On Sun, 25 Dec 2005, sudip mukherjee wrote: Bear in mind that closed form estimators for GARCH models do not exist, so estimation must be carried out by means of numerical maximisation. For more detail, see chapter 10 of the gretl guide at http://ricardo.ecn.wfu.edu/pub//gretl/manual/PDF/gretl-guide.pdf The algorithm we use for assessing convergence in GARCH models is rather picky, I am afraid. You might want to have a look at the discussion that took place in the list about 1 year ago on a very near subject: http://ricardo.ecn.wfu.edu/pipermail/gretl-users/2004-October/000060.html Other packages sometimes print out "estimates" even when a form of convergence we deem acceptable hasn't occurred. My very personal opinion, however, is that the garch module needs to undergo serious revisions, though I also believe this is not the most urgent thing in our todo list. Riccardo `Jack' Lucchetti Dipartimento di Economia Università Politecnica delle Marche jack(a)dea.unian.it http://www.econ.univpm.it/lucchetti