A robust test for error cross-section correlation in panel models
A wild bootstrap test of the null hypothesis that the errors of a panel data model are not correlated over cross-section units is proposed. The new test is more generally applicable than others that use the restrictive assumptions of normality and homoskedasticity. Monte Carlo results indicate that the new test is reliable.
Year of publication: |
2010-07
|
---|---|
Authors: | Godfrey, L ; Yamagata, T |
Institutions: | Department of Economics and Related Studies, University of York |
Subject: | Cross-section correlation | Wild bootstrap | Robust test |
Saved in:
freely available
Saved in favorites
Similar items by subject
-
Testing Conditional Independence via Rosenblatt Transforms
Song, Kyungchul, (2007)
-
Specification analysis of linear quantile models
Escanciano, J.C., (2014)
-
Testing for Nonlinearity in Mean in the Presence of Heteroskedasticity
Hurn, Stan, (2004)
- More ...
Similar items by person