The effect of female education on health in Bangladesh

Female education is believed to affect health through its influence on health behaviors. However, the effect of female education may be estimated incorrectly when female education is correlated with unobserved variables at the community and individual levels that also influence health. This dissertation estimates the causal effect of female education on health in Bangladesh and addresses the potential sources of endogeneity. I apply instrumental variables (IV) constructed from education programs introduced nationwide in the 1990s in Bangladesh to analyze integrated data from the 2007 Demographic and Health Survey, the 1981 population census, and the secondary school census in 2006. In the first paper, I assess the causal effect of female education on adolescent reproductive health outcomes in order to understand the mechanisms through which education influences adolescent fertility. I find that a one-year increase in the highest grade achieved reduced significantly the probability of first marriage by age 15 by .050, the probability of first live birth by age 16 by .013, and the number of live births by age 20 by .072 births. In the second paper, I examine the causal effect of maternal education on sex bias in child survival between the first and fifths birthdays in Bangladesh in order to assess the influence of maternal education on parental son preference and differential parental behaviors by gender of child. I find that a one-year increase in highest grade achieved increased significantly the survival probability for both boys and girls by .012. However, there was no incremental effect of maternal education by gender of child, implying that girls do not benefit any more than boys from educated mothers. The difference between the IV and ordinary least squares (OLS) estimates of the effect of female education on health vary depending on the health outcomes. The finding suggests that the direction and magnitude of bias due to endogeneity are not universal across health outcomes and cannot be determined as a priori knowledge. Lastly, the reduced form results of the two papers suggest that the education programs significantly enhanced female education, and reproductive and child health