Spatial durbin model

[Jonas Ngouhouo]

Good Evening,
Please, I would like somebody help me on one of both problems (or both if possible) I faced when estimating a Spatial Durbin Cross Sections Model in Stata?

1) I am using the module spmstard for Multiparametric Spatio Temporal AutoRegressive Regression. I have a problem when generating the spatial weight matrix using spweight. Indeed, I have latitude and longitude cordinates of 1035 households surveyed with a GPS and I would like you to informe me on the type of data to use. Please, how should I proceed to obtain the weight matrix with such data ? (The coordinates unit is the decimal system in decimal degrees: e.g. 2.77934, 13.356)

2) I have also tried estimating the Spatial Durbin Cross Sections Model using the module spmlreg and the spatwmat (for the spatial weight matrix) . I got the results. But, as mentioned by Lesage, the SDM model estimates cannot be interpreted as partial derivatives. Please, could some one indicates me how to calculate the direct, the indirect as well as the total effect after running the spmlreg module?

I look forward to having your reply.

Thank you

Best regards.

[Emad Shehata]

Check here

http://www.ats.ucla.edu/stat/stata/faq/morans_i.htm

To convert latitude and longitude cordinates to binary spatial weight matrix
spweight is valid only with binary or inverse matrix

[Jonas Ngouhouo]

Dear Pr. Emad elmessih Shehata.

I Thank you very much for your reply. I thank you also for making many stata modules including the spmstard module available to researchers.
I am going to follow the website you are advising to me.

Best regard.

Jonas Ngouhouo Poufoun
PhD Student and Temporary Lecturer (ATER)
BETA - Univ of Lorraine
INRA - Laboratory of Forest Economics
[email protected]
54000 Nancy, Fce.