The RATS Software Forum

dear Tom
when i do garch estimation
Is QMLE estimator not efficiency when the the distribution is not normal ?
that means i should use the "robust" option?
thank you very much

If the residuals aren't Normal, but you use the standard GARCH, then, yes, you are getting a QMLE, and an asymptotically valid covariance matrix for the coefficients is computed with the ROBUST option.