Page 1 of 1

GMM and selecting best model

Posted: Thu May 05, 2016 3:55 am
by willyg09139
I have two models estimated via GMM, how do I compare them?

Presumably with AIC, BIC (which is the case for OLS). However, information criterion is not provided in the output (information criteria is provided with OLS). Is there a way to get the information criteria using GMM methods?

If information criteria is not available in GMM, could I choose the model with the best R^2 & lowest standard error of the regression?

Apology if I am asking a silly question.

Re: GMM and selecting best model

Posted: Mon May 09, 2016 9:29 pm
by dakila

Re: GMM and selecting best model

Posted: Wed May 11, 2016 12:11 pm
by nemanja100
Hi guys!
I have estimated c ar(1) ar(12) model, because there was certain autocorrelation on k=12. Its monthly data, series has one unit root.
How to write this model?
reduced ARIMA(12,1,0) or
ARIMA(1,1,0)x(1,0,0)?

I must say sar(12) is not significant.