Women's Schooling, the Selectivity of Fertility, and Child Mortality in Sub-Saharan Africa Book

This paper estimates the determinants of child mortality in the fourteen sub-Saharan countries for which Demographic and Health Surveys (DHS) data are available. It differs radically from the usual approach of estimating reduced-form equations of child mortality from samples of children by allowing for the possibility that such samples are choice-based, reflecting prior selective fertility decisions. If parents care about the health outcomes of potential births, then any unobserved factors (heterogeneity) that affect those outcomes will influence fertility decisions. Changes in women's schooling thus affect the survival outcomes of those born by: 1) altering the population of women, classified by inherent healthiness, who bear a child in any time period; and 2) directly altering the survival probabilities of those selected to be born. Inattention to the first effect was shown to result in underestimation of the effect of women's schooling in reducing child mortality in eleven of the fourteen countries studied. Methods that disregard the potentially selective effects of fertility underestimate the importance of women's schooling by a factor of 3 in the case of Tanzania and by a factor of 2 in the case of Nigeria. The issue of identification of the empirical model complicates estimation that incorporates selective fertility. The discrete (binary) nature of mortality makes estimation particularly difficult. With these data, choice of a parametric distribution for the errors (normality) was insufficient to identify the determinants of mortality from the determinants of fertility. Identification was achieved by assuming that first births are exogenous - in particular, that the lack offertility in all periods prior to the first birth was not selective. The intuition is that in a high-fertility environment such as sub-Saharan Africa, where almost all fecund women bear at least one child during their reproductive lives, a sample of one child per woman reflects the distribution of health heterogeneity of the full (uncensored) population of women. Selective fertility choice begins subsequent to the first birth. Although somewhat restrictive, this identification is less onerous than assuming all choices in a woman's reproductive life are nonselective.