This paper evaluates the effect of AIDS-related mortality on per-capita incomes of surviving household members, using a large nationally representative sample of rural households from Zambia. To minimize selection bias that may arise because AIDS is likely to be the endogenous outcome of individual behavior, we employ a difference-in-difference propensity score matching estimator. We find that the death of a prime-age member has no significant impact on per-capita household income. This result continues to hold when we control for spillover effects by excluding households from the control group if members departed or joined for reasons related to AIDS. A likely explanation for this finding is that surviving household members pursue a mix of income and demographic coping strategies that prevents income losses in the short to medium run.
HIV/AIDS prime-age mortality per adult equivalent income difference-in-difference propensity score matching spillovers Zambia