Copula-Based Tests For Cross-Sectional Independence In Panel Models

This paper considers tests of cross-sectional dependence using copulas in panel models. It is important to test the cross-sectional dependence in panel models because the existence of cross-sectional dependence will invalidate conventional tests such as t-tests and F-tests which use standard covariance estimators of parameter estimators. Moreover, the choice of estimation methods may depend upon whether there exists cross-sectional dependence in the errors of panel models. When the errors are cross-sectionally dependent in panel data models, for example, the computation of MLE and GMM could be rather complicated, and the feasible GLS estimator will be invalid or have to be modified substantially.