Can I use a spatially lagged Dependent Variable while using Spatial Error Model?

Question

I have following issue:

I run spatial diagnostics on dependencies for my Log-Log Transformed regression model. LM Tests (including Robust) are highly significant.

Since I am using GeoDa, I cannot run robust Error or Lag models because it is not an option. I thought of manually creating a spatially lagged Y and include it in the Spatial Error model. Likelihood ratio test says it's much better than just using either one.

Is this approach valid or is it wrong?

Answer 1

Yes, but that's essentially the Spatial Durbin Model, now. Looks like the equation below:

y = ρWy + Xβ + θWX + u

SDM synthesizes the advantages of a spatial lag model and a spatial error model. You can work your way back to both models: for the lag, if θ = 0; and for error, if θ = -ρβ.

GeoDa is great for visualizations, but not regression modeling. Use R, STATA, Python or MATLAB.

For more depth in the math logic, see the following:

• http://rei.revues.org/3887 : One of the more popular articles on the subject, by James P. Lesage
• http://www.atlantis-press.com/php/download_paper.php?id=25840534
• Golgher, A. B., & Voss, P. R. (2016). How to interpret the coefficients of spatial models: Spillovers, direct and indirect effects. Spatial Demography, 4(3), 175-205.