Working Papers

The Design of Exam-based University Admission System and the STEM Gender Gap (with YingHua He and Jin-Tan Liu) 

• Abstract: A centralized university admission system can give students discretion over which subject tests to take, and give university-major programs discretion over which subject tests to count and the corresponding weights. We study how this two-sided design shapes the STEM gender gap, leveraging economy-wide data on Taiwan's exam-based centralized university admissions. Women are substantially less likely than comparable men to take the STEM-heavy mathematics test, narrowing their feasible set of STEM programs. We estimate an equilibrium model of test-portfolio choice, score production, and university-major program choice, allowing for latent types linking the sequential decisions. We further evaluate counterfactual designs of the exam-based admission system, including resetting STEM programs' subject weights, merging the two mathematics tests, and splitting capacity by mathematics requirement.

The Global Gender Gap in STEM Applications: Pipeline vs. Choice (with Isaac Ahimbisibwe, Adam Altjmed, Georgy Artemov, Andrés Barrios-Fernández, Aspasia Bizopoulou, Martti Kaila, Jin-Tan Liu, Rigissa Megalokonomou, José Montalbán, Christopher Neilson, Sebastián Otero, and Xiaoyang Ye), R&R at Economics of Education Review

Decentralizing the Centralized University Admission Market: Evidence from Taiwan's Batched Multi-offer Deferred Acceptance Mechanism (with Jin-Tan Liu)

• Abstract: The application-based university admission channel in Taiwan operates a hybrid mechanism, combining decentralized holistic evaluation by programs with centralized assignment after students receive batched multiple offers (and waitlists). The mechansim has a unique feature: before submitting her final rank-ordered list, every applicant observes her exact position in the published ranking of each program that offers or waitlists her. Leveraging administrative records on complete program rankings and student choices, we identify and estimate preferences of both sides of the matching market via a structural model, disentangling their vertical (common) and horizontal (idiosyncratic) components. Counterfactual simulations quantify the allocative and welfare consequences of ranking-position revelation, the value of decentralized evaluation relative to exam-based assignment, and how expanding the application-based quota reshapes matching outcomes and inequality across sociodemographic groups.

Spatial Inequality and High School Track Choice: Implications from Mexico City

Selected Work in Progress

Right-Sizing an All-Charter School District: An Equilibrium Analysis of New Orleans Public Schools System (with Dong Woo Hahm, Douglas Harris, Qinyou Hu, and Shangkun Xie)

Priority Design, Immigration Policies, and Educational Track Choice in Finland (with Qinyou Hu)

Everybody's Normal: A Field Experiment to Destigmatize Mental Illness among Adolescents and Parents in China (with Qinyou Hu and Yiming Xia)