Right specification of GARCH models?

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

Posts: 43
Joined: Wed May 31, 2017 7:37 am

Right specification of GARCH models?

Postby AfonsoRod » Wed May 22, 2019 12:37 pm

Dear all,

I would like to know how can we decide on the proper specification of our GARCH model in Eviews. Mainly, I would like to know:

1) Witch values of 'p' and 'q' should we use in our GARCH(p,q)?

2) Witch error distribution should we use (between "Normal (Gaussian)"; Student's t; Generalized Error (GED); Student's t with fixed df.; and GED with fixed parameter)

Note that my dependent variable is the electricity price with daily frequency, for a ten year period. I have more than one explanatory variables including a considerable amount of dummies. I have also had a lagged dependent variable [ price(-1 )].
As said before, my main struggle is to know whether my GARCH model is rightly specified or not. Also, in the residuals diagnostics I've been facing problems with the Correlogram Q-statistics test (Correlogram of Standardized Residuals), since the p-values are zero (therefore, significant) for all lag values. The rest of the tests (Correlogram Squared Residuals, Histogram Normality Test and ARCH LM test - ARCH) seem to be fine.

Can anyone help me with this?

I thank you all in advance.

Yours sincerely,
Afonso Rodrigues

Return to “Econometric Discussions”

Who is online

Users browsing this forum: No registered users and 4 guests