Is Gender Destiny? Gender Bias and Intergenerational Educational Mobility in India

Abstract

Many recent studies provide evidence of gender bias against girls in India, for example, in health, education expenditure, breast feeding, and sex selection. In contrast, the gender gap in schooling has narrowed substantially over the decades. Does gender convergence in schooling attainment imply that the girls in the younger generation in India enjoy equal educational opportunities as the boys? To analyze this question, we study intergenerational schooling persistence addressing both empirical and theoretical challenges. We incorporate gender bias against girls in the family, school and labor market in a Becker-Tomes model and derive mobility and investment equations that can be taken to data. Parents may underestimate a girl's ability, expect lower returns, and assign lower welfare weights (“pure son preference”). The model delivers the widely used linear conditional expectation function (CEF) for mobility under constant returns but generates strong predictions: parental bias cannot cause gender gap in relative mobility. With diminishing returns, the CEF is concave, and parental bias affects both relative and absolute mobility. Since coresidency causes severe underestimation of the gender gap, we use data from India Human Development Survey that includes nonresident children and parents. Evidence rejects the linear mobility CEF in favor of a concave relation (both rural and urban). The daughters of uneducated fathers face lower relative and absolute mobility irrespective of rural/urban location. We find gender equality in absolute mobility for the children of college educated fathers in urban areas, but not in villages. Theoretical insights help understand the mechanisms, suggesting underestimation of academic ability and unfavorable school environment for girls. Rural parents exhibit pure son preference. Differences in the incidence of unwanted girls and the impact of parental nonfinancial inputs explain the rural-urban differences. The standard linear model misses important heterogeneity and yields misleading conclusions such as no son preference in rural India.