Unbalanced panel data

For questions regarding the import, export and manipulation of data in EViews, including graphing and basic statistics.

Moderators: EViews Gareth, EViews Jason, EViews Steve, EViews Moderator

Aktar
Posts: 61
Joined: Thu May 21, 2009 5:08 am

Unbalanced panel data

Postby Aktar » Sun Oct 07, 2018 6:30 am

Hi all,

I have a datafile with seven cross-sections (countries) and 131 time observations. I have several explicative variables available for each of the seven countries, but for some of other explicative variables, they are available for four, five or six countries only.

I tried to structure my workfile as undated or balanced panel. But when running the estimate (OLS), eviews excludes the three countries for which all explicative variable are not available and run the estimates as if the panel data was balanced (only four cross-sections are used in the estimation process). When I exclude these variables that are not available for the seven countries, all the seven cross-sections are taken into account.

So my question is: Why I can't run panel estimates over unbalanced data where some explicative variables are not available (i.e. any point/value) for some inputs of the cross-section?

I attach my workfile (equation object "linear"). The variables that are not available for the seven cross section are china_fdix china_pfm china_fdim.

thank you
Attachments
untitled.wf1
(262.61 KiB) Downloaded 5 times

startz
Non-normality and collinearity are NOT problems!
Posts: 3379
Joined: Wed Sep 17, 2008 2:25 pm

Re: Unbalanced panel data

Postby startz » Sun Oct 07, 2018 7:25 am

Observations are only included in a regression when all the variables have values for a that observation.

Aktar
Posts: 61
Joined: Thu May 21, 2009 5:08 am

Re: Unbalanced panel data

Postby Aktar » Sun Oct 07, 2018 7:57 am

So an unbalanced panel is a one where there are missing values while the explicative variable remains available for all cross-sections?
Thank you

startz
Non-normality and collinearity are NOT problems!
Posts: 3379
Joined: Wed Sep 17, 2008 2:25 pm

Re: Unbalanced panel data

Postby startz » Sun Oct 07, 2018 8:08 am

Almost. You don't have to have all values for all cross-sections, but you do need all values for some cross-sections.


Return to “Data Manipulation”

Who is online

Users browsing this forum: No registered users and 8 guests