[R-sig-Geo] Estimating extended SDM (different weight matrix for independent and dependent variables)

[Roger Bivand]

On Mon, 19 Mar 2012, Roosbeh Nowrouzian wrote:

> Hello:
> I an trying  fit a different version of Spatial Durbin Model (SDM) model in
> R. My question is is it possible to define 2 different weighting matrix in
> models specification of SDM. So in the model specification instead of y =
> ?Wy + ?? + X ? + W X ? + ?, we specify model as  * **y = ?W1y + ?? + X ? +
> W2 X ? + ?*  . Is there a way to estimate such a model in R?
>

Yes, by fitting a lag model on the covariates and their manually 
constructed lags - use the lag method for listw objects ?lag.listw to 
apply the W2 listw object to your X variables, usually omitting the 
intercept.

The impact methods may however need attention, so you would have to 
re-read LeSage & Pace (2009) very carefully before trying to interpret any 
results.

Hope this helps,

Roger
>
>
> Thanks in advance for your help and consideration.
>
> Roosbeh Nowrouzian
>
> PhD student
>
> Department of Civil and Coastal Engineering
>
> University of Florida
>
> 	[[alternative HTML version deleted]]
>
>

-- 
Roger Bivand
Department of Economics, NHH Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no