EViews.com

Posted: **Mon Sep 16, 2019 1:58 pm**

Hi Eviews team!

I need a help to understand how Eviews calculates EGARCH log-likelihood. I firstly estimated an EGARCH and, using Eviews coefficients results, tried to replicate Eviews log-likelihood output. However, I am unable to get the same value. The difference is not that big, but I am curious since I can match it using GARCH(1,1) with the information available on Eviews Help.

I am sending attached a file with my manual calculation that includes a random series as an example. There's a GARCH and an EGARCH tab with Eviews and manual calculation results.

Could you please let me know if I am missing anything when replication EGARCH log-likelihood?

Thank you very much for your support

Posted: **Wed Sep 18, 2019 9:24 am**

For eGARCH the first observation for the variance is calculated as:

Code: Select all


log_garch = omega + alpha*sqrt(2/pi) + beta*log_garch0

If you change your cell value in C3 to:

Code: Select all


=EXP($M$2+$M$3*ABS(SQRT(2/PI())+$M$4*LN(C2)))

you should match (up to rounding error)

Posted: **Thu Sep 19, 2019 3:02 pm**

Perfect! Many Thanks!

EViews.com

EGARCH Log-likelihood calculation

EGARCH Log-likelihood calculation

Re: EGARCH Log-likelihood calculation

Re: EGARCH Log-likelihood calculation