Page 1 of 1

ARMA structure for Multivariate GARCH model

Posted: Mon Oct 04, 2010 9:20 am
by DB916
Hi all,

Im using Eviews 6 at the moment and right now i have a series of data. I want to get rid of the autocorrelation by introducing a ARMA structure model. How would I do this? The topic im doing at the moment involves the multivariate GARCH models such as the BEKK model.

Can u give me the instructions to add the ARMA structure on top of the multivariate GARCH?

However if EViews 7 can still do the job. Please give me the instructions anyways.

Thanks heaps! I've been losing a lot of sleep over this!

Re: ARMA structure for Multivariate GARCH model

Posted: Mon Oct 04, 2010 9:23 am
by EViews Gareth
You can't add MA terms, but you can add AR terms to each equation's specification in your System object. In systems, AR terms are added like so:

Code: Select all

y=c(1)*x1+c(2)*x2 +[AR(1)=c(3)]

Re: ARMA structure for Multivariate GARCH model

Posted: Mon Oct 04, 2010 9:30 am
by DB916
Thanks Gareth.

So the command is Object > New Object > System?

Can you explain to me the code that you sent to me please?

How would I say write the code to denote an AR(3) process which was required to get rid of the autocorrelation?

Does Eviews 7 permit MA structure to be introduced into the equation?

thanks heaps again!

Re: ARMA structure for Multivariate GARCH model

Posted: Mon Oct 04, 2010 10:28 am
by EViews Gareth
You're probably best off reading the manual chapter on Multivariate GARCH.