4 variate BEKK MGARCH

[anarello]

Hello Fellows,
I am trying to estimate BEKK MGARCH (Engle and Kroner ,Multivariate simultaneous generalized ARCH.Cambridge University press,1995) model . I was told that RATS is capable to do this. Does anybody have a code for the 4 variable(4 variables would be stock indices return) MGARCH BEKK representation model,I mean I would like to see all the elements of the Matrices A and G, not only diagonal elements or a scalar.
Thank you guys

[TomDoan]

With a relatively current version of RATS (6 or later), this does a three variable BEKK. Four variables just needs another variable added to the end of the GARCH instruction. (Everything except the last line is data handling).

Code: Select all

all 6237
open data g10xrate.xls
data(format=xls,org=columns) / usxjpn usxfra usxsui
*
set xjpn = 100.0*log(usxjpn/usxjpn{1})
set xfra = 100.0*log(usxfra/usxfra{1})
set xsui = 100.0*log(usxsui/usxsui{1})
*
* Standard BEKK with GARCH instruction
*
compute gstart=3,gend=6237
garch(p=1,q=1,mv=bekk,pmethod=simplex,piters=5) gstart gend xjpn xfra xsui

[anarello]

Dear Tom,
Thank you for your promt reply.Today is the first day I was introduced to RATS.Could you please explain more in deail what should I do.You mean I should add one more variable on the very last row stating the name of the variable,and leave everything the same as in your reply to Jessie "Trivariate VAR-GARCH model"
ll 6237
open data g10xrate.xls
data(format=xls,org=columns) / usxjpn usxfra usxsui
*
set xjpn = 100.0*log(usxjpn/usxjpn{1})
set xfra = 100.0*log(usxfra/usxfra{1})
set xsui = 100.0*log(usxsui/usxsui{1})
set xdji = 100.0*log(usxdji/usxdji{1})
*
* Standard BEKK with GARCH instruction
*
compute gstart=3,gend=6237
garch(p=1,q=1,mv=bekk,pmethod=simplex,piters=5) gstart gend xjpn xfra xsui xdji

[TomDoan]

Yes. If you had the USXDJI variable from some source. (There's nothing in your modified program that defines it). You can list as many variables as you want on the GARCH instruction --- the first two parameters set the estimation range, the remainder are the dependent variables. Note that it's sometimes hard to estimate BEKK models once you get beyond five variables because of the number of free parameters and the way they interact.