Parents have to decide how much and how to invest in each of their children. Becker proposed that parents choose among different types of investments for each child efficiently and that they also choose investments to equalize wealth across their children. Existing empirical tests of this hypothesis using across family variations in investments suffer from unobserved family heterogeneity. Family fixed effects methods have been suggested and used, although their interpretation has not been well articulated.
The contributions of this paper are two fold. First, we provide an empirical framework to study substitution effects in parental investments in the presence of unobserved family heterogeneity, unequal parental valuations of their investments across children, and unobserved di§erences in child abilities. Second, we implement this framework using a unique data set on parental investments in education and marital transfers in rural China.
Our household fixed effect regressions show that marital transfer is negatively correlated with educational investment. If the educational investment in one son is lower than that of his brothers by 1 yuan, he will be compensated by receiving 33 cents more than his brothers in marital transfers; the corresponding result for daughters is 12 cents. Differences in marital transfers across children do not fully compensate for differences in educational investments. These results cannot be explained by unequal valuations by parents or measurement problems. Instead they suggest that there are strategic considerations within the family.
We also found evidence against equal valuations across parental investments for different children in the same family. Across daughters, parents invest more in their older daughters. This first child bias is not present among sons.