Essays on education production in China and the U.S.

This dissertation explores the importance of three key education inputs--fellow students, teachers, and school size--in the production of education in the context of Chinese and American education system. The first essay examines the effects of roommates on students' academic performance by taking advantage of the randomness of the college roommate assignment system at a selective Chinese university for the class of 2008. I find that randomly assigned roommates' average pre-treatment academic abilities are not significant determinants of students' academic performance across a number of different specifications. However, students are likely to be influenced by roommates' decision in terms of number of course credits taking and choice of major. The second paper pinpoints the effects of teachers on students' grades at a selective Chinese university, utilizing the conditional random assignment of students into different sections of a university-wide English course. Estimated teacher fixed effects show substantial variation across teachers in terms of teaching effectiveness. However, teaching effectiveness is only weakly related to teacher characteristics such as teacher rank, education attainment, gender, and experience. This latter result should be interpreted as suggestive rather than definitive due to the small sample size of teachers and low statistical power. The third essay, coauthored with Thomas S. Dee and Brian A. Jacob, attempts to establish bounds on the causal effects of school size on measures of parental involvement and social capital by using the differences in observed traits across parents connected to smaller and larger schools as a guide to the size and direction of their potentially confounding unobserved traits. Results suggest that smaller schools are more effective at promoting both parental involvement (e.g. parent's participation in PTA activities and volunteering at schools) and social capital (e.g., knowledge of children's classmates and community identification) especially in rural setting.