Dear all,

I was looking to run a generalized hausman test for panel models with cluster-robust standard errors and found that the -xtoverid- could be used for this purpose. However, as I am applying multiway clustering, I use the vcemway option before my xtreg. This option allows me to run both random effects and fixed effects models with multiway clustering.

The various error messages returned by stata when applying -xtoverid- include "xtoverid not compatible with xtreg model fe", "cluster(): too many variables specified" or "no observations".

I also tried the -rhausman- command which returns error "__000000 not found".

Do you know of a hausman test version that works with multiway clustering?

Thank you and kind regards,
Anna