General question on redundant fixed effects test

[jmd]

Hi,

I am fairly new to EViws and I am using it for a thesis-assignment.

It was suggested by my supervisor to use fixed effects models for analyzing some company data (however as I have not read much econometrics despite basics I have now had a crash-course in this) so please bare with me if my question seems outright stupid.

I have estimated an equation with both cross-section and period fixed effects and as most littertaure (i.e. Brooks 2008) suggest I have now done a Redundant Fixed Effects - Likelihood Ratio Test. So far so good.

The output i get is as below, while in all examples in books they get 0.0000 on everything.

Cross-section F 14.054642 (11,86) 0.0000
Cross-section Chi-square 116.253598 11 0.0000
Period F 1.985655 (9,86) 0.0506
Period Chi-square 21.334559 9 0.0112
Cross-Section/Period F 9.127958 (20,86) 0.0000
Cross-Section/Period Chi-square 128.675804 20 0.0000

My question is thus, how do i acertain that the model with both period and cross-section fixed effects is the most correct one, as i cannot reject the hypothezis that the Cross section-fixed effects is correct?

Any help would be great!

Regards

[akmalamiruddin]

Dear All

I have the similar result with yours, thus my question is same with your as well,
I am really appreciate, if you could share the answer or the other friend coment for me
thanks a lot, I am very greatfull if you sent it to my e-mail,akmal9@gmail.com

thanks very much

best reagard
akmal

[EViews Glenn]

I'm afraid that I don't quite understand the question...The test results compare three sets of models

Period/Cross vs. Period-alone
Period/Cross vs. Cross-alone
Period/Cross vs. None

You are correct in noting that the second test is borderline, but one would argue that given the results you would still be better off including the period effects for robustness at the cost of the efficiency..