A0746
Title: Adjusted MLE for the spatial autoregressive parameter
Authors: Federico Martellosio - University of Surrey (United Kingdom) [presenting]
Abstract: The quasi-maximum likelihood estimator (QMLE) of the autoregressive parameter in a spatial autoregression can suffer from substantial bias. Part of this bias is due to the presence of other parameters in the model, and a simple approach to reduce the impact of nuisance parameters is to recenter the profile score. We study properties of the resulting estimator, named the adjusted QMLE. Despite being first-order asymptotically equivalent, the QMLE and the adjusted QMLE estimators behave quite differently in many cases of practical interest. In particular, we show that the supports of their distributions may be different, which implies that care needs to be taken when comparing the two estimators.
