Workfile structure for unbalanced panel data

[AbnormalDistribution]

I did my best to stay abreast of what is already out there on this topic, as well as read the EViews primer on panel data:

http://www.eviews.com/help/helpintro.ht ... l#ww100941

However, a step-by-step procedure for the correct workfile settings for unbalanced panel was not included.

My data set frequency/cross sections:

• 5 years of monthly data

• 2000 cross sections

Note: I won't be imputing/ad hoc data wrangling to fill the missing values. I would like to embrace the unbalanced panel framework to allow cross-sections to enter and leave, hopefully keeping survivorship bias to a minimum.


Can someone clarify the industry standard settings in EViews for this type of workfile structure?

(I'm using Eviews 9 and only have: Unstructured/undated, dated regular frequency and balanced panel as drop down menu options.)

[EViews Gareth]

You cannot create a new unbalance panel. You can only import unbalanced data in from an external source.

[AbnormalDistribution]

Thanks, that's good to know.

But going back to finding the EViews way to set this up, we'd have to just set the workfile to unstructured undated? Then File->Import->Import from file?

Also I guess we need to specify the number of observations to time Periods * cross sections? In my case that would be 60*6000.

Maybe it would be easier to copy and paste the unbalanced panel into series objects once the proper workfile is set up?

Curious about any differences on how estimation is handled under these settings too. (E.g: if the linear algebra would break down from not having an observation/how EViews users deal with that)

If there is any documentation on unbalanced panel in EViews, I would appreciate that too.

Cheers

(Postscript: be as verbose as you like, it won't be wasted on me. To borrow from Margin Call: treat me like a single-celled organism.)

[AbnormalDistribution]

Would this approach seen in the Harrison Rubinfeld panel example work?

http://www.eviews.com/help/helpintro.ht ... es.html%23

... and enter “TOWNID” as the Identifier series. EViews will prompt you twice to create a CELLID series to uniquely identify observations. Click on OK to both questions to accept your settings.

EViews restructures your workfile so that it is an unbalanced panel workfile. The top portion of the workfile window will change to show the undated structure which has 92 cross-sections and a maximum of 30 observations in a cross-section.

Seems this gives us an unbalanced panel, not 100% sure though.

[EViews Glenn]

You'll want to provide the date series too.