Abstract
This study examines the relationship between unemployment and life expectancy in South Africa, Nigeria, Kenya, Ethiopia, and Ghana, from 2000 to 2022. To achieve this aim, annual country-level data from the World Bank’s World Development Indicators and the WHO Global Health Expenditure Database are used. The study adds to the literature by employing a panel quantile regression framework to examine the effect of unemployment on life expectancy, along with key covariates: wage-salaried workers, external health expenditure, and GDP per capita. The study further uses three comparative estimation approaches: the fixed-effects mean benchmark, panel quantile regressions at the 25th, 50th, and 75th quantiles with country and year effects, and interaction-augmented specifications. This helps to capture moderating channels. Furthermore, clustered bootstrap inference and functional-form robustness are used to assess the stability and distributional properties of the estimates.
The estimated results indicate that unemployment has the largest adverse effect on life expectancy at the lower and median parts of the distribution. However, the effect attenuates and becomes statistically indistinguishable from zero at the upper quantile. External health expenditure and GDP per capita are positively associated with life expectancy, with stronger effects at higher quantiles. The results also show that interaction terms indicate buffering: higher formal employment share and greater health-financing intensity reduce the marginal harm of unemployment. This is particularly the case where baseline life expectancy increases are higher. Nonetheless, in lower-life expectancy contexts, unemployment shocks more readily translate into foregone care and excess mortality. Moreover, the wage-salaried employment share displays a counterintuitive negative association at the 25th and 50th quantiles. This suggests composition and measurement effects rather than a direct detrimental role of formal work. Furthermore, robustness checks with alternative specifications, lag structures, and policy-regime controls confirm the direction and distributional gradient of the main findings.
The overall distributional results suggest that the unemployment-life expectancy association is heterogeneous across the life expectancy distribution. Larger penalties are concentrated at the lower quantiles. It is concluded that health-financing intensity, macro capacity, and employment structure variables play important roles in explaining...