What drives and forecasts population longevity beyond income? Evidence from a cross-country panel

Population longevity represents a central dimension of sustainable development. Although national income and life expectancy are closely associated, recent development initiatives increasingly emphasize health-system capacity, public health interventions, environmental quality, and demographic change. The extent to which these factors are associated with longevity conditional on income, and the potential of flexible predictive models to generate forward-looking longevity scenarios, remain uncertain. This study examines life expectancy at birth as a function of six drivers, using data from 28 countries over the period 2003–2022 and employing a range of Machine Learning (ML) techniques. The results reveal nonlinear associations, including a concave relationship between income and longevity that emerges without imposing a functional form a priori. The preferred ML model is then used to project life expectancy over 2023–2032 under baseline, accelerated health-expenditure, and integrated health-system and prevention scenarios. Mean life expectancy in the sample increases from 78.10 years in 2022 to 78.21 years in 2032 under baseline trends, 78.34 years under accelerated health expenditure, and 78.98 years under the integrated health-system and prevention scenario, with larger gains concentrated in countries that have not yet closed key coverage gaps. These projections are predictive scenarios rather than causal counterfactuals, but they illustrate how ML models can support forward-looking population-health planning, provide decision-makers with interpretable estimates of alternative longevity trajectories, and offer a basis for extending the analysis to broader country coverage.