Driving SDG ‐Oriented Economic Development in Emerging Economies: Machine Learning Evidence From Energy, Human Capital, and Governance

ABSTRACT

This study examines the sustainability‐related determinants of economic development within an SDG‐oriented framework by integrating carbon productivity, energy efficiency, education, life expectancy, governance, and corruption control into a machine learning (ML) approach. Using panel data for the Next‐11 economies covering the period 1990–2023, the study evaluates the predictive performance of seven alternative models: OLS, Ridge, Lasso, Random Forest, Gradient Boosting, XGBoost, and Extra Trees, to identify the relative importance of environmental, human capital, and institutional factors in explaining GDP per capita as a proxy for sustainable development progress. The findings reveal that ensemble learning algorithms substantially outperform conventional linear models. Among all specifications, the Extra Trees model achieved the highest predictive accuracy, with a test R ² of 0.987, RMSE of 0.053, and MAPE of 0.345%, compared with an R ² of 0.782 and RMSE of 0.225 for the benchmark OLS model. SHAP analysis further indicates that energy efficiency and carbon productivity are the most influential predictors of economic development, with mean SHAP contributions of approximately 0.23 and 0.19, respectively. Human capital variables exhibit moderate but context‐dependent effects, whereas governance and corruption control display relatively lower direct contributions. These findings suggest that policies aimed at improving energy efficiency, environmental productivity, education, health, and institutional quality can support SDG‐oriented economic development across Next‐11 economies.