News impact curves for various GARCH models in the rugarch-package

After having slept on it for a nightI now think that both the fGARCH
submodels (I am especially interested in the NAGARCH) and the eGARCH
model have a NIC that has to be expressed in terms of the z_{t-1}
since in both cases the effect of epsilon_{t-1} on sigma_t does depend
on sigma_{t-1} which is of course nonconstant.

But as stated before, the "newsimpact(fit)"-output of the eGARCH model
tells me that here the NIC was given in terms of epsilon_{t-1}.
Is this a typo?  Should it mean "z_{t-1}"?


Am 2014-06-14 7:25 PM, schrieb Johannes Moser:

Dear all,

I'm working with the really nice "rugarch"-package and currently have 
an issue with respect to the news impact curves (NIC).

In an attempt to plot several NIC into the same plot I realized that 
while the NIC for the sGARCH, the gjrGARCH and the eGARCH are given 
with respect to the epsilon_{t-1}, the NICs for the submodels of the 
fGARCH model are given in terms of z_{t-1}.

Firstly I am a bit confused since just like the fGARCH model, the 
eGARCH model (as to the eGARCH-model-setup in the vignette) is also 
given in terms of the z_{t-k} , k={1,...,q}.
But nevertheless the NIC of the eGARCH is given in terms of 
epsilon_{t-1} .
At least this is what the "newsimpact(fit)"-output tells me.
Why is it this way for the eGARCH but not for the fGARCH?

Secondly I'd like to have the NIC of all the different models 
depending on the epsilon_{t-1} for better comparison.
So for the fGARCH case I thought about calculating the epsilon_{t-1} 
values given the z_{t-1} values and the conditional mean and 
volatility.
Is this a good idea or is there an important reason why the NICs for 
the fGARCH submodels are NOT given this way?

Many thanks and kind regards,
Johannes

--