Robust test for spatial error model:considering changes of spatial layouts and distribution misspecification

This paper suggests a robust LM (Lagrange Multiplier) test for spatial error model which not only reduces the influence of spatial lag dependence immensely, but also presents robust to changes of spatial layouts and distribution misspecification. Monte Carlo simulation results imply that existing LM tests have serious size and power distortion with the presence of spatial lag dependence, group interaction or non-normal distribution, but the robust LM test of this paper shows well performance.