GMM Estimation of Lattice Models Using Panel Data: Application

We propose an empirical application of lattice models to actual household-level data based on the generalized method of moments. We take advantage of the two dimensional structure of panel data to construct a lattice specification. Then, a class of nonparametric, positive semidefinite covariance matrix estimators that allow for a general form of spatial dependence characterized by a metric of economic distance is introduced. This framework is applied to estimating spatial patterns in the residential demand for drinking water. Estimation results indicate that accounting for spatial dependence yields efficient estimate of the asymptotic variance matrix. Compared to non-spatial strategies, spatial dependence implies higher standard errors for all parameter estimates so as to strongly modify patterns of significance.