Maternal and child mortality rates remain high in many developing countries. According to the most recent estimates, 358,000 maternal deaths occurred worldwide in 2008. 99 percent of these deaths took place in a developing country. India alone accounted for about 18 percent of all maternal deaths. Neonatal mortality also remains high, accounting for about 40 percent of all deaths in children under-5. Many of these deaths are believed to be preventable with access to better quality care during pregnancy and childbirth, and global health policies accordingly focus on promoting deliveries in health facilities as a strategy for reducing maternal and neonatal mortality in developing countries. These policies implicitly assume that there are large quality gains for women who deliver in a health facility, that translate to improvements in health outcomes. In India 55 percen of births take place at home. The expected large "returns" drives the huge amounts of expenditure annually on programs to incentivize women to deliver in a health facility. There is however surprisingly little good evidence about the "value" of giving birth in a health facility (measured in terms of reductions in mortality or morbidity). Previous research trying to measure mortality differences between women who deliver in a health facility and women who deliver at home has suffered from selection problems. Clearly, women who choose to give birth at home are different from women who choose a facility delivery. This unobserved heterogeneity represents the single biggest hurdle to obtaining reliable estimates of the returns to facility deliveries. In this study using publicly available secondary data, we adopt an innovative approach to trying to estimate the mortality impacts of giving birth in a health facility. We explot exogenous variation in rainfall patterns in India at the time of delivery and use this variation to study the impact of delivering in a health facility on neonatal and maternal mortality and morbidity. Our approach essentially compares two groups of women;one group that deliver in a health facility and the other group that deliver at home, where the only difference between the two groups is that one group was "lucky" to have had good weather at the time of delivery and so was able to get to the health facility, while the other group was "unlucky" to have bad weather at the time of delivery and so could not get to the health facility. Estimating the size of these "returns" is critical because whether or not these policies pass a cost-benefit test, and are a good investment of scarce development dollars, depends on the size of the "returns" to delivering in a health facility.
This study is important for public health because we provide evidence about the value of policies promoting hospital deliveries as a strategy to reduce maternal and infant mortality in developing countries.