Speaker: 王宇龙 (Yulong Wang)

Title: Non-Robustness of the Cluster-Robust Inference  

Schedule: June 27, Tu 1:30-3:30 PM

Location: 诚明楼 (Chengming Hall）RM 315

 

Introduction: 王宇龙，普林斯顿大学经济学博士，雪城大学麦克斯韦公共政策学院经济系副教授。主要研究领域为计量经济学理论，在国际计量顶刊 Journal of Econometrics、Journal of Business & Economic Statistics、Journal of the American Statistical Association、Econometric Reviews、Econometric Theory、Journal of Applied Econometrics 发表多篇学术论文。

 

Abstract: Conventional cluster-robust (CR) standard errors may not be robust when data contain a few very large clusters. For instance, when using the 51 U.S. states as clusters, the largest cluster (California) comprising around 10% of the total sample violates the assumptions necessary for reliable CR inference. We formally show that the conventional CR methods can fail when the distribution of cluster sizes follows a power law with an exponent of less than two. The implications of our findings are signicant, suggesting the need to reexamine numerous empirical results reported in the economics literature, including examples from recent top journal publications.