Small Sample Properties of Maximum Likelihood Versus Generalized Method of Moments Based Tests for Spatially Autocorrelated Errors

By Peter H. Egger, Mario Larch, Michael Pfaffermayr, Janette F. Walde. Social Science Research Network, 2009

70 citations recorded by the source corpus.

This paper undertakes a Monte Carlo study to compare MLE-based and GMM-based tests regarding the spatial autocorrelation coefficient of the error term in a Cliff and Ord type model. The main finding is that a Wald-test based on GMM estimation as derived by Kelejian and Prucha (2005a) performs surprisingly well. Our Monte Carlo study indicates that the GMM Wald-test is correctly sized even in small samples and exhibits the same power as their MLE-based counterparts. Since GMM estimates are much easier to calculate, the GMM Wald-test is recommended for applied researches.

Subfields: Economics and Econometrics

View Paper: https://www.semanticscholar.org/paper/678d309c5cbef73ce811cc011f94f5a18d088690