Gender bias and intergenerational educational mobility : theory and evidence from China and India

We incorporate gender bias against girls in the family, the school and the labor market in amodel of intergenerational persistence in schooling where parents self-finance children's educationbecause of credit market imperfections. Parents may underestimate a girl's ability, expect lowerreturns, and assign lower weights to their welfare ("pure son preference"). The model delivers thewidely-used linear conditional expectation function (CEF) under constant returns and separability, but generates an irrelevance theorem: parental bias does not affect relative mobility. Withdiminishing returns and complementarity, the CEF can be concave or convex, and gender biasaffects both relative and absolute mobility. We test these predictions in India and China usingdata not subject to coresidency bias. The evidence rejects the linear CEF, both in rural and urbanIndia, in favor of a concave relation. The girls face lower mobility irrespective of location in Indiawhen born to fathers with low schooling, but the gender gap closes when the fathers are collegeeducated. In China, the CEF is convex for sons in urban areas, but linear in all other cases.The convexity for urban sons supports the complementarity hypothesis of Becker et al. (2018),and leads to gender divergence in relative mobility for the children of highly educated fathers. Inurban China, and urban and rural India, the mechanisms are underestimation of ability of girlsand unfavorable school environment. There is some evidence of pure son preference in rural India.The girls in rural China do not face bias in financial investment by parents, but they still facelower mobility when born to uneducated parents. Gender barriers in rural schools seem to be theprimary mechanism, with no convincing evidence of parental bias.