While estimating a pooled panel model through the ols command I realized that the F
statistics is different if the --cluster option is used although the reported degrees of
freedom remain unchanged (as it should be).
Besides, does the --cluster option overrides the selection of HAC?
Show replies by date