A penalized quasi-maximum likelihood method for variable selection in the spatial autoregressive model

This paper investigates variable selection in the spatial autoregressive model with independent and identical distributed errors. A penalized quasi-maximum likelihood method is developed for simultaneous model selection and parameter estimation. Under some regular conditions, theoretical properties of the proposed estimators, including consistency and the oracle property, are established. In addition, a computationally feasible algorithm is designed to realize variable selection procedure. Simulation studies are conducted to examine the finite sample performance of the proposed method and a real example about Boston housing data is presented for illustration purpose.