EViews.com

Hi,

In the Eviews help file I noticed two specifications for an EGARCH model: 29.22 and the one directly underneath. The only difference is E[e(t-1)/s(t-1)]. It says that this term reduces to SQRT(2/pi) under normally distributed errors. But what will term be under GED and student-t errors? Am I correct that Eviews uses formula 29.22 to model EGARCH? Why does it leave out the expectancy term than? Does Eviews use 29.22 under all three error distributions: normal, GED and student-t?

Thank you in advance.

EViews uses 29.22 in all cases.

Note that the manual points out that the equation below is Nelson's parameterization of the model (which is not used by EViews). As noted in the manul, the difference between the two models simply amounts to an adjustment to the intercept term.

Thank you for clarifying this for me Glenn.
Do you happen to know anything about setting up log likelihood functions?
http://forums.eviews.com/viewtopic.php?f=4&t=2661