Predicting Long-Run Income Growth from Women’s Indicators in Developing Countries

Cross-national studies usually reduce women’s economic position to one or two composite indices, so it remains unclear which dimensions of gender inequality are most closely associated with long-run growth in low-income economies. Drawing on 128 sex-disaggregated indicators from the World Bank Gender Data Portal for 62 low- and lower-middle-income countries over 1960–2023, we combine principal component analysis and clustering with machine-learning models to identify the health, educational, and demographic correlates of the long-run trend in gross domestic product (GDP) per capita. Under repeated cross-validation, a random forest explains 38 percent of out-of-sample variation. Health indicators occupy six of the ten most influential predictors: the share of deaths among women aged 15–59 attributable to injuries dominates under two importance measures, followed by the injury share of all female deaths, deaths from communicable and maternal causes, and female primary-school enrollment. Female schooling and child survival improved fastest where they started lowest, and those countries recorded flatter GDP per capita trends. We interpret these associations, which are predictive rather than causal, through human-capital theory, and identify preventable female mortality, female schooling, and fertility-related constraints as priorities for causal evaluation rather than proven levers of growth.