[River Huang]

Hi, Nicu, I think you can give it a try, and consider small sample correction (xtdcce2/xtcd2) if possible.

[mirza aboubakar]

I have panel data withT=5 and N=607 based on the Hausman test my model is fixed but my data is also suffering from cross-sectional dependence according to Pesaran cd test, I also checked through Friedman but this test is showing no cross-section dependence then I did frees test that shows the presence of cross-sectional dependence so first on what test I should rely and if I assume that the problem of cross-sectional dependence than what test should I use to manage cross-sectional dependence at the same time my model is suffering from AR1 and GroupWise hetro found from xttest3.

[mirza aboubakar]

is there any command in Stata to check that whether the autocorrelation is general or panel specific moreover how to check that whether my data is suffering from moving average correlation.

[Carlo Lazzaro]

Mirza:
I would probably go -xtreg- with clustered standard errors.

[mirza aboubakar]

Mirza:
I would probably go -xtreg- with clustered standard errors.

But as per my case xtreg cluster or robust option don't remove the cross sectional dependence how I would justify and if I use xtscc then accordingbto the author we first need to confirm that wherher fixed effect model is still applicable or not he suggested pannel robust hausman to run again so I am unable to run in the stata to assess that what is my suggested model and last thing I also have dummy variable it is also ommited when I run fixed effect any suggestions how to run dummy variable with fixed effects

[Carlo Lazzaro]

Mirza:
it's strange that with a N>T panel dataset (where N is remakably larger than T) cross-sectional dependence can bite that hard.
Moreover, the test you performed are sensible for T>N panel datasets: hence, I would first check whether the p-values you got from them are trustworthy when the N dimension is dominant.
As far as the easier question is concerned, there's nothing we can gainst the -fe- machinery: time-invariant predictors will be wiped out.