Testing for spatial autocorrelation in simultaneous equation models

Abstract

In spatial econometrics, several tests for spatial autocorrelation in single-equation regression models have been developed. Simultaneous equation models are often estimated using spatial, cross-sectional data, yet little attention has been paid to spatial autocorrelation problems in such models. The sole discussion has been by Anselin and Kelejian (1997). This paper constructs an alternative spatial autocorrelation test for models estimated by two-stage leastsquares, and builds closely on the work of Harvey and Phillips (1980) for time-series. The test requires only one auxiliary regression and can then be applied in the same way as the standard Moran/Cliff-Ord test, and takes account of the finite-sample impact of the exogenous variables on the distribution of the test. It is also argued that these tests are only the first stage in a much needed interfacing of simultaneous equations modelling and spatial econometrics.