Monte Carlo properties of spatial dependence LM tests

Abstract

This paper compares the power and size of both spatial one-directional and robust to local misspecification LM test using Monte Carlo techniques under different set ups. Confirming the results found in Anselin et al . (1996), under local misspecification the robust LM test improves its size and power of tests. However, under no misspecification, the robust LM test is severely penalized and the usual one-directional LM test has a much better performance.

Acknowledgement

We are grateful to Luc Anselin for his comments.