Old-Age Security and Gender Preference Hypotheses: A Duration Analysis of Malaysian Family Life Survery Data

Abstract

This paper demonstrates that an accelerated hazard framework is more appropriate than commonly used proportional hazard framework to model the timing of marriage and timing and spacing of children. Using the 1976 Malaysian Family Life Survey Data and selecting an appropriate duration model, the paper tests the old-age security and gender preference hypotheses and estimates the replacement effect. The paper finds strong evidence for the old-age support hypothesis: If parents have sufficient wealth to support themselves during their old-age, they have longer birth intervals; the Chinese have higher likelihood of depending on their children for old-age support than the Malays, and higher is the husband's earnings, the less likely is the couple's dependence on children for such support. The paper finds weak evidence for son preference hypothesis: The number of sons has no effect on the birth intervals of the Malays whereas it has significant negative effect up to three children for the Chinese. Regarding the replacement effect the paper finds that in response to a child death, parents like to have shorter durations up to fourth child and thereafter, the effect is not significant; the desired family size of the Chinese is higher than that of the Malays.