Inference on Union Bounds with Applications to DiD, RDD, Bunching,
and Structural Counterfactuals
主讲人：Xinyue Bei（Duke University）
主持老师：（北大8797威尼斯老品牌）王熙
参与老师：（北大经院）王一鸣、王法、刘蕴霆
（北大国发院）黄卓、张俊妮、孙振庭
（北大新结构）胡博
时间：2024年3月22日（周五）10:00-11:30
地点（线下）：8797威尼斯老品牌107会议室
报告摘要：
A union bound is a union of multiple bounds. Union bounds occur in a wide variety of empirical settings, from relaxations of the difference-in-differences parallel trends assumption to counterfactual analysis with partially identified structural parameters. In this paper, I provide the first general and systematic study of inference on these kinds of bounds. When the union is taken over a finite set, I propose a confidence interval based on modified conditional inference. I show that it improves upon existing methods in a large set of data generating processes. When the union is taken over an infinite set, I consider the set defined by moment inequalities, as is common in practice. I then propose a calibrated projection based inference procedure that generalizes results from the moment inequality subvector inference literature and is computationally simple. Finally, the new procedures give statistically significant results while the pre-existing alternatives do not in two empirical applications, the sensitivity analysis in Dustmann, Lindner, Schönberg, Umkehrer, and Vom Berge (2022) and the counterfactual analysis in Dickstein and Morales (2018).
主讲人简介：
Dr Xinyue Bei is an econometrician working primarily on inference in partially identified models with applications in structural models, sensitivity analysis, and counterfactual analysis. She is also interested in hypothesis tests in nonstandard situations, with applications to normal mixtures and regime switching, sample selectivity, skew normal distributions, and serial correlation.
