Am 08.11.2012 21:14, schrieb Tomasz Bielawski: > > I have question if it is possible in gretl to prepare similar prediction > with one regressant in variance equation for normal,t and GED > distributions.
Tomasz, it may well be that I'm missing something here, but if you can estimate the variance equation, then you can use it for forecasting. If you have extra terms in there, you need to forecast them, too. But the problem with certain types of distributions would seem to apply to model specification and estimation, not to the subsequent step of using estimated equations for forecasting. Of course, if you want to have some confidence intervals, then things become more complicated, but it sounded as if you "just" wanted the point forecast (of the variance). cheers, sven
