Page 1 of 1

Convergence issue in DCC GARCH

Posted: Fri Aug 05, 2016 6:01 am
by Vik_pa
Dear Tom,
I am trying to estimate my model using bivariate BEKK GARCH and DCC GARCH. I have using weekly data for 15 years. For DCC GARCH model, I am facing the No Convergence issue when using the ar1/var model in the mean equation. With out the var model in the mean equation and with mean equation containing the constant only, it converge in 32 iterations. Also for BEKK model there is no issue in convergence with var model as mean equation.

Can you please help me in this issue.

Kind Regards,

My code for BEKK is

Code: Select all

system(model=var1)
var Crude_return Gold_return
lags 1
det constant
end(system)

garch(model=var1, p=1,q=1,mv=bekk,rseries=rs,mvhseries=hs,stdresids=z,derives=d) / crude_return gold_return
DCC GARCH Code

Code: Select all

GARCH(MODEL=VAR1,P=1,Q=1,MV=DCC,METHOD=BHHH,PMETHOD=SIMPLEX,PITERS=10) / CRUDE_RETURN GOLD_RETURN

Re: Convergence issue in DCC GARCH

Posted: Fri Aug 05, 2016 9:33 am
by TomDoan
That's probably a good indication that DCC is a bad choice.