meaning of corr (u_i, Xb)

[Timea De Wispelaere]

Hello,

I know, thanks to this forum, that a high corr (u_i, Xb) means that the RE is likely not to be appropriate (given that it is assumed to be zero). However, I do not find this in any econometrics textbook. What does it mean? Does it mean the correlation between the error term and the regressors? And what does it imply? Could anyone advice me an econometrics text book that gives some more explanation on this issue?

Kind regards,
Timea De Wispelaere

[Carlo Lazzaro]

Timea:
the -xtreg- entry in Stata .pdf manual can give you both details and references.
Another valuable source on this topic is the help file of the community-contributed programme -xtoverid-.

[Timea De Wispelaere]

Thank you Carlo!