The Dimensions of Child Replacement.

Abstract

This study is devoted to one of the central questions of economic demography--how is it that couples respond to the death of a child, where their environment is one in which the risks of child mortality are significant? The issues being considered are conveniently summarized under the heading of "child replacement" effects. The study has both theoretical and empirical components, with the empirical work being focused on mortality and fertility in postwar Malaysia. The theoretical model of child replacement effects advances a proposition not previously seen in the literature: that parents may pursue mixed strategies in their efforts to respond to a child death. They may attempt to have additional births to make up for the loss of a child; they may also attempt to devote more care to the births which follow, whether or not the number of births is increased. Results from the economic theory of rationing, duality, and maximization under nonlinear constraints are used to draw out the implications of parity and investment effects as they shape child replacement efforts. It is suggested that investment effects are perhaps most likely to be displayed in high-fertility, high-mortality settings. The empirical work, based on the 1976 Malaysian Family Life survey, centers on the implications of child deaths for subsequent parental decisions on breastfeeding, one of the most prominent in a set of important child investment variables. Logit models and a discrete-state, continuous time stochastic process model are estimated to determine whether a child death affects the probability of breastfeeding a subsequent birth or breastfeeding duration. While preliminary, the results show that child deaths do not appreciably affect the probability of breastfeeding, but that, conditional on breastfeeding occurring, they are quite important in lengthening the duration of breastfeeding; the latter finding may be taken as initial evidence for an investment effect. The three-state stochastic process model also supplies information on the effect of exogenous variables on the risks of mortality for breastfed and not-breastfed children, and provides a framework for evaluating the protective influence of breastfeeding on child health.