Page 1 of 1

How can I run this eqation (GARCH)

Posted: Thu Mar 05, 2009 12:28 am
by pedcheng
Hi ,

I would like to ask how can I run this equation

Rit = α + µσt2 - (φ0 + φ1σt2) Rit-1 + εt

the model use GARCH(1,1) with GED

I try to put it in the mean equation but dont know how to
deal with "φ0 + φ1σt2) Rit-1" term

Thanks in advanc

Re: How can I run this eqation (GARCH)

Posted: Thu Mar 05, 2009 9:39 am
by EViews Gareth
Could you define your terms a bit?

Are φ0 and φ1σt2 coefficients?

Is φ1σt2 a single term, or is it actually φ1*σt2

Re: How can I run this eqation (GARCH)

Posted: Thu Mar 05, 2009 9:32 pm
by pedcheng
φ0 and φ1 are coeffients but σt2 is conditional variance

so the term (φ0 + φ1σt2) Rit-1 is

(coefficient + coefficient * conditional variance) * Rit-1

I try to use ARCH-M menu when I run equation using GARCh model

but it only put the term " σt2 " in my equation, I dont know how to put this " σt2 " to multiply with Rit-1

Guess this will help you

Thanks

Re: How can I run this eqation (GARCH)

Posted: Fri Mar 06, 2009 5:10 pm
by Gene
EViews doesn't handle "non-standard" GARCH in Mean specifications.

Here is one solution:

- You can estimate your garch model without the garch in the mean part.
- Save the conditional variance.
- Then estimate you model again this time with the conditional variance saved from the previous estimation in the mean.
- If you want you can save the cond. variance again and then re-estimate the garch again. You can keep doing this until the log likelihood doesn't change much.


For another solution, you can set and estimate the model up in LogL.

Re: How can I run this eqation (GARCH)

Posted: Sat Mar 07, 2009 7:21 am
by pedcheng
thanks QMS Gareth , Gene

I will try it by the way you told me.

Re: How can I run this eqation (GARCH)

Posted: Tue Mar 10, 2009 6:20 am
by pedcheng
How can I run the model in LogL?

thanks