R squared equals 1 when running spatial fixed effects

[Sara Restrepo]

Hello everyone,

I´ve been asked to show that, when you run your outcome variable against your geographical fixed effects, the adjusted R squared equals 1.
The idea behind this is that, as your database is in that unit of analysis when running:

reg y i.spatial_fixed_effect

you should get that that spatial fixed effects explain all the variance you have.

I have tried several times with a dataset but haven´t succeed.

Has anyone proved this with a constructed database so I can take a look at it?

Thank you very much,

Regards

Sara R

[Phil Bromiley]

If this is a homework, please respect the listserve's FAQ and not ask about homework.

Whether a fixed effect of any sort explained all the variance would depend on whether you have observations varying within the domain of the fixed effect.