Difference b/w xtpcse Y X i.id, corr(psar1) hetonly and reg Y X i.id ,vce(cluster id)

[Changho Moon]

Dear Statalists,

my panel data is consist of industry x time, which means large differences across industries.

So, I assume that the error terms are autocorrelated within industries and heteroscedastic across industries. i also assume that the error terms are uncorrelated across industries.


Under the assumption,

Which syntax is preferable among [xtpcse Y X i.id, corr(psar1) hetonly] and [reg Y X i.id ,vce(cluster id)]?

Thank you in advance.

[Carlo Lazzaro]

Changho:
I'd go

xtpcse Y X i.id, corr(psar1) hetonl

[Changho Moon]

Thank you for your help and warmth as always.