Application Of Vector Autoregressive Graph Model In Econometric Field

Nowadays,the rapid development of China's real estate industry has a crucial impact on the national economy.Practice has shown that there is a correlation between the housing prices in various cities,and the housing price changes have a "ripple" effect.Therefore,the study of the linkage between housing prices is particularly crucial to the healthy development of the national economy.In this paper,spatiotemporal chain graph and spatiotemporal autoregressive graph models are firstly constructed,and the linkage relationship between spatiotemporal variables is identified by VAR graph model method.In the graph VAR model,assuming that the autoregressive matrix and inverse covariance matrix of the model are sparse,the Granger causality is represented by digraph,and the Partial correlation of the same period is represented by undigraph.In order to estimate model parameters simultaneously,nets estimation algorithm based on Lasso regularization is used.This algorithm avoids the limitations of traditional methods and establishes conditions for consistent selection and estimation of variable parameters.Space-time autoregressive graph model was applied to 70 large and medium-sized cities in our country house prices linkage analysis,Based on the housing price index data of 70 large and medium-sized cities in China from June 2005 to December 2019,the graph vector autoregressive model was constructed,and in view of the linkage relationship between network characteristics analysis.the block model analysis depict the relationship between China's 70 large and medium cities housing prices market.The results show that the linkage of housing prices in large and medium-sized cities in China presents a remarkable network structure,with good accessibility and high stability,and the housing price fluctuations in first-tier cities and second-tier stronger cities have a great influence,which makes them the center of the network.Beijing,Shanghai,Guangzhou,Zhengzhou and so on have the function of "leading" the housing price,and the fluctuation of the housing price will affect the surrounding cities and then spread,while the third-tier and fourth-tier cities and relatively partial areas,such as Guiyang,Yueyang and Zunyi,the housing price is mainly affected by other cities,which is in the "following" position in the network.And then to block of 70 large and medium-sized cities according to the relationship strength analysis,mainly divided into 13 cities such as Beijing,Shanghai and Zhengzhou to overflow plate,Haikou,Guiyang,Sanya and other 15 cities to benefit from plate,25 cities such as Hangzhou,Nanjing,Tianjin for strong correlation mediation plate,Dali,Tangshan,Jiujiang,etc for the weak link in 17 cities of intermediary plate.Finally,in view of the linkage between housing prices in 70 large and medium-sized cities in China,scientific and reasonable housing price regulation suggestions are put forward.