1 NONPARAMETRIC ESTIMATION OF THE SPATIAL CONNECTIVITY MATRIX BY THE METHOD OF MOMENTS USING SPATIAL PANEL DATA

We use moments from the covariance matrix for spatial panel data to estimate the parameters of the SAR model, including the spatial connectivity matrix W. In the unrestricted SAR model the parameters are exactly identified by the moments in the covariance matrix. The restricted SAR model is over-identified, but we suggest that its parameters may be estimated by GMM and its restrictions may be tested empirically. If Gini covariances are used instead of conventional (Pearson) covariances, the estimated W matrix is asymmetric. We also suggest that the reduced rank covariance matrix will estimate W and the SAR parameters more efficiently