集群稳健推断如何改变应用计量经济学

How Cluster-Robust InferenceIs Changing Applied Econometrics

作者：詹姆斯·麦金农（James G. MacKinnon）

In many elds of economics, and also in other disciplines, it is hard to justify theassumption that the random error terms in regression models are uncorrelated. It seemsmore plausible to assume that they are correlated within clusters, such as geographicalareas or time periods, but uncorrelated across clusters. It has therefore become verypopular to use clustered standard errors, which are robust against arbitrary patternsof within-cluster variation and covariation. Conventional methods for inference usingclustered standard errors work very well when the model is correct and the data satisfycertain conditions, but they can produce very misleading results in other cases. Thispaper discusses some of the issues that users of these methods need to be aware of.
在经济学的许多领域以及其他学科中，很难证明回归模型中的随机误差项不相关的假设是合理的。假设它们在群集内（例如地理区域或时间段）相关，但在群集之间不相关似乎更合理。因此，使用“聚类”标准错误变得非常流行，该标准错误对于聚类内变化和协变的任意模式都非常可靠。当模型正确且数据满足特定条件时，使用聚类标准误差进行推论的常规方法效果很好，但是在其他情况下它们可能会产生非常误导的结果。本文讨论了使用这些方法的用户需要注意的一些问题。