Interpreting xtreg output

[Bart Boersma]

hi,

I have searched this and many websites in order to completely understand the output of xtreg, fe. I am currently writing my thesis and this is my first time using paneldata.
my questions are mainly about this part of the table:

Fixed-effects (within) regression Number of obs = 50,407
Group variable: indiv_index Number of groups = 14,896

R-sq: Obs per group:
within = 0.0138 min = 1
between = 0.0443 avg = 3.4
overall = 0.0364 max = 23

F(14,35497) = 35.51
corr(u_i, Xb) = -0.0963 Prob > F = 0.0000

and this part:

sigma_u | .66461587
sigma_e | .43401624
rho | .70103988 (fraction of variance due to u_i)
----------------------------------------------------------------------------------
F test that all u_i=0: F(14895, 35497) = 6.54 Prob > F = 0.0000


What does the final f-test mean? (in red not purple)
How can I intepret the rho?
how can i intepret the sigma_u and the sigma_e?

thanks in advance.

[Amin Sofla]

I think that the best place to get a good grasp of panel data is the Stata manual.

[Carlo Lazzaro]

Bart:
see Examples under -xtreg- entry, in addition to any decent panel data econometrics textbook (Stata users read, among Others valuable textbooks, the following one: https://www.stata.com/bookstore/micr...metrics-stata/).
In brief:
- the purple F-test is equivalent to the F-test appearing in OLS;
- the red one tells you that -xtreg,fe- fits your data better than pooled OLS.

PS: crossed in the cyberspace with Amin's helpful reply, that I've echoed somehow.

[Bart Boersma]

thank you