Page 1 of 1

mgarch code

Posted: Fri Sep 28, 2012 4:31 am
by hl545
Hi, I am a newer Eviews programming user. I want to estimate a multivariate garch in mean model but my code seems not working. The error message says "overflow". I checked my code several times but it seemd ok. Is it because of the data themselves not fitted or there are some bugs in the code.

atached is the Eviews files.
thanks,

Re: mgarch code

Posted: Fri Sep 28, 2012 7:09 am
by EViews Gareth
The starting values are bad. EViews is unable to achieve good convergence in the univariate ARCH model used for starting values. You could try using other starting values.

Re: mgarch code

Posted: Fri Sep 28, 2012 7:14 am
by hl545
Thanks very much! For the starting values, I followed the previous examples. Could you possibly suggest me some good starting values or how do I do to make some?
thanks,