Abstract
If fertility reflects the choice of households, results of their choice (duration between successive births and health of the children) cannot be considered to be determined randomly. Most existing studies of child health, however, tend to overlook the effects of fertility selection on child health. This paper argues that not accounting for this selection issue yields biased estimates and it is difficult a priori to predict the direction of this bias. We find that the estimates of birth spacing on child mortality are different when we do not account for fertility selection. Additionally, the correlated hazard estimates that we present here better fit our samples than the corresponding bivariate probit estimates used in the literature. A comparison of the fertility behaviour of households in the Indian and the Pakistani Punjab highlights the differential nature of institutions on demographic transition in these neighbouring regions.