Panel regression

noetsi

Fortran must die
#1
I am brand new to this method, but the federal government requires we run it next year with no information provided on that other than it will occur. I was wondering if any were familiar with it. It combines, in a basic way time series and cross sectional regression, but not in any way I have seen time series regression done before. I was wondering if the data has to be stationary or not (no mention is made in the literature I have seen on this topic).

It reminds me of multilevel regression, I was wondering if the cross sectional data is nested inside time.

It does not seem to assume homoskedasticity (in the same way that multilevel regression does not). I am unsure it assumes linearity or normality either.
 
Last edited: