Income or Knowledge? An Exact Three-Way Split of the Rise in World Life Expectancy along the Preston Curve, 1970–2023, with India, China and Nigeria

Preston found that most of the rise in life expectancy between the 1930s and 1960s came from an upward shift of the curve relating life expectancy to income, not from movement along it. I revisit the question with World Bank data for up to 196 economies and an exact decomposition. Fitting the curve as life expectancy on log GDP per capita in each year, the population-weighted change in life expectancy splits into three parts that add up exactly: movement along the curve as incomes rise, the shift of the curve, and the change in the weighted mean residual. Between 1990 and 2019 world life expectancy rose by 7.80 years across 191 economies. Rising incomes account for 4.42 years, the upward shift of the curve for 5.38 years, and the residual change for minus 2.00 years. The residual term is negative because populous countries that were far above the curve in 1990 moved towards it: China's residual fell from 10.90 to 3.17 years. India's residual was 4.19 years in 1990 and 3.84 in 2019, so India stayed about four years above the life expectancy predicted for its income. Over 1970–2019 the shift share is larger (11.13 of 15.81 years). The curve is concave in log income in both 1990 and 2019, and it has flattened: the slope fell from 5.291 to 4.688 years per log point.