[Joro Kolev]

On somewhat related note, and promoting my awesome command -suregr-: I have written a post estimation command for -sureg- that calculates robust, and cluster robust standard errors/ variances for -sureg-. So now you can get robust and cluster robust standard errors/variances for -sureg-.

Type from within Stata

ssc install suregr

and it will all be revealed.

[Jane Ruu]

I have just solved it by myself. The code is right but it was not running because I had too many observations.

Can you tell me how you solved your problem? I'm facing the same problem now. It only works when I use a limited set of variables. But I want to keep all variables I have now (3 independent variables, 5 controls, and year dummies). I have about 520 observations (47 firms, 15 years panel data).

[Nariman Sayed]

Dear All,

hope everyone's problems have been solved. I am wondering how to embrace FE with XTSUR, if anyone reached an answer.

thanks in advance !

[Nariman Sayed]

Dear Prof Joro Kolev .. many thanks for your important contribution !!.. However, I wonder if this command can also address hetero problem as well, like the serial correlation was addressed by robust.

thanks in advance !

[Joro Kolev]

Dear Prof Joro Kolev .. many thanks for your important contribution !!.. However, I wonder if this command can also address hetero problem as well, like the serial correlation was addressed by robust.

thanks in advance !

Yes, when we compute variances robust to cluster correlation, or Newey-West autocorrelation in time series, our variances are automatically robust to heteroskedasticity.

[Nariman Sayed]

Much thanks prof!!..

I have another inquiry if you don't mind.. does the assumption of contemporaneous correlation, which made us resort to SUR, exclude correlation of past values as well ? so that if I have like reversal causality(feedback from the dependent variable to some of independent ones) cannot I use SUR ?