xsmle: two-regime Spatial Durbin Model

[Monika Banaszewska]

Dear Statalist Users,


I want to implement two-regime Spatial Durbin model in which spatially-lagged dependent variable is interacted with a dummy (d = {0;1}):

y_it = α + ρ₁ d_it Wy_it + ρ₂ (1 - d_it) Wy_it + β X_it + θ W X_it + μ_i + λ_t + Ɛ_it

Without this interaction it would be:

xsmle y x1 x2 x3, wmat(W) emat(W) type(both) model(sdm) fe type(,leeyu) cluster(code)

How to include interacted spatially-lagged dependent variable using 'xsmle' command?

[Andrzej Stasio]

Dear Monika,
I was wondering if you have found a solution to your problem?
Thanks for letting me know,
Best regards, Andrzej

[Sebastian Kripfganz]

Short answer: It is not possible to do this with xsmle because the log-likelihood function would have to be adjusted accordingly to account for the endogeneity of the interaction term.