Page 1 of 1
Problems with estimating Realized GARCH
Posted: Tue Jun 28, 2011 1:52 am
by evaeva
Hi,
I am trying to estimate Realized GARCH by Peter Hansen (2011), but have some problems with it - please see the attachment with description. I am using Eviews 6.
I would be very grateful if you could help me. Thank you very much in advance!
Eva
Re: Problems with estimating Realized GARCH
Posted: Tue Jun 28, 2011 3:50 am
by trubador
It seems you have correctly specified the model. It is hard to tell what is wrong going on with the model without seeing your data. It would be helpful, if you can post your workfile that contains the data. That said, you can first estimate a garch model to gain insight into domain of your model's parameters.
Re: Problems with estimating Realized GARCH
Posted: Tue Jun 28, 2011 5:51 am
by evaeva
Hi,
thank you for the answer. I attach the initial workfile (realized_garch.wf1) with only two variable "x" and "r" and then the code and then the resulting workfile (realized_garch_results.zip).
Many thanks!
Eva
Re: Problems with estimating Realized GARCH
Posted: Tue Jun 28, 2011 7:05 am
by trubador
I modified the following section so as to get initial values from regular garch estimation. You can delete unnecessary output once you decide which ones to keep.
Code: Select all
'declaration of coefficient vectors to use in RealGARCH likelihood
equation init.arch(1,1) r @ x(-1)
init.makegarch h0
series z0 = r/@sqrt(h0)
equation ols.ls x c h0 z0 (z0^2-1)
coef(1) omega = init.@coefs(1)
coef(1) beta = init.@coefs(3)
coef(1) gamma = init.@coefs(4)
coef(1) xi = ols.@coefs(1)
coef(1) phi = ols.@coefs(2)
coef(2) tau
tau(1) = ols.@coefs(1)
tau(2) = ols.@coefs(2)
coef(1) sigma=ols.@se^2