I am an undergraduate student at Harvard University living in Cambridge, Massachusetts. I currently
work as a research assistant for Rahul Singh.
I will start a PhD in Economics at Harvard University in Fall of 2025.
As a college student, I am pursuing an BA degree in Honors Statistics with an expected graduation
date of May 2025. From 2022-2024, I spent two years working closely with
Jesse Shapiro and Isaiah Andrews
as a research assistent.
Research interests: My undergraduate research focused on causal inference under misspecified models in Economics.
Currently my research is centered around nonparametric causal inference and identification. I hope to continue
this theme in my graduate studies.
Contact me: mosesstewart [at] college [dot] harvard [dot] edu
Work Experiences
Research Assistant for Rahul Singh at Harvard University (2024 - present)
Research Assistant for Jesse Shapiro and Isaiah Andrews at National Bureau of Economic Research (2022 - 2024)
Teaching Assistant at Harvard University (2022 - present)
Proctor at MIT MITES (2022 - present)
Working Papers
Singh, R., & Stewart, M. (2025, July). Placebo Discontinuity Design. https://arxiv.org/abs/2507.12693
@inproceedings{Singh2025-dc,
title = {Placebo Discontinuity Design},
author = {Singh, Rahul and Stewart, Moses},
url = {https://arxiv.org/abs/2507.12693},
month = jul,
year = {2025},
publisher = {arXiv},
eprint = {2507.12693},
doi = {10.48550/arXiv.2507.12693},
file = {pdd.pdf}
}
Stewart, M. (2025, March). Constructing an Instrument as a Function of Covariates. https://arxiv.org/abs/2503.10929
@inproceedings{stewart_constructing_2025,
title = {Constructing an {Instrument} as a {Function} of {Covariates}},
doi = {10.48550/arXiv.2503.10929},
url = {https://arxiv.org/abs/2503.10929},
publisher = {arXiv},
author = {Stewart, Moses},
month = mar,
file = {cifc.pdf},
year = {2025},
keywords = {Economics - Econometrics}
}
Researchers often use instrumental variables (IV) models to investigate the causal relationship between an endogenous variable and an outcome while controlling for covariates. When an exogenous variable is unavailable to serve as the instrument for an endogenous treatment, a recurring empirical practice is to construct one from a nonlinear transformation of the covariates. We investigate how reliable these estimates are under mild forms of misspecification. Our main result shows that for instruments constructed from covariates, the IV estimand can be arbitrarily biased under mild forms of misspecification, even when imposing constant linear treatment effects. We perform a semi-synthetic exercise by calibrating data to alternative models proposed in the literature and estimating the average treatment effect. Our results show that IV specifications that use instruments constructed from covariates are non-robust to nonlinearity in the true structural function.
