When using weighted and non-weighted on Panel regression

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

trinhndt
Posts: 8
Joined: Mon Mar 07, 2011 11:46 pm

When using weighted and non-weighted on Panel regression

Postby trinhndt » Sun May 08, 2011 2:51 am

Hi all, I try to learn panel regression. But, I can not understand that when we should use weighted and non-weighted in running the regression for panel data. Because, I find that there are many option in equation estimation windows for choosing such as cross-section (three options: none, fixed effect, random); period (three options: none,fixed effect, random) and each of them has another option: weighted and non-weighted.
So, could you guys help me to learn when we should choose weighted and non-weighted.
Thank you so much!!!!

christina.sakali
Posts: 11
Joined: Mon May 09, 2011 3:18 am

Re: When using weighted and non-weighted on Panel regression

Postby christina.sakali » Mon May 09, 2011 3:19 pm

I am not expert, but from what I know, if you select the weighted option you will estimate a Weighted Least Squares regression (WLS) which is a special case of the Generalised Least Squares Method (GLS).
It is used in case you have detected the problem of heteroskedasticity in the cross-section or period dimension. Through this method the equation is transformed, which results in residuals being omoskedastic (which is what we want).

EViews Help decribes briefly the different estimation methods in the section titled 'Panel Estimation', so you can have a look there.

Also, I am sure you can find info on GLS methods and fixed VS random effects models in any econometrics textbook (try looking in the chapter on heteroskedasticity and the chapter on panel data).

Hope this helps.

trinhndt
Posts: 8
Joined: Mon Mar 07, 2011 11:46 pm

Re: When using weighted and non-weighted on Panel regression

Postby trinhndt » Mon May 09, 2011 11:49 pm

christina.sakali wrote:I am not expert, but from what I know, if you select the weighted option you will estimate a Weighted Least Squares regression (WLS) which is a special case of the Generalised Least Squares Method (GLS).
It is used in case you have detected the problem of heteroskedasticity in the cross-section or period dimension. Through this method the equation is transformed, which results in residuals being omoskedastic (which is what we want).

EViews Help decribes briefly the different estimation methods in the section titled 'Panel Estimation', so you can have a look there.

Also, I am sure you can find info on GLS methods and fixed VS random effects models in any econometrics textbook (try looking in the chapter on heteroskedasticity and the chapter on panel data).

Hope this helps.


Hi Christina,
Thanks so much for your sharing. I will looking for the textbook as your suggestion!
have a nice week!!!!


Return to “Econometric Discussions”

Who is online

Users browsing this forum: No registered users and 22 guests