xtreg, fe and xtreg, re when T>N

[Sidra Ilyas]

Hi Dear,

Can we use xtreg, fe and xtreg, re when T>N in our panel data?

Many Thanks.

[Carlo Lazzaro]

Sidra:
you would be better of with -xtregar- and -xtgls-.

[Sidra Ilyas]

Thanks Carlo,

But if we still apply xtreg, fe and xtreg, re when T>N, how will our results be affected?

Can I please say that the Maximum T I have is 28, while N is ranging between 3 and 17 in different estimations. How the results will be affected if I apply xtreg, fe or re here?

Also, can you please tell me when was the xtregar first introduced?

Regards,
Sidra

[Carlo Lazzaro]

Sidra:
- in sum: it's a matter of serial corrrelation, that deserves more precise ways of modelling when T>N (something that -xtreg- can't do that well);
- for more details on -xtregar-, see the related entry in Stata .pdf manual.

[Sidra Ilyas]

Hi Carlo,

Thanks for the response. Is specifying vce(robust) with xtreg is insufficient in this case?

Regards,
Sidra

[Sidra Ilyas]

In addition to my query above and given our conversation before that, do you think xtgls and xtregar work when T>N and the explanatory variables are a mix of I(0) and I(1) variables? Is there any requirement of the stationarity of the variables for xtgls and xtregar?

Please recall that T could be around 28, while N ranges between 3 and 17 in different estimations.

Please respond at your earliest.

Regards,
Sidra