Simple Regression Based Tests for Spatial Dependence

We propose two simple diagnostic tests for spatial error autocorrelation and spatial lag dependence. The idea is to reformulate the testing problem such that the test statistics are asymptotically equivalent to the familiar LM test statistics. Specically, our version of the test is based on a simple auxiliary regression and an ordinary regression t-statistic can be used to test for spatial autocorrelation and lag dependence. We also propose a variant of the test that is robust to heteroskedasticity. This approach gives practitioners an easy to implement and robust alternative to existing tests. Monte Carlo studies show that our variants of the spatial LM tests possess comparable size and power properties even in small samples.