EViews.com

Dear all

Why does Eviews uses equation 7.22 in the manual(page 209) as the variance equation in the EGARCH model
and not the original Nelson (1991) specification as in the equation below equation 7.22 (also on page 209)?

Also, in the manual they explain that this will only affect the intercept.
They argue with p=1 and normal distribution the difference will be alpha_1*(2/pi)^(1/2). Could somebody explain this?

Cheers
S

It's a relatively harmless normalization that simplifies the calculation since we don't need to compute the expected value term for all possible distributions.

Having said that, I'm glad you pointed this out as I notice that there's an error in the documentation. The correct terms with respect to alpha are

|e/sigma| - E(|e/sigma|)

They put the absolute values in the wrong place in our docs.

The absolute value is why you get the sqrt(2/pi) term for the normal.

Ok, Thanks!

Now it is indeed consistent with Nelson (1991).

S