Including an extra independent variable in a panel regression reports missing t-statistics/std. errors

[Utsav Saksena]

Hello everyone,

I have a panel dataset, where the panel variable is industry code (I am doing an industry level analysis).

I am trying to run the following regression -

 xtreg A B C D i.year, fe

This essentially means that I want to see what relationship independent variables B, C, & D have with A, accounting for year specific fixed effects and industry (panel variable) fixed effects.

Running the following model gives no errors -

 xtreg A C D i.year,  fe

In fact, the coefficients and signs are as expected

Upon including variable B, somehow, the results just show a blank dot (.) for all t-statistics, standard errors and F-value/statistic.
Additionally, the R squared values (within, between, overall) all show '1'.

I thought this was a multicolinearity problem, but seeing the correlation table does not show anything very unusual, at least to me -

corr A B C D
(obs=945)

| A B C D
------------------------------------------
A | 1.0000
B | 0.6814 1.0000
C | 0.0781 -0.6765 1.0000
D | 0.2688 0.1620 0.0498 1.0000


Would you have any suggestions as to what I could be doing wrong? Thanks a lot!

[Carlo Lazzaro]

Utsav:
welcome to this forum.
As you may be already aware of having read FAQ, you can increase your chances of getting helpful replies by:
- reporting what you typed and what Stata gave you back via CODE delimiters;
- sharing an excert/example of your data via -dataex-. If your data.

If your data are confidential, try to anonimize/mask them or to reproduce tha same problel with an official Stata dataset;

[Sarah Edgington]

In addition to Carlo's advice on how to help us help you, I would also suggest looking at the relationship between year and industry and your B variable.