[Ale Zapata]

I mean if there is a command that explain why STATA adjusted for 5o cluster in the variable you chose and not 51 or 49. Is It possible to make a graph that would show the data clustered by STATA (vce(cluster var)) command? Thank you in advance.

[alexander Zangria]

Hi there, I have a dataset with heterogeneity, serial correlation and cross-dependance. I need to run a fixed effect . In my dateset T>N so if I understood correctly from previous comments I should use xtregar, fe command? With this command these issues will be fixed? Why not vce(cluster var)? Thank you in advance.

[Carlo Lazzaro]

Ale:
standard errors are clustered on the variable decided by the researcher, provided that the procedure is feasible.
The fact that the number of clusters appearing in the outcome table is lower than expected, may be due to missing values in any of the variables.

[Carlo Lazzaro]

Alexander:
in such a situation I would consider -xtpcse- with -i.company- among the predictors.
See also the options available from -xtpcse- for taking disturbances into account.