The impact of expected child survival on husbands' and wives' desired fertility in Malaysia: A log-linear probability model

Abstract

The Malaysian Family Life Survey (MFLS) conducted in 1976-77 provides an opportunity to explore quantitatively the relationship between expected child survival on husbands and wives desired fertility. A valuable feature of the MFLS is that several questions were addressed to both husbands and wives. This study incorporates both sets of answers into the analysis. The hoarding hypothesis suggests that parents who expect a small fraction of children to survive would be more likely to desire additional births than their counterparts with more optimistic expectations. The results presented were obtained using the log linear probability model. In view of the fact that the desire to have at least 1 child is almost universal in Malaysia couples with no children at time of the interview were excluded from the analysis. The figures under in the table measure the impact of the various exogenous variables on the probability that H equals 1 i.e. that the husband wants to have additional children. If an exogenous variable say x has a positive sign in the H column this means that other factors held constant a rise in x increases the odds that H will equal 1. The figures under are the corresponding estimates for wives. The interaction term captures the influence of husbands desired fertility on the wifes and vice versa. Aside from the child mortality variables several other covariates are included in the analysis as controls. As expected the interaction term was positive and highly significant; the fertility desires of husbands and wives were positively associated. Other factors held constant if the husband wanted more children this makes it more likely that the wife will want more children and vice versa. The constant term was significantly larger for husbands than wives. Ceteris paribus husbands had preferences for larger family sizes than wives. Son preference was confirmed only for fathers and did not seem to be prevalent among women. Results showed an insignificant impact of female education on desired future fertility. For male education there was both a negative and a postive sign; the former for wives and the latter for husbands. The 2 coefficients associated with death were positive as expected but significance was low. The t ratio was 0.14 in the H column and 1.2 in the W column. Holding constant actual child deaths in the family and the control variables fathers perceptions of child survival probabilities affected desired fertility but mothers perceptions did not. The husbands perceptions had no influence on W; the wifes perceptions had no influence on H.