I am working on an Eviews program to estimate volatility spillovers in a bivariate BEKK-GARCH framework. My univariate GARCH series are estimated using the Component GARCH (C-GARCH) model. My unrestricted bivariate BEKK is a modification of Eviews own example program of a restricted bivariate BEKK-GARCH.
My idea is that since the C-GARCH(1,1) model could be rewritten as a GARCH(2,2) model, I should add the estimated coefficients as follows to get the initialisation values for my bivariate BEKK:
'C-GARCH(1,1): GARCH = Q + alpha(RESID(-1)^2 - Q(-1)) + beta(GARCH(-1) - Q(-1)),
'Q = omega + rho(Q(-1) - omega) + phi(RESID(-1)^2 - GARCH(-1)) can be rewritten as a GARCH(2,2) process:
'GARCH(2,2) = omega + alpha(1)*RESID(-1)^2 + alpha(2)*RESID(-2)^2 + beta(1)*GARCH(-1) + beta(2)*GARCH(-2)
'C-GARCH(1,1) = GARCH(2,2) = (1-alpha-beta)(1-rho)*omega + (alpha + phi)*RESID(-1)^2 - (alpha*rho + (alpha + beta)*phi)*RESID(-2)^2 + (beta - phi)*GARCH(-1) - (beta*rho - (alpha + beta)*phi)*GARCH(-2)
'omega(1) = (1-alpha-beta)(1-rho)*omega
'alpha(1) = (alpha + phi) - (alpha*rho +(alpha + beta)*phi)
'beta(1) = (beta - phi) - (beta*rho - (alpha + beta)*phi)
Now, I am not used to programming this kind of complex models, and I just want to know if I've got this correctly. The version attached is very basic and I know that the unovariate series aren't well-fitted. I just wanted to make it as simple as possible to read. If anyone would want to give me some feedback, it would be greatly appreciated!
For questions regarding programming in the EViews programming language.
1 post • Page 1 of 1
Who is online
Users browsing this forum: Google [Bot] and 8 guests