[Openspace] error correction models with spatial dependence

[Mario Gellrich]

Hi,

I have panel data (measurements at the community level from two years), the 
dependent variable is the rate of change of the forest area between both 
years. The idea is to calculate a model that considers both serial as well 
as spatial dependence. Recently, I found an article by Paul Elhorst in the 
book "Spatial Econometrics and Spatial Statistics" by A. Getis et 
al.(Eds.)(2004), where the author discusses a model with first order serial 
and spatial autocorrelation. This corresponds approximately to what I'm 
looking for but I'm not interested in the status of y at time t but in the 
rate of change between both years, which to my mind would mean that the 
spatial error correction model would be the appropriate modelling 
technique. My question is, does anybody know, whether I' m on the right way 
and if so, are there some articles about this topic (I did not found 
references on spatial error correction models in the literature). If I'm 
wrong, what alternatives are possible?

Mario Gellrich


Mario Gellrich

Swiss Federal Research Institute(WSL)
Mario Gellrich, Dipl. Forstwirt
Forest Department, Economic Section
Zürcherstrasse 111
CH-8903 Birmensdorf

Tel:  ++41 (0)1 739 2520
Fax: ++41 (0)1 739 22 15
Mail: mario.gellrich at wsl.ch
Web: http://www.wsl.ch/phonebook/STAFF/2409.ehtml