Sub-Saharan Africa continues to face the world’s highest under-five mortality, with one in fifteen children dying before their fifth birthday. Because half of these deaths occur within the first month of life, encouraging mothers to give birth in healthcare facilities rather than at home has been a central policy focus in recent decades as a means to reduce maternal and neonatal mortality.

Yet, despite major gains in access to facility birth, sub-Saharan African countries remain far from closing the mortality gap with higher income countries. One important reason is that quality of care remains strikingly low, especially at the last mile where vulnerability is highest. If the most vulnerable require more specialized interventions, expanding access alone may fail to deliver substantial health gains. Moreover, a poor experience with low-quality care can discourage future use of healthcare services.

This paper addresses two research questions:

1. Does access to facility birth reduce neonatal mortality, and how do effects vary jointly with facility quality and baseline risk at the last mile?

2. How does an exogenous encounter with formal healthcare affect mothers’ future use of health services?

I study these questions in rural Malawi. I combine georeferenced birth histories from the Demographic and Health Survey with a census of health facilities with detailed information on staffing, services, and supplies.

A concern is that this shock could also capture how rainfall affects agriculture and health differently in more remote areas, so I control for monthly rainfall in the twelve months preceding birth and its interaction with distance, which absorbs these slower moving effects and leaves the short term rainfall around birth interacted with distance as the variation that identifies the causal effect of access. 

I find three main results.

1. Access to facility birth substantially lowers neonatal mortality for high-risk births when the nearest facility has a doctor, but has no benefit for low-risk births or where facilities lack doctors. The takeaway is that the share of facility births, despite being the metric used in international policy guidelines, is not a sufficient statistic for progress in health, as its effectiveness can be very heterogeneous.

2. Greater access to facility birth reduces the likelihood that mothers later visit a healthcare facility when their nearest facility has low resources, while access to better equipped facilities has no significant effect. Exposure to low-quality care can therefore discourage future engagement with the health system.

3. Delivery choices display persistence. An increase in access to facility birth raises the probability that a mother gives birth in a facility for her next child, regardless of facility quality, while mothers who are prevented from reaching a facility because of the rainfall shock become more likely to deliver at home in subsequent births. This highlights the importance of increasing access as it can help break the cycle of persistent home births