Construction Method Of Spatial Weight Matrix In Spatial Econometric Model

Currently,the spatial econometric models are popular.And as an important part of spatial models,the spatial weight matrix reflects the spatial correlation between individuals and has a significant influence on the estimation of the models.Firstly,the SAR model and SEM are described in the paper.MLE,GMM and 2SLS estimation methods of these two models can be seen in the article.Secondly,this paper extends the model to the spatial panel model and gives the MLE method and the GMM estimation of the models accordingly.Thirdly,the Monte Carlo simulation of the four spatial panel models are carried out.It is found that the recognition effect of the four models on positive spatial correlation is better than the recognition effect on negative spatial correlation.Also,it is showed that the spatial weight matrix has a great influence on parameters' estimation.So the last chapter of the article proposes a new spatial weight construction method—K-order lag attenuation method,then the article estimates the spatial auto-correlation coefficient with different spatial weight matrix.Simulation result shows that new weight matrix is better for estimation.