Time Is of The Essence: Experimental Evidence on The Effects of Job Application Timing
Abstract: Leveraging a nationwide correspondence experiment that sent over 83,000 fictitious applications to large US firms, I provide novel evidence on the causal effects of job application timing. Specifically, I uncover a significant early application advantage: the early half of applications submitted to a job in the experiment are 1.2 percentage points more likely to be contacted, an effect size equivalent to 57% of the Black-White contact gap observed in the same context. Furthermore, each day earlier an application is submitted increases the likelihood of contact by 0.4 percentage points. This advantage is driven by recruiters' evaluate-while-searching strategy, wherein early applicants are contacted and evaluated while later applications continue to be accepted. By constructing the contact history prior to an application, I uncover two competing mechanisms, Quota and Learning, and demonstrate how they jointly shape recruiters' decision-making dynamics. Together, these findings add a new dimension to our understanding of labor market discrimination, highlighting the importance of considering the timing aspect in further research and policy discussions. draft | slides | poster
presented at: All-California Labor Economics Conference 2024 (poster), PolMeth Europe 2025, WEAI 100th Annual Conference, 3rd Annual AEA CSQIEP Mentoring Conference, 6th World Labor Conference, 20th EGSC at WashU St. Louis, Oligo Workshop 2025 (scheduled, poster)
Judicial Learning from Feedback: Evidence from U.S. Asylum Courts
Abstract: We study how immigration judges learn from appeal feedback in a setting with substantial discretion and noisy feedback. Using comprehensive EOIR and BIA administrative data, we construct high-dimensional prediction models to proxy judges' expected reversal probabilities and identify surprise reversals where the appeal outcome sharply deviates from judicial expectations. Event-study estimates show that surprise reversals increase grant rates by 1.57 percentage points in the following week, with no evidence of reduced procedural diligence. An attention-based learning model rationalizes these dynamics, predicting that only attentive judges update strongly in response to unexpected feedback. Reduced-form and Marginal Treatment Effect (MTE) analyses confirm this heterogeneity: attentive judges converge to appellate standards, while inattentive judges remain threshold-misaligned. Learning operates through both appeal filing and appeal success, and is stronger in high-salience defensive cases. Our findings highlight attention as a key channel through which appeal courts effectively discipline lower-court decision-making. draft | slides | poster
All the Things We Could Have Been: Labor Market Outcomes of Not Getting into the Ideal Major
Abstract: Education and major selection have an enduring impact on labor market outcomes. Leveraging the quasi-experimental setting of the College Entrance Exam in China and a mix of administrative, survey, and online labor market data, I provide causal evidence on the impact of being marginally excluded from one's preferred major on long-term career development for elite college students. Those narrowly missing their ideal majors invest more in career-focused skills and industry reputation both before and after graduation. Over time, they self-select into industries aligned with their preferred majors but face significantly higher job instability and slightly lower wages. A policy reform in the major application process supports these findings, linking them to over-inference and the frustration of narrowly missing their ideal majors. draft | slides | poster
presented at: All-California Labor Economics Conference 2023 (poster), USC Economics PhD Alumni Conference (poster).
More exicting projects to be listed :)