The RATS Software Forum

I have a three variables. The Covariance is matrix H and defined in Garch statements. I would like a equation for the mean of one of the variables (yt) in this way:

Beta1 = H(3,1) / H(1,1)
Beta2 = H(3,2) / H(2,2)

yt = b0 + d1*(Beta1*wt) + d2*(Beta2*zt) + et

b0, d1 and d2 are coefficients to be estimated, wt and zt are the other 2 variables in Multivariate Garch; et is the error. Variables wt and zt in mean could be defined in a VAR structure with l lags

How can I put this specification in a Multivariate GARCH (DCC by now)?

Regards,

Use the MVHSERIES and HADJUST options to save the covariance matrices (MVHSERIES option) and create the BETA1 and BETA2 values
(HADJUST option). Something like:

set beta1 = 0.0
set beta2 = 0.0

equation yeq y
# constant beta1 beta2
define equations for other dependent variables
group meanmodel yeq other equations
garch(model=meanmodel,mvhseries=mvh,$
hadjust=(beta1=mvh(3,1)/mvh(1,1),beta2=mvh(3,2)/mvh(2,2)),$
other options)

Thanks for your advice.