Determinants of life expectancy in high longevity countries: evidence from machine learning

Health is directly aligned with Sustainable Development Goal (SDG) 3: Good Health and Well-Being, which emphasizes ensuring healthy lives and promoting well-being for all at all ages. The present study investigates the determinants of life expectancy (LEX) by incorporating a comprehensive set of factors: CO₂ emissions as an environmental factor; GDP, health expenditure, and research and development (R&D) as economic factors; education and individual internet use as social factors; and rule of law and government effectiveness as institutional factors. Using panel data for the top 20 high-life-expectancy countries covering the period 2001-2023, this study applies both traditional econometric techniques namely, PMG, fixed effects, and FMOLS estimators and advanced machine learning approaches, specifically Gradient Boosting and Random Forest. The regression results reveal that CO₂ emissions negatively affect LEX, whereas GDP, health expenditure, education, internet use, rule of law, government effectiveness, and R&D exert positive influences. The machine learning results further indicate that GDP, health expenditure, and education are the three most critical predictors of LEX in both Gradient Boosting and Random Forest models, with GDP emerging as the most dominant factor. Institutional variables such as rule of law, government effectiveness, and R&D display moderate importance, while CO₂ emissions and individual internet use consistently rank as the least influential. In terms of predictive performance, Gradient Boosting outperforms Random Forest across evaluation metrics, demonstrating lower errors and higher explanatory power. In light of these findings, this study also provides important policy implications to enhance LEX.