ESSAYS IN THE ECONOMICS OF EDUCATION

This dissertation comprises three essays in the economics of education. I use natural and field experiments to evaluate education policies and programs. I also bring insights from psychology to understand how to improve learning and work efficiency. The first chapter shows how a Chinese city was successful in helping its low-performing schools catch up. The city's education bureau identified several low-performing middle schools and guaranteed elite high school admission to their top ten-percent graduates. I document that schools affected by this top-ten percent policy improved their performance by 0.3 standard deviation. To understand the underlying mechanisms, the city's lottery system for middle school assignment is used to test for changes in composition and value-added. The study suggests that incentives for better students to attend lower-performing schools help narrow not only the school performance gap but also the school quality gap. The second chapter evaluates a peer mentoring program that matches high-performing students as mentors to their low-performing classmates and provides non-monetary incentives for them to study together and stay in school. We implemented the program in two rural Chinese middle schools. The program did not improve the mentees' math scores, but instead increased their learning stress. However, the program did significantly improve the mentors' math scores by 0.57 standard deviations and lowered their dropout rate by 3%, with no impact on their mental health scores. We discuss possible reasons for these surprising results and propose changes in program design that may help mentees benefit as well. The third chapter studies the effect of time abundance on work efficiency. I propose a strategic framework of efficient completion of time-constrained tasks. Facing a task with a deadline, an agent is under-motivated when there is abundant time and over-motivated when the deadline is too close. This generates a hump-shaped relationship between efficiency and time available for the task. I use online homework tracking data for a large introductory microeconomics class to test this theory. Within-subject analysis provides evidence supporting the predictions: when a student starts work neither too early nor too late, he/she has a higher class ranking and a lower time cost