FGLS on a cross-section data set

[Adam Red]

Hi all,

I am not good with econometrics, so I have a stupid question: "Is it possible to perform a FGLS regression on a cross section data set?"

The command xtreg asks me to define a time variable and panel identifier for my dataset. From this I am guessing the answer is no.

If no, are there bad implications if I set the time variable constant to 1 for every observation, and I list every single observation as the panel identifier (i=1,...N)?

[Jorrit Gosens]

See xtgls here.
Your suggestion will not help. With one observation per group you essentially have no panel shape to your data anymore.

[Adam Red]

Thanks Jorrit Gosens,
The link is very helpful.