The RATS Software Forum

Hi, do you know how to test for panel heterogeneity in rats?

Thanks a lot!!

What type of heterogeneity?

To know whether to use Levin and Lin unit root tests or Im et al unit root tests; whether we can assume that rho (in the unit root test) is the same for all individuals or not.

Thanks!!

If you have a choice, use @IPSHIN. The question is what is the behavior under the alternative. It's extremely unlikely that if the null rejects that a common stationary root would be the appropriate alternative.

So even if my data ended up being homogeneous (which I doubt) I am better off taking the conservative way of using @ipshin ? What about for cointegration tests, I have been using @paincoint which also allows for heterogeneity, do I need to test for heterogeneity first before performing it?

Thanks so much!

It sounds like you have your null and alternatives confused. Where there are tests, they would have a null of homogeneity, not heterogeneity so you would be testing for homogeneity. The unit root and cointegration tests need a separate decision as to the structure of the alternative, and heterogeneity makes more sense there.