I am new to spatial statistics and recently came across a problem I can't figure out.
I am running spatial models that predict county-level poverty rate using a set of county-level variables (unemployment, racial composition, etc.) and a few state-level variables (state minimum wage and education spending).
In my 1st model - a spatial lag model - I use the white poverty rate as my outcome. In my 2nd model - spatial error model - I use black poverty as my outcome. I use rook's spatial weights matrix for both models.
However, since my models includes 2 state-level covariates, I wanted to cluster my standard errors at the state-level. Can I cluster SEs in a spatial regression model, and does it make theoretical sense to do that? If so, any input on how I can do this in R would be appreciated.