Spatially lagged endogenous variable

[Dario Maimone Ansaldo Patti]

Dear All,

I estimate a spatial model (specifically a SARAR). One of the variables in the model is likely to be endogenous. Therefore I use the command spivreg. Additionally, I believe that this endogenous variable is also spatially correlated.

When reading the user manual about spivreg, I found:

"You cannot include in the model spatial lags of the endogenous regressors or spatial lags of the excluded exogenous regressors."

Now my question is whether it is impossible to do that or whether Stata cannot do that. I tried to search online but I did not find any conclusive answer about whether it is possible to estimate a model with a spatially lagged endogenous variable.

Any suggestion would be highly appreciated.

Best regards,

Dario

[Sebastian Kripfganz]

That's a limitation of spivreg/spivregress. My guess is that the programmers did not want to deal with the extra complications of allowing that much flexibility. I do not see any reason why it should not be possible from a theoretical point of view.