UC Berkeley

This dissertation empirically investigates the effects of large-scale education policies in developing countries. Chapter 1 examines how a government-endorsed binary college label—the World-Class University (WCU) label—shapes students’ college choices in China. As part of a higher education development initiative, the 2017 reform added 25 institutions to the original list of 115 labeled universities without significantly increasing fiscal investment. These newly labeled universities experienced a 0.07 standard deviation increase in the average college entrance exam (CEE) scores of admitted students relative to matched non-labeled institutions. Using administrative college application data, we estimate a random utility model and find that students are willing to travel an additional 294 km to attend early-labeled WCUs and 57 km for newly labeled ones, relative to a baseline of 910 km. The WCU label emerges as one of the most important factors influencing students’ college choices. Chapter 2 builds on these findings by quantifying the value students place on labeled universities and disentangling the perceived monetary return (labor market premium) from the non-monetary value (willingness to pay). An information experiment reveals that students perceive an 11% earnings premium from attending a World-Class University (WCU). In addition, a stated preference survey using hypothetical choice scenarios shows that students are willing to forgo 13% of their expected earnings to attend a WCU, underscoring the substantial non-monetary value associated with the label. Taken together, these findings suggest that the combined perceived value of the WCU label amounts to 24% of expected earnings. Chapter 3 investigates the impact of parental education on the "missing women" phenomenon in the second generation, characterized by a significant gender imbalance favoring males. Using Indonesia’s 1970s INPRES school construction program as a natural experiment, it examines whether expanding educational opportunities can reduce gender bias. Leveraging a difference-in-difference identification strategy with data from the 2010 Indonesian Census, I find that in ethnic groups where only sons are entitled to inherit, the program reduced the gender ratio at birth from 113 boys per 100 girls to 111, and increased the likelihood of having at least one daughter by 9 percentage points. In contrast, for ethnic groups that traditionally allow daughters to inherit, I observe a balanced baseline gender ratio and no effect of parental education on fertility outcomes. The program closes half of the gender ratio gap between inheritance-biased and inheritance-neutral groups. These findings underscore the potential of education to shift cultural preferences and improve gender equity, particularly within patrilineal societies.