Intra-Household Gender Disparity in School Choice: Evidence from Private Schooling in India

Saturday, 4 July 2015: 8:30 AM-10:00 AM
CLM.7.03 (Clement House)
Soham Sahoo, Indian Statistical Institute, New Delhi, India
This paper explores the gender inequality within households in the decision of private versus government school choice in India. During the last decade, India has experienced a huge surge in both demand for and supply of private schools. These fee-charging private schools have been perceived by households as a better alternative to government schools where quality of education has remained a concern. In rural India, education of girls is often considered to be less worthy than boys' education. Since private schools charge fees and are more expensive than their government counterpart, households may be less likely to send girls, as compared to boys, to private schools. Using a three period longitudinal data from rural Uttar Pradesh, this paper seeks to identify the intra-household gender gap in private school choice. Since the school choice decision is observed only for children who are enrolled, this study estimates a household-fixed effects model that also takes into account any non-random enrollment decision in the analysis. The results show a significant gender gap in private school choice, which is present for both younger and older children, and is rising over time. The study explores further to find that a higher village level difference in average cost of private and government schooling is associated with a larger gender gap in private school choice, even after controlling for average school level quality. Among the cost components, the direct cost, school fees in particular, comes out to be the most significant factor to be correlated with the gender gap. Acknowledging that cost differences could be endogeneous, the analysis shows that the result is robust to village level confounding trends. The findings also remain unaltered in a separate exercise that inspects the sensitivity of the effects with respect to potential omitted variable bias.