N is smaller than T.

[Benjamin Onah]

Hello everyone.

Please, I have a concern. I'm studying the impact of ICT on youth unemployment in two regions. My N for the first region is 16 and T is 30 years. N for the second region is 6 and T is 30 years. So, N is smaller than T. In this regard, please which econometric model/method can I use and why? Thanks in advance.

[Carlo Lazzaro]

Benjamin:
tas you have a T>N panel dataset, take a look at: -xtgls- and -xtregar- Stata commands.

[Benjamin Onah]

Benjamin:
tas you have a T>N panel dataset, take a look at: -xtgls- and -xtregar- Stata commands.

Thanks for your comment. Any justification for using xtgls or xtregar?

[Carlo Lazzaro]

Benjamin:
-xtregar- supports -fe-, too.
The two T>N estimators also differ for their options about heteroskedasticity and autocorrelation (see the related help files and Stata.pdf manual entries for more details and examples).

[Fei Wang]

Benjamin, I think another option is to use -xtscc-, a user-written command. This method is powerful when you have large T (no matter N is large or small) and accounts for general forms of cross-sectional and serial correlations. Well, T = 30 is not that large, but at least you may have a try.

[Benjamin Onah]

Benjamin:
-xtregar- supports -fe-, too.
The two T>N estimators also differ for their options about heteroskedasticity and autocorrelation (see the related help files and Stata.pdf manual entries for more details and examples).

well appreciated.

[Benjamin Onah]

Benjamin, I think another option is to use -xtscc-, a user-written command. This method is powerful when you have large T (no matter N is large or small) and accounts for general forms of cross-sectional and serial correlations. Well, T = 30 is not that large, but at least you may have a try.

well appreciated.