Hi,
I have panel dataset for 85countires. The time period is for 40 years(1978 to 2017), but the max of the object per group is only 14years because of unbalancing.
So I understand N is 85 and T is 40 and N > T.
I checked the heteroscedastic and autocorrelation in my dataset, to dealing with these I can use xtgls and xtreg , fe vce (robust), right?
Many other FAQs recommend to use xtreg , fe vce (robust) when N is larger than T, but I think my N and T are both enough large.
In that case, can I use xtgls ? The result of analysis is very different, I prefer to use xtgls rather than xtreg , fe vce (robust).
I have panel dataset for 85countires. The time period is for 40 years(1978 to 2017), but the max of the object per group is only 14years because of unbalancing.
So I understand N is 85 and T is 40 and N > T.
I checked the heteroscedastic and autocorrelation in my dataset, to dealing with these I can use xtgls and xtreg , fe vce (robust), right?
Many other FAQs recommend to use xtreg , fe vce (robust) when N is larger than T, but I think my N and T are both enough large.
In that case, can I use xtgls ? The result of analysis is very different, I prefer to use xtgls rather than xtreg , fe vce (robust).
Comment