URAG Working Paper #03-03
An Alternative Estimator for Fixed Effects Spatial
Models (pdf)
by Mary G. McGarvey and Mary Beth
Walker, August 2003
Keywords: spatial correlation; fixed effect model; spatial correlation
consistent covariance estimation; monte carlo evidence
This paper examines inference in finite samples based on estimation using
maximum likelihood and ordinary least-squares estimators in fixed effect
models with spatially autocorrelated errors. The focus of the Monte Carlo
experiments is on the effect of covariance misspecification on finite
sample inference. We find that the maximum likelihood estimator of the
conditional mean parameters is more efficient than OLS under most populations
but the quasi-t statistics are most reliable using OLS with a spatial
correlation consistent covariance estimator.
Comments and questions regarding this paper may be directed to Mary
Beth Walker.
|