Hausman test: failure to meet asymptotic assumptions

[Tina Samuel]

Hi I am running a simple model on countries green exports. So I wanted to check if I should random or fixed effects.
I ran Hausman test, but got very strange results. Any ideas what might be wrong with this dataset?

[Andrew Musau]

In applied work, the hausman command is limited because one would likely want to cluster standard errors due to potential heteroskedasticity and serial correlation, and the command does not offer a robust version of the Hausman test. I would suggest installing xtoverid from SSC and using that instead.

ssc install xtoverid, replace
help xtoverid

[Tina Samuel]

Yes this analysis suffers from autocorrelation and I use cluster to fix for it.
Thank you the xtoverid worked well.