xtpcse and xtgls with fixed effects

[lucho sanchez]

Hello community:
I want to robustly estimate (xtgls) a panel data model, for both FE and RE. The following commands are correct:

For FE:

xi: xtgls Y X i. id i.year, p (c) c (a)

For RE:

xtgls Y X, p (c) c (a)

Thank you very much!

[Joro Kolev]

If you mean that you want robust standard errors/variance, you can get them with my post estimation command -xtglsr-. Type from within Stata

 ssc install xtglsr

[lucho sanchez]

Thanks Joro!!
but beyond that, I have the doubt if the commands - xtgls or -xtpcse are by default for fixed effects, or do I need to adapt them.

[lucho sanchez]

If you mean that you want robust standard errors/variance, you can get them with my post estimation command -xtglsr-. Type from within Stata

ssc install xtglsr

that is looking beautiful!

[Joro Kolev]

For -xtgls- you definitely need to include the fixed effects manually, so in general it is not a fixed effects estimator, you of course can include fixed effects manually as you showed in your first post.

I have not used -xtpcse-, but I think it is same with it, you need to include manually the fixed effects.

Thanks Joro!!
but beyond that, I have the doubt if the commands - xtgls or -xtpcse are by default for fixed effects, or do I need to adapt them.

[lucho sanchez]

For -xtgls- you definitely need to include the fixed effects manually, so in general it is not a fixed effects estimator, you of course can include fixed effects manually as you showed in your first post.

I have not used -xtpcse-, but I think it is same with it, you need to include manually the fixed effects.

Thanks a lot, is very clear for me.