A1007
Title: Jackknife inference with two-way clustering
Authors: Morten Nielsen - Aarhus University (Denmark)
Matt Webb - Carleton University (Canada)
James MacKinnon - Queen\'s University (Canada) [presenting]
Abstract: For linear regression models with cross-section or panel data, it is natural to assume that the disturbances are clustered in two dimensions. However, the finite-sample properties of two-way cluster-robust tests and confidence intervals are often poor. Several ways to improve inference with two-way clustering are discussed. Two of these are existing methods for avoiding, or at least ameliorating, the problem of undefined standard errors when a cluster-robust variance matrix estimator (CRVE) is not positive definite. One is a new method that always avoids the problem. More importantly, a family of new two-way CRVEs is proposed based on the cluster jackknife, and it is proven that they yield valid inferences asymptotically. Simulations for models with two-way fixed effects suggest that, in many cases, the cluster-jackknife CRVE combined with the new method yields surprisingly accurate inferences. A simple software package is provided, twowayjack for Stata, that implements the recommended variance estimator.
