Unbalanced Panel Random Effects Fixed Effects Dependent Variable divided through 0

[Seb Bau]

Hi ,
I have some values as dependent variables calculated like 7/0. Does Stata treat them like missings?
How does STATA work with them in an originally balanced Panel set with xtset (firm, time) ?
When I do xtreg, re or fe does it exclude the total firm or only the year?
I have to exclude then th firm on my own, because I Need for These estimation methods a balanced Panel?

[Carlo Lazzaro]

Seb:
Stata will treat 7/0 as missing and, via listwise, deletion, rule it out from any descriptive/inferential statistics.
Under -xtreg-, Stata will exclude only the year the firm has missing values in any of the variables.
There's no need to create a balanced panel dataset, as Stata ca handle both unbalanced and balanced panel datasets.
Besides, if missingness is informative, you will end up with a biased subsample of the original dataset.

[Phil Bromiley]

To add to Carlo's points, you should also think hard about whether there is another way to represent your dv. If you have 7/0, you may also have some close to zero denominators that result in extreme outliers.

One option to consider is using the 7 as the dv and controlling for the 0 variable. This may introduce problems with error terms, but those are generally easily handled.

You'll have a greater chance of a useful answer if you follow the FAQ on asking questions -- providing Stata code in code delimiters, Stata output, and sample data using dataex.