A0530
Title: GMM estimation of spatial autoregressive models with cluster-dependent errors
Authors: Takaki Sato - Musashi University (Japan) [presenting]
Abstract: The generalized method of moment (GMM) estimation of spatial autoregressive (SAR) models is considered, with unknown cluster correlations among error terms. In the presence of cluster correlations within errors, nonlinear moment conditions suitable for independent errors lose their validity, and GMM estimators obtained from the moment conditions are inconsistent. A GMM estimator obtained from another nonlinear moment condition is proposed, suitable for cluster-dependent error terms, and its asymptotic properties are shown. Because the asymptotic variance of the GMM estimator depends on the choice of a weight matrix for GMM estimation, an optimal weight is also discussed that minimizes the asymptotic variance, and a feasible optimal GMM estimator is introduced based on a consistent estimator of the weight. Monte Carlo experiments indicate that the proposed GMM estimator has a small bias and root mean squared errors when error terms have cluster correlation compared to two-stage least squares estimators and GMM estimators for independent errors.
