A Pairwise Difference Estimator for Partially Linear Spatial Autoregressive Models

<title>Abstract</title>We propose a pairwise difference estimator for partially linear spatial autoregressive models with heteroscedastic or/and spatially correlated error terms. In comparison with other competing estimators, e.g. the profile QMLE (Su & Jin, 2010) and the semiparametric GMM estimator (Su, 2012), our estimator has the advantage of computational simplicity particularly when one is interested in estimating the finite dimensional parameters in the model. Large sample properties of the estimator are formally established and a consistent estimate of the asymptotic CV matrix is provided. We then use the method to robustly estimate the effect of strategic interaction in deciding local school spending.