Constraints Space Set Of Weighting Matrix And Matrix Structure Parameter Estimation

In regional science, geographical science research, and an increasing number of applications of economics, spatial dependence usually modeling by spatial weight matrix, which each element represents a pair of spatial unit of the direction and strength of the spillover effect, and also represents the spatial weight matrix space unit and the diffusion model of interaction, thus spatial weight matrix is a kind of meaningful, more easy to explain, represents the spatial interaction (space) since the correlation of spatial autocorrelation model. The choice of appropriate spatial weight matrix is an important part of modeling, because it assumes that the prior structure space since the correlation. The goal of this article is to set a more accurately express the real space depend on the weight matrix.This paper first introduces the spatial weight matrix in use today is set and comparative analysis, to draw out in this paper, we study the object------in the spatial lag model within the framework of spatial weight matrix structure constraint matrix setting and estimate;To impose constraints on spatial weight matrix of the matrix structure, using the sample distribution estimation method by spatial weight matrix estimation, and gives the specific algorithm to estimate the method, prove the uniqueness of the estimates of the spatial weight matrix, and the monte carlo simulations are done for the estimate of the spatial weight matrix and Boostrap inspection;The use of spatial weight matrix estimation methods of setting and the five northwestern provinces, we made empirical analysis on regional GDP:The innovation point of this essay is:(1) Set of sample constraints spatial weight matrix, the matrix structure of the method to get the spatial lag model is estimated based on the sample distribution parameter estimation.(2) Proved the uniqueness of spatial weight matrix estimation.The estimate of the parameters of spatial weight matrix are given the asymptotic distribution. Esti-mate of the spatial weight matrix and the monte carlo simulation is carried on the Boostrap inspection, stochastic simulation results also show that in this paper, the resulting parameter estimates is very close to the true value, Boostrap test results show that the parameter estimation is feasible.