Education is an important factor in reducing poverty, improving child health, and empowering women, which are key indicators of economic development and are supported by the United Nations Millennium Development Goals. While education is positively associated with these development indicators, identifying the extent that education has a causal impact in developing countries requires exogenous variation in education. Two dissertation chapters rely on a change in compulsory schooling law (CSL) in Turkey as a natural experiment to identify the causal impact of education.

In 1997, the Turkish government extended compulsory schooling from five to eight years, and in addition created new schools and classrooms, hired additional teachers, and provided transportation to rural areas. As a result, enrollment rates and education attainment significantly increased, especially for females. I use the change in the CSL as a natural experiment, using variation in the exposure to the CSL across cohorts as instruments to education.

In the second chapter, I find that mother's primary school completion improves infant health, as measured by very low birth weight, and child health, as measured by height-for-age and weight-for-age z-scores, even after controlling for many potential confounding factors. Exploring heterogeneous effects indicates that maternal education improves child health more in provinces with lower income and urbanization rates. I also demonstrate that primary school completion leads to earlier preventive care initiation and reduces smoking. Using parameters from various studies, I find that the benefits outweigh the cost of the CSL.

The third chapter demonstrates that an extra year of female schooling reduces teenage fertility by 0.03 births, which is a reduction of 33%. Exploring heterogeneous effects indicates that female education reduces teenage fertility more in provinces with higher initial fertility and lower population density. Finally, the CSL postpones childbearing by delaying marriage, thereby reducing fertility.

The fourth chapter explores the effect of teenage childbearing on health outcomes and behaviors of mothers using a nationally representative sample of twins from the Midlife Development in the US dataset. I employ within-family estimations using samples of siblings, twin pairs, and identical twin pairs, to overcome the bias generated by unobserved family background and genetic traits. The results suggest that teenage childbearing does not affect long-term health; however, it adversely affects exercise and preventive care behaviors. Further, I find that the effects of teenage childbearing may operate through education and the quality of the spouse.