Missing Cross sections in Panel regresion

[MaStein]

Hi!

I am currently working on a data panel project, and I have some trouble with Eviews in that some cross sections is automatically removed when I run regressions.

The full panel consists of 46 companies (cross sections) and yearly observations from 2000 to 2014.
The dependent variable is leverage ratio, which I regress upon several accounting variables such as log of sales, profitability, tangible assets etc.

The problem is that, when I run the regression, Eviews only includes 42 cross sections in the regression output. I know this is due to the fact that some data points are missing (some of the firms didn't exist the whole 14 year period). The weird thing is that the number of firms with missing values is more than 4 (companies automatically removed).

There are five likely candidates that has been removed:

Company 21 is missing 11 years
Company 24 is missing 8
Company 28 is missing 9
Company 30 is missing 8
Company 33 is missing 11

Is there any way to find out exactly which companies that has been removed? And is there any way to get Eviews to include them?

Best Regards,

Marie

[EViews Gareth]

EViews will only drop a cross-section entirely if there are no valid observations for it at all.

The easiest way to see which observations were included in a regression is to use Proc->Make Residuals to look at the residual series. Any observations for which the residuals are NA are ones that were not used during estimation.

There is no way to include a missing observation in a regression.

[MaStein]

Thank you for replying so quickly!

When I use Proc->Make residuals, a list of all 46 cross sections appears.
However, in the regression output it is stated that only 42 is included. The output also states that the adjusted sample period is 2001, but I'm guessing that is probably due to the fact that there is only a few of the companies with data in 2000.

Does this mean that all the cross sections is included in the regression?

[EViews Gareth]

No, it doesn't. There are probably some cross-sections for which all observations of the residuals are NA. Those are the ones that were dropped completely.