Page 1 of 1

Estimating the conditional CAPM

Posted: Wed Nov 11, 2009 3:01 am
by CHRIS83
Hello,
I would like to modify the EViews sample program tv_garch.prg according to the conditional CAPM.
Does anybody know how to adjust the estimating procedure in that program?
In particular the mean equation deviates from a normal multivariate GARCH-M process as it includes conditional covariances.
The specification of the CAPM mean equation is shown in the attachment. The variance equation follows a simple BEKK-process.
Thank you in advance

CHRIS83

Re: Estimating the conditional CAPM

Posted: Thu Nov 12, 2009 2:58 am
by trubador
You can find several examples if you search the forum. For instance:
Example-1
Example-2

Re: Estimating the conditional CAPM

Posted: Mon Nov 16, 2009 7:23 am
by CHRIS83
Hello trubador,

thank you for your help.
However, I still have the problem that the GARCH-term in the mean equation has to be multiplied by an exogenous vector (the vector of portfolio weights). Is there any possibility to adjust the program to that conditional CAPM specification?

Many thanks,
CHRIS83

Re: Estimating the conditional CAPM

Posted: Tue Nov 17, 2009 6:22 pm
by trubador
Unfortunately, you cannot use matrix multiplication in LogL object. You can either write the resulting terms one by one or try to specify them in terms of series multiplication...