Impact of Piped Water Supply on Infant Mortality Rate in Brazil

Shanti Gamper-Rabindran, University of Pittsburgh
Christopher Timmins, Duke University

We examine the causal impact of piped water supply on the under one infant-mortality-rate (IMR) using a novel census dataset that divides Brazil into 3659 minimally comparable areas in 1991 and 2000. The first difference models, controlling for time invariant unobservables, indicate that the 14 unit improvement in percentage household with piped water supply between 1991 and 2000 reduced the under-1-IMR by 0.56 deaths per 1000 live-births, amounting to 3.5% of the mean decline in IMR between 1991 and 2000. The quartile regressions, with IMR for children aged 1-5, who are more resilient to water-borne diseases than children under 1, to control unobserved health inputs, indicate that the impact of piped water is more pronounced in areas with higher under-1-IMR. In 1991, a percentage point increase in piped water coverage reduced under-1-IMR at the 90th percentile by 0.25 deaths per 1000 live-births, but showed no reductions at the 10th percentile.