Spatial Measurement Model Variable Selection Method And Its Application

Spatial econometrics,as the main method of analyzing spatial economic data,has become an important part of econometric analysis after thirty years of development.Variable selection method has been a hot issue in econometrics research.The development of their own trajectory,rarely cross.However,due to the introduction of spatial correlation,the Gauss-Markov hypothesis is no longer valid,and the variable selection based on the classical linear model is no longer applicable.However,with the rapid development of information technology,therefore,how to choose the explanatory variables and construct the optimal model is an urgent problem in the process of spatial measurement modeling.In this paper,the variable selection method is introduced into the spatial measurement model,and a series of variable selection methods of spatial measurement model are proposed and validated,and the results of theoretical research are applied to the selection of financial factors affecting stock return.Based on the spatial autoregressive model(SAR model),this paper first uses the KL information and the Bayesian method to estimate the AIC criterion and the BIC criterion of the classical linear model under the condition that the model residuals obey the normal distribution.(SAIC criterion)and spatial BIC criterion(SBIC criterion)based on the SAR model are presented and demonstrated,and it is proved that the SAIC criterion and the SBIC criterion based on the SAR model under the condition of their variables selection.Secondly,the above method is extended to a more generalized spatial econometric model,the Spatial Autoregressive Model with Autoregressive Disturbance(SARAR model),and it is proved that under certain conditions,the SARIC criterion and SBIC criterion based on SARAR model Variable selection is consistent.In this paper,the residual assumptions are relaxed and the generalized spatial information criterion(SGIC criterion)is constructed based on the SARAR model based on the assumption that the residuals are independent and identically distributed.The SAIC and SBIC guidelines are integrated into the unified analysis framework.Spatial information criterion is divided into spatial AIC class criterion and spatial BIC class criterion.In the theoretical derivation,this paper also designs a computer simulation experiment.Monte Carlo simulation is used to study the finite sample properties of the variable selection of spatial model.It is found that for the spatial data,compared with the classical linear model,the variable selection method The proposed method is more effective in variable selection of spatial model.In the empirical part,this paper applies the variable selection method of spatial measurement model,which is theoretically researched to the research of financial index variable selection,which affects the stock return rate.Secondly,this paper uses the Moran's I test and the spatial autoregressive model without the explanatory variables to calculate the spatial effect of the stock return rate,and finds that the stock price of the stock market in our country is much higher than that of the stock market.We can find out the financial indicators,which reflect the profitability and development ability of the company.We can see the financial indexes,which influence the return of the stock to the stock.The result shows that,the financial index,which reflects the ability of solvency of the company,is the second,and the financial index,which reflects the operating ability of the company,is the smallest.Finally,this paper carries on the robustness test and finds that the conclusion of this paper is robust...