Page 1 of 1

Importing unbalanced panel

Posted: Wed Mar 14, 2018 8:39 am
by mikeguima
Hello everyone,

I'm currently trying to figure out the correct way to set up a workfile correctly for my dataset. My data (I send the excel file attached) is unbalanced panel data in the sense that not all the firms (my dissertation is in finance) have data for all years (from 2005 to 2016). My goal is to setup a pooled OLS model for my data.

When I imported the data, Eviews set the workfile data as "unstructured/undated", but it doesn't look correct since my data is dated. Therefore, i'd like to know what the best way to structure the data would be for the purpose of my study and how I could do that. I looked into Eview's users guide but I didn't find it very clear.

I'm running Eviews 10.

Thank you in advance!

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 9:27 am
by EViews Gareth
Once you have brought it in as undated, click on Proc->Structure/Resize Current Page and change the structure type to dated panel. Then enter the name of your cross-section and date series, and you should be good.

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 9:37 am
by mikeguima
Should I leave the default "balance between starts and ends" ticked? In addition, the software reports "Indicies do not uniquely identify observations. 32 duplicate date(s) found". Create CELLID to identify observations". What does this mean and should I click yes or no?

Thank you!

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 10:03 am
by EViews Matt
It looks like the data for several companies, such as Bollore, is present in the Excel file twice. That will mess up the restructuring to a dated panel.

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 11:08 am
by mikeguima
EViews Matt wrote:It looks like the data for several companies, such as Bollore, is present in the Excel file twice. That will mess up the restructuring to a dated panel.

Indeed!! :shock: Thank you very much! I would have never noticed that! It's working perfectly now! Thank you!

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 11:37 am
by mikeguima
One more question that arose: I need to include year dummies in this estimation. One of the variables Eviews generates is "year", menaing the year for each observation. If I include "year" in the estimation of my regression is it analogous to including year dummies or if not, is there other way to do it?

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 11:45 am
by startz
year is just the value of the year. If you want dummies try @expand(year)

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 11:46 am
by EViews Gareth
Or include period fixed effects.

Re: Importing unbalanced panel

Posted: Wed Mar 14, 2018 11:58 am
by mikeguima
I tried @expand(year) but it returns a "near singular matrix" error.

Period fixed effects seems to be working! I went to the panel options tab, period and fixed effects, is it correct?

I also wanted to estimate using the White or Huber-White Var/Covar matrix estimator to validate inference under heteroscedasticity and autocorrelation, but I can't seem to find it in the options now that the data is set as panel. Do you now where I can find that option?

Thank you again!

Re: Importing unbalanced panel

Posted: Wed Aug 08, 2018 10:38 pm
by msyed
Dear All,

I am trying to run a regression for unbalanced panel data set. I have lot of empty cells at my spreadsheet for missing data.

1. Is there a recommended way to turn the unbalanced panel data into a balanced panel data using EViews?
2. May the regression result vary if an unbalanced panel data set is turned into a balanced panel data set?

Many thanks for your help.