Testing for the Appropriate Level of Clustering in Linear Regression Models

TY - JOUR

T1 - Testing for the Appropriate Level of Clustering in Linear Regression Models

AU - MacKinnon, James G.

AU - Nielsen, Morten Ørregaard

AU - Webb, Matthew D.

PY - 2023/8

Y1 - 2023/8

N2 - The overwhelming majority of empirical research that uses cluster-robust inference assumes that the clustering structure is known, even though there are often several possible ways in which a dataset could be clustered. We propose two tests for the correct level of clustering in regression models. One test focuses on inference about a single coefficient, and the other on inference about two or more coefficients. We provide both asymptotic and wild bootstrap implementations. The proposed tests work for a null hypothesis of either no clustering or “fine” clustering against alternatives of “coarser” clustering. We also propose a sequential testing procedure to determine the appropriate level of clustering. Simulations suggest that the bootstrap tests perform very well under the null hypothesis and can have excellent power. An empirical example suggests that using the tests leads to sensible inferences.

AB - The overwhelming majority of empirical research that uses cluster-robust inference assumes that the clustering structure is known, even though there are often several possible ways in which a dataset could be clustered. We propose two tests for the correct level of clustering in regression models. One test focuses on inference about a single coefficient, and the other on inference about two or more coefficients. We provide both asymptotic and wild bootstrap implementations. The proposed tests work for a null hypothesis of either no clustering or “fine” clustering against alternatives of “coarser” clustering. We also propose a sequential testing procedure to determine the appropriate level of clustering. Simulations suggest that the bootstrap tests perform very well under the null hypothesis and can have excellent power. An empirical example suggests that using the tests leads to sensible inferences.

KW - CRVE

KW - Cluster-robust variance estimator

KW - Clustered data

KW - Grouped data

KW - Robust inference

KW - Wild bootstrap

KW - Wild cluster bootstrap

U2 - 10.1016/j.jeconom.2023.03.005

DO - 10.1016/j.jeconom.2023.03.005

M3 - Journal article

SN - 0304-4076

VL - 235

SP - 2027

EP - 2056

JO - Journal of Econometrics

JF - Journal of Econometrics

IS - 2

ER -