Testing for a structural break in the weight matrix of the spatial error or spatial lag model

Testing for a structural break in the weight matrix of the spatial error or spatial lag model. Spatial Economic Analysis. This paper studies the weight matrix in the two generic spatial econometric models, allowing the friction-of-distance parameter to be freely estimated by (Gaussian) maximum likelihood. It shows that in samples of moderate size it is perfectly feasible to estimate both the distance-decay coefficient that shapes the weights, and the usual SEM/SLM coefficient, and to test for a change in the former at an unknown change point. The procedure is easily extended to test for changes in other parameters. Tests of empirical size and power are investigated by Monte Carlo experiment, and the test is applied to a model of exchange rate pass through, with clear results.