Something people might like to test: in CVS I've added Bruce Hansen's asymptotic p-values for the "qlrtest" command, and also configured it so that if the original model was estimated with the --robust flag, the sup-Wald test uses the robust variance matrix. As for the issue I mentioned a week ago (the dating of the "break"), I notice that Hansen defines it the same way gretl does (the break period being the first period of the "new regime") so I'm not going to change that at present. Allin _______________________________________________ Gretl-devel mailing list Gretl-devel(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-devel