Panel specific ar vs common ar

[Michael Lee]

Hello everyone,
what are the tests/ criteria that specify which type of options should I choose when I'm using xtpcse command?
1. Exclusively when I should corr(psar1) rather than corr(ar1)?
2. Also when I should use rhotype(tsc)?

Thank you in advance,
Best regards,
Michael

[Joro Kolev]

If you do this manually, you would simply calculate a panel specific AR coefficient, and then test whether these are the same across panels with any test you can. Say a Wald test or anything else that is convenient.

I do not know whether -xtpcse- offers automatic testing facilities to carry out this test.

The rhotype option is a technicality, there are a couple of ways to calculate the autoregressive parameter, and they are asymptotically equivalent. Just let the -xtpcse- choose this for you, or choose -regress- if there is such an option.

[Michael Lee]

Joro:
Actually, I didn't get the answer.

[Michael Lee]

Could you explain a little more about your answer?