Birth Spacing, Child Survival and Fertility Decisions: Analysis of Causal Mechanismsa

Abstract: We jointly analyze infant mortality, birth spacing, and total fertility of children in a rural area in Bangladesh, using longitudinal data from the Health and Demographic Surveillance System (HDSS) in Matlab. To distinguish causal mechanisms from unobserved heterogeneity and reverse causality, we use dynamic panel data techniques. We compare the results in a treatment area with extensive health services and a comparison area with standard health services. Simulations using the estimated models show how fertility and mortality can be reduced by, for example, breaking the causal link that leads to a short interval after a child has died. Eliminating this effect would reduce fertility and increase birth intervals, resulting in a fall in mortality by 0.14 and 2.45 per 1000 live births in treatment and comparison area, respectively. The effects of the numbers of (surviving) boys and girls on birth spacing provide evidence of son preference: having more boys has a stronger effect on the birth interval than having more girls, though both effects are significantly positive. A simulation suggests that if families would behave as if their all children were sons, fertility levels would be reduced by 3.5% and 5.7% in the ICDDR,B and comparison areas, respectively.