In most developing countries children provide some form of insurance against risks when parents are old, which, in turn, justifies parental preference to have more children. In this paper, we examine the causal effect of number of children on several measures of health status of elderly parents using newly available China Health and Retirement Survey data. Because number of children in a family is not exogenously determined, we use a natural experiment (variations in China’s one child policy) and preferences for a son to account for exogenous variation in family size. We show that both variation in the one-child policy and having a first born child who is a daughter significantly increase the family size. Overall, our results suggest that having more children has a negative effect on self-reported parental health, but generally no effect on other measures of health. We find no difference between the effect of number of children on maternal and paternal health. We find some evidence that having an adult daughter living at home, or in close geographical proximity, has a positive effect on parental health. The results also suggest that upstream financial transfers have a positive effect on parental health.
