Omitted Factors and Spatial lags in the Dependent Variable
It’s been suggested a number of times that the significance of the estimated coefficient of the spatial lag of the dependent variable may be the result of an omitted common factor which may be spatially correlated-see, among others, (Gibbons and Overman in J Reg Sci 52:172–191 2012 and Corrado and Fingleton in J Reg Sci 52(2):210–239 2012). The concern is important because this parameter is often of primary interest. The purpose of this paper is to consider this issue in a formal context. We give conditions under which the estimator of that parameter is consistent and asymptotically normal, and when it is not. We consider two cases. In one case the omitted factor is exogenous, and in the other it is endogenous. We give results for both cases.