEC, followed by governance, while social factors are smaller in magnitude but show increasing importance over time. Lag analysis indicates short-run effects for social and governance variables at one-year lags and medium- term cumulative effects for environmental variables, peaking around four years. Country-level profiles highlight clear heterogeneity: environmental drivers dominate in China and Russia, governance factors are most important in Korea and the United States, and social influences are stronger in Canada and Japan. The study provides a comprehensive, externally validated, and interpretable framework for fertility prediction, while emphasizing that the analysis remains predictive and associational rather than causal. We regard this as a first step toward more rigorous causal evaluation of the highlighted drivers.