Page 1 of 1

Filtering regressions with significant parameters

Posted: Fri Apr 14, 2017 7:44 pm
by Rantul
I have calculated 37,200 regressions and I would like to filter out those that have all their significant parameters at 5%, is it possible to program this? :P

Re: Filtering regressions with significant parameters

Posted: Fri Apr 14, 2017 8:16 pm
by EViews Gareth
Yes.

Re: Filtering regressions with significant parameters

Posted: Fri Apr 14, 2017 8:43 pm
by Rantul
how?? :| :| :| :| :| :|

Re: Filtering regressions with significant parameters

Posted: Sat Apr 15, 2017 7:30 am
by startz
I'm just curious as to why you want to do this?

Re: Filtering regressions with significant parameters

Posted: Sat Apr 15, 2017 8:54 am
by Rantul
I need to choose within a combination of time series models (garch models), and I need the model with the best information criterion to have all the parameters significant at least 5% (and that the residuals are random). with 37,200 calculated regressions, I wanted to know if there is a faster way

PD: for a AG-DCC model

Re: Filtering regressions with significant parameters

Posted: Sat Apr 15, 2017 10:28 am
by startz
I can see looking at the information criterion. But why would you specify that all the coefficients should be significant?

Re: Filtering regressions with significant parameters

Posted: Sat Apr 15, 2017 3:22 pm
by Rantul
To make sure than the GARCH model is well specified. If the variance has no asymmetry, it does not make sense to use a GARCH with asymmetry to model the residuals. (i need to calculate the standard residuals)