Working Paper: NBER ID: w13683
Authors: Abhijit V. Banerjee; Esther Duflo
Abstract: This paper uses household survey data form several developing countries to investigate whether the poor (defined as those living under $1 or $2 dollars a day at PPP) and the non poor have different mortality rates in old age. We construct a proxy measure of longevity, which is the probability that an adult's mother and father are alive. The non-poor's mothers are more likely to be alive than the poor's mothers. Using panel data set for Indonesia and Vietnam, we also find that older adults are significantly more likely to have died five years later if they are poor. The direction of causality is unclear: the poor may be poor because they are sick (and thus more likely to die), or they could die because they are poor.
Keywords: poverty; mortality; health; developing countries; household survey data
JEL Codes: I12; I32; O12; O15
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| poor health (I14) | poverty (I32) |
| poverty (I32) | mortality (I12) |
| lower levels of household consumption per capita (D12) | probability of an individual's mother being alive (J19) |
| extremely poor (I32) | mortality (I12) |