Shunzhuang Huang

About

I am a PhD student in Econometrics and Statistics at the University of Chicago, Booth School of Business. I am fortunately advised by Panos Toulis. I also worked/have been working with Yuehao Bai, Xinran Li and Azeem Shaikh on specific projects. I did my undergraduate at the University of Hong Kong.

My research interests lie in causal inference broadly, with a focus on randomization tests, interference and partial identification.

Contact: shunzhuang.huang [AT] chicagobooth [DOT] edu

Research

"Sharp Testable Implications of Encouragement Designs" (2025) [arxiv] (with Y. Bai and M. Tabord-Meehan)

• We characterize a set of inequalities depending only on the observed data that hold if and only if people's (unobserved) potential treatment choices are consistent with an "encouragement design", under which each value of the instrument only encourages towards at most one choice.

"Finite-Sample Valid Randomization Tests for Monotone Spillover Effects" (2025) [arxiv] (with X. Li and P. Toulis)

• We develop randomization tests that are valid regardless of sample sizes for whether the spillover effect, i.e., the effect of your "neighbors'" treatments on your outcomes, varies monotonically with the strength of neighbors' treatments.

"Inference for Treatment Effects Conditional on Generalized Principal Strata using Instrumental Variables" (2024) [arxiv] (with Y. Bai, S. Moon, A. Santos, A. M. Shaikh and E. J. Vytlacil)

• We establish identification results and inference procedures for treatment effects conditional on "generalized principal strata", a set of possible responses to treatments and instruments.

"On the Identifying Power of Monotonicity for Average Treatment Effects" (2024) [arxiv] (with Y. Bai, S. Moon, A. M. Shaikh and E. J. Vytlacil)

• We show that imposing a monotonicity-type restriction on potential treatments (and anything stonger), a generalization to the classical Imbens-Angrist monotonicity assumption, does not reduce the size of the identified sets of average treatment effects whenever the restriction is consistent with the observed data.