Research On Setting Method And Application Of Spatio-temporal Weights Matrix

In recent years,Spatial econometrics has developed from econometrics into an independent branch,and has received more and more attention in the economic field.When spatial econometrics modeling and analysis,one of the most important steps is to build a spatial weight matrix,which is also a hot topic of spatial econometrics research.The spatial weight matrix represents the mutual relation between individual regions or economic variables in the spatial section element,and it is an important bridge linking the theoretical spatial econometric model with actual life,and the suitability of the spatial weight matrix will directly affect the final estimation result of the model.Therefore,it is necessary to construct a proper spatial weight matrix.The paper first combed the development of the spatial weight matrix in detail.Then summarizes the methods of testing spatial correlation in spatial econometric modeling and analysis,as well as the characteristics and classification of spatial econometric models.Then introduced the general definition of spatial weight matrix and some common matrix construction methods.Then,the paper constructs a method for setting the spatio-temporal weights matrix.And the Markov chain Monte Carlo(MCMC)method is used to simulate and compare different spatial temporal weight matrices.The results show that the spatial temporal weight matrix is more suitable.Finally,the static space durbin model and dynamic space durbin model with double fixed effects of space and time are used to study the impact of innovation drive on China's regional economic development.Make the model conclusion more in line with the actual situation.The possible innovations of the paper are as follows: First,the annual global Geary 's C index ratio is used to construct time weight matrix.It is used to represent the dynamic change of space spillover effect in time.Then combined with the spatial weight matrix through the Kronecker product to obtain the spatio-temporal weights matrix.Second,according to the data generation process of different spatial econometrics models,This paper uses Markov chain Monte Carlo method to simulate the spatial econometric model under different spatial temporal weight matrices.It gets the posterior model probability.Then select the appropriate spatial temporal weight matrix according to the posterior model probability.