Mortality and fertility: how large is the direct child replacement effect in China?

"The direct child replacement effect, that is the extent to which parents will have a child to replace one that died, was estime d by Olsen's method in China, using microdata from a sample survey of 4785 ever-married women in Hebei province taken in 1985. Olsen's method corrects for bias in the ordinary least squares estimate and uses mortality rates as an instrumental variable. Data for all age groups were pooled. Estimates of the direct replacement effect before and after correction for bias were 0.836 and 0.556, showing that over half of child deaths were replaced. The instrumental variable estimate produced a result of 0.614, similar to 0.613, computed by considering p1 to be random. A computation without instrumental variables yielded 0.774. The results showed that this method for calculating the direct replacement effect works well with these data because several of the alternative methods gave close estimates. There appears to be little family specific variation in child mortality rates, pointing out the effectiveness of Chinese primary health care. The direct replacement effect of about 0.6 is, however about 3 times that found by Olsen in Colombia and Malaysia, and close to the theoretical maximum. The reason for such a high value may be the small family size in China, where death of a child is more likely to provoke a response. Or it might suggest that severe government policy causes couples to attempt to have another child. Policy implications in this setting are that rather than make more contraceptives available, or improve infant survival, Chinese policymakers should make a small family size attractive, e.g., by improving education or pensions.