Work

Essays in the Economics of Education

Year: 2025
Type: dissertation
Abstract: This dissertation examines the effects of expanding public school choice, regulating class sizes, and altering student beliefs, by combining large-scale administrative data with quasi-experimental and field-experimental methods. Chapter 1 examines the expansion of Early French Immersion programs in British Columbia. My analysis takes advantage of rich administrative data covering all public and private students in the province. Using the distance to new programs as a source of variation in exposure, I implement a heterogeneity-robust difference-in-differences estimator that leverages the quasi-random timing of program introductions. The introduction of French Immersion programs draws students from regular public schools and, to a lesser extent, private schools, with shifts concentrated among higher-income and English-speaking families. Affected areas see gains in Grade 7 test scores and higher post-secondary attendance at age 19. Chapter 2 studies how schools respond to class size regulations using two policy changes in British Columbia as natural experiments: the removal of average class size limits in 2012 and the introduction of stricter caps in 2017. I show that schools often form mixed-grade classes rather than hire more teachers when enrollment slightly exceeds class size caps. These responses align with cost-minimizing incentives across different policy regimes. I also find evidence of enrollment bunching just below thresholds, indicating strategic behavior. The results show how class size rules affect school organization and classroom composition. Chapter 3 uses a field experiment to show that misperceptions about the noisiness of performance signals hinder accurate belief updating. In a large first-year calculus course, students consistently overestimate test randomness and exhibit overconfidence. A randomized intervention correcting these beliefs improves grade predictions, especially among students who adjust their beliefs accordingly. The treatment reduces up to one-third of the gap between students' errors and a Bayesian benchmark, highlighting that inaccurate beliefs about the noisiness of signals are a key but addressable barrier to accurate updating.
Language: English
Subfield: Education
Open Access status: green
OpenAlex ID: W7139847805
Cites: 109
Cited by:
Related to: 10