On Joint Modelling and Testing for Local and Global Spatial Externalities

This paper concerns the joint modeling, estimation and testing for local and global spatial externalities. Spatial externalities have become in recent years a standard notion of economic research activities in relation to social interactions, spatial spillovers and dependence, etc., and have received an increasing attention by econometricians and applied researchers. While conceptually the principle underlying
the spatial dependence is straightforward, the precise way in which this dependence should be included in a regression model is complex. Following the taxonomy of Anselin (2003, International Regional Science Review 26, 153-166), a general model is proposed, which takes into account jointly local and global externalities in both modelled and unmodelled effects. The proposed model encompasses all the models discussed in Anselin (2003). Robust methods of estimation
and testing are developed based on Gaussian quasi-likelihood. Large and small sample properties of the proposed methods are investigated.