Research
Working Paper
Learning Preference from Observed Rankings (with Chen Chian Fuh and Shang En Tsai). [Paper]
Estimating consumer preferences is central to many problems in economics and marketing. This paper develops a flexible framework for learning individual preferences from partial ranking information by interpreting observed rankings as collections of pairwise comparisons with logistic choice probabilities. We model latent utility as the sum of interpretable product attributes, item fixed effects, and a low-rank user-item factor structure, enabling both interpretability and information sharing across consumers and items. We further correct for selection in which comparisons are observed: a comparison is recorded only if both items enter the consumer’s consideration set, inducing exposure bias toward frequently encountered items. We model pair observability as the product of item-level observability propensities and estimate these propensities with a logistic model for the marginal probability that an item is observable. Preference parameters are then estimated by maximizing an inverse-probability-weighted (IPW), ridge-regularized log-likelihood that reweights observed comparisons toward a target comparison population. To scale computation, we propose a stochastic gradient descent (SGD) algorithm based on inverse-probability resampling, which draws comparisons in proportion to their IPW weights. In an application to transaction data from an online wine retailer, the method improves out-of-sample recommendation performance relative to a popularity-based benchmark, with particularly strong gains in predicting purchases of previously unconsumed products.
The Proximal Surrogate Index: Long-Term Treatment Effects under Unobserved Confounding (with Ting-Chih Hung). [Paper]
We study the identification and estimation of long-term treatment effects under unobserved confounding by combining an experimental sample, where the long-term outcome is missing, with an observational sample, where the treatment assignment is unobserved. While standard surrogate index methods fail when unobserved confounders exist, we establish novel identification results by leveraging proxy variables for the unobserved confounders. We further develop multiply robust estimation and inference procedures based on these results. Applying our method to the Job Corps program, we demonstrate its ability to recover experimental benchmarks even when unobserved confounders bias standard surrogate index estimates.
Institution or Major? Understanding Student Preferences in College Admissions (with Kuan-Ming Chen, Chi-Chao Hung, Hau-Hung Yang). [Paper]
In many countries outside the United States, college admissions require students to choose both an institution and a major simultaneously. This structure forces applicants to weigh the trade-off between attending a higher-ranked institution and enrolling in their preferred field. Using data from Taiwan’s centralized college admissions system, we examine how students make these choices. Our findings indicate that students strongly prioritize institutional ranking over their field of study. After accounting for national exam scores, we observe little heterogeneity in preferences across gender and geographic regions. Additionally, we find only a weak correlation between the program choice and expected earnings, suggesting that non-pecuniary factors, such as institutional reputation, may play an important role in shaping student decisions.
On the Consistency of Bayesian Adaptive Testing under the Rasch Model (with Hau-Hung Yang and Chia-Min Wei). [Paper]
This study establishes the consistency of Bayesian adaptive testing methods under the Rasch model, addressing a gap in the literature on their large-sample guarantees. Although Bayesian approaches are recognized for their finite-sample performance and capability to circumvent issues such as the cold-start problem; however, rigorous proofs of their asymptotic properties, particularly in non-i.i.d. structures, remain lacking. We derive conditions under which the posterior distributions of latent traits converge to the true values for a sequence of given items, and demonstrate that Bayesian estimators remain robust under the mis-specification of the prior. Our analysis then extends to adaptive item selection methods in which items are chosen endogenously during the test. Additionally, we develop a Bayesian decision-theoretical framework for the item selection problem and propose a novel selection that aligns the test process with optimal estimator performance. These theoretical results provide a foundation for Bayesian methods in adaptive testing, complementing prior evidence of their finite-sample advantages.
Personalized Subsidy Rules (with Haitian Xie). [Paper] [Talk @ AMES 2022]
Subsidies are commonly used to encourage behaviors that can lead to short- or long-term benefits. Typical examples include subsidized job training programs and provisions of preventive health products, in which both behavioral responses and associated gains can exhibit heterogeneity. This study uses the marginal treatment effect (MTE) framework to study personalized assignments of subsidies based on individual characteristics. First, we derive the optimality condition for a welfare-maximizing subsidy rule by showing that the welfare can be represented as a function of the MTE. Next, we show that subsidies generally result in better welfare than directly mandating the encouraged behavior because subsidy rules implicitly target individuals through unobserved heterogeneity in the behavioral response. When there is positive selection, that is, when individuals with higher returns are more likely to select the encouraged behavior, the optimal subsidy rule achieves the first-best welfare, which is the optimal welfare if a policy-maker can observe individuals’ private information. We then provide methods to (partially) identify the optimal subsidy rule when the MTE is identified and unidentified. Particularly, positive selection allows for the point identification of the optimal subsidy rule even when the MTE curve is not. As an empirical application, we study the optimal wage subsidy using the experimental data from the Jordan New Opportunities for Women pilot study.
Publication
The Impact of Loosening Concealed Carry Laws on Firearm Demand (with Jessica Jumee Kim). Marketing Science 44(3), 2025 , 496-504. [Paper]
Since 2021, 14 states have loosened Concealed Carry Weapon (CCW) laws. This research investigates the impact of loosening CCW laws on legal firearm purchases. Specifically, we explore CCW Shall Issue adoption, which removes local authority discretion on permit issuance, and CCW Permitless Carry, the least restrictive policy. We construct both state and county-month panel datasets covering 2010 to 2017, using background checks and online firearm retail purchase data. We find that CCW Shall Issue adoption increases gun purchases, particularly new handguns. Over 70% of this increase is driven by repeat gun buyers. The increase in handguns induced by CCW Shall Issue is substantially greater in high crime and urban areas. In contrast, CCW Permitless Carry has no effect on gun purchases.
Global Representation of the Conditional LATE Model: A Separability Result (with Haitian Xie). Oxford Bulletin of Economics and Statistics 84(4), 2022, 789-798. [Published Version] [WP Version]
This paper studies the latent index representation of the conditional LATE model, making explicit the role of covariates in treatment selection. We find that if the directions of the monotonicity condition are the same across all values of the conditioning covariate, which is often assumed in the literature, then the treatment choice equation has to satisfy a separability condition between the instrument and the covariate. This global representation result establishes testable restrictions imposed on the way covariates enter the treatment choice equation. We later extend the representation theorem to incorporate multiple ordered levels of treatment.