-
首页
-
- 首页
- WORKSHOP预告
计量、金融和大数据分析workshop:Cluster Robust Inference in Linear Regression Models with Many Covariates
发布日期：2023-03-03 12:00 来源：
主题：Cluster Robust Inference in Linear Regression Models with Many Covariates（线性回归模型中多协变量下的聚类稳健推断）
主讲人：巩爱博（北京大学经济学院助理教授）
主持老师：（北大经院）王熙
参与老师：（北大经院）王一鸣、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：2023年3月3日（周五） 10:00-11:30
地点（线下）：北京大学经济学院107会议室
报告摘要：
Researchers often include many covariates in their linear regression models to control for confounders in empirical research in economics, statistics and social sciences. It is also common practice in empirical work to use cluster-robust standard errors. In this project, we develop inference methods that are robust to the presence of many covariates and to clustering. We find that when the number of included covariates grows at the same rate as the sample size, the commonly used Liang-Zeger and HC-k cluster robust standard errors are invalid in general. We propose cluster robust standard error formulas that are robust to the inclusion of possibly many covariates from different approaches and apply the standard errors under different setups. Simulation evidence supporting our theoretical results is also provided.
主讲人简介：
巩爱博，现为北京大学经济学院助理教授，密西根大学博士，主要研究领域为计量经济学，相关研究已经发表在Journal of Economic Theory国际顶级期
国家发展研究院官方微信
Copyright© 1994-2012 北京大学 国家发展研究院 版权所有, 京ICP备05065075号-1
保留所有权利，不经允许请勿挪用
-
首页
-
计量、金融和大数据分析workshop:Cluster Robust Inference in Linear Regression Models with Many Covariates
发布日期：2023-03-03 12:00 来源：
主题：Cluster Robust Inference in Linear Regression Models with Many Covariates（线性回归模型中多协变量下的聚类稳健推断）
主讲人：巩爱博（北京大学经济学院助理教授）
主持老师：（北大经院）王熙
参与老师：（北大经院）王一鸣、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：2023年3月3日（周五） 10:00-11:30
地点（线下）：北京大学经济学院107会议室
报告摘要：
Researchers often include many covariates in their linear regression models to control for confounders in empirical research in economics, statistics and social sciences. It is also common practice in empirical work to use cluster-robust standard errors. In this project, we develop inference methods that are robust to the presence of many covariates and to clustering. We find that when the number of included covariates grows at the same rate as the sample size, the commonly used Liang-Zeger and HC-k cluster robust standard errors are invalid in general. We propose cluster robust standard error formulas that are robust to the inclusion of possibly many covariates from different approaches and apply the standard errors under different setups. Simulation evidence supporting our theoretical results is also provided.
主讲人简介：
巩爱博，现为北京大学经济学院助理教授，密西根大学博士，主要研究领域为计量经济学，相关研究已经发表在Journal of Economic Theory国际顶级期
访问学者申请
EMAIL
北京大学
ENGLISH
