Page 1 of 1

Eviews 9 clustered standard errors

Posted: Thu Nov 02, 2017 10:43 am
by oksanakim
Dear Moderators

I am using Eviews 9 currently and am working with Undated/Unstructured panel. I need to estimate an OLS with se clustered at firm level. I did the follows: modified the structure of the file by specifying Firm ID as a Identifier Series. Then, when estimating the regression, in the Panel option I left "none" in both the cross-section and time-periods field. I also chose "white cross-section" in the "coefficient covariance method". Did I do it correctly? Please just say Yes or No:-).

Also, do I correctly understand that in Eviews 10 clustering estimation is made much easier and that there are drop down boxed upon OLS estimation that allow choosing a clustering variable? Thank you!!

Re: Eviews 9 clustered standard errors

Posted: Mon Nov 13, 2017 11:15 pm
by EViews Glenn
You want White period as this allows for within cross section between period correlation.

Yes. http://www.eviews.com/EViews10/ev10ecest_n.html#cluster

Re: Eviews 9 clustered standard errors

Posted: Wed Nov 15, 2017 11:39 am
by oksanakim
Thank you Glenn.