The Spatial Econometric Model And Its Empirical Study Of Financial Risk Contagion Based On The Directed And Asymmetric Information

With the continuous development of global economic integration,diversification and liberalization,and the rapid development of internet technology and computer technology,the links among regions,economies and financial markets have become more and more close.And the spatial effects among them have also been becoming increasingly significant.This largely increases the probability of financial risk contagion.At the same time,the over-virtualization of financial markets seriously impacts the real economy,which is its foundation.This makes the financial risk contagion show some newly multidimensional and mixed features.On the other hand,In recent years,global financial events have occurred frequently.Especially the three black swan events in 2016,which have added many uncertainties to the global economic situation and political environment,and have caused extreme panic among market investors and fierce fluctuations in global financial markets.Therefore,the research conducted on the multidimensional and mixed spatial effects and propagation channels of financial risks among financial markets and the real economies,to a certain extent,has important practical and theoretical significance to market investors,policy makers,financial experts and related academic personnel,as well as portfolio management and risk assessment.However,the traditional econometric model ignores the ubiquitous spatial effects among variables,which is obviously inconsistent with social reality and may cause a certain degree of bias in the estimation results.While the spatial econometric method introduces the information of the spatial structure of variables into models by constructing the spatial weight matrix as the carrier of spatial effects,and then constructs the spatial econometric models,which reshapes the analytical framework of econometrics.And the spatial econometrics method has now become the mainstream branch of econometrics and a research hotspot.While the newly multidimensional and mixed characteristics of financial data may,to a certain extent,cause loss of information on the spatial effects of financial data obtained from the traditionally undirected and symmetric spatial weight matrix constructed based on “distance” and irrelevance of directionality.Therefore,how to construct a spatial weight matrix based on the objective information between economic units in the current economic situation to describe the spatial correlation structure between them as reasonably and fully as possible,and then construct a spatial econometric model is the core issue of the current econometric theory.Financial market is a complex and non-linear system.The interactions among financial systems have a certain degree of difference.That is,the mutual influence between them is asymmetric.And the transfer entropy method in information theory not only can capture the information between nonlinear systems,but also can capture the direction of information flows.So the characteristic of transfer entropy can to a certain extent fit very well with the essence of financial markets.And recently some scholars have applied it to the study of financial issues.Based on this,this dissertation firstly uses the transfer entropy method to introduce the weighted information between two economic units into the GARCH model to improve the traditional econometric model,and then constructs the the directed and asymmetric information(DAI)economic measure model to measure the interrelationships among financial markets and to verify the effectiveness of the transfer entropy method.Then,based on the verification of validity,the transfer entropy method is extended to the application of multiple economic units,i.e.,it is used to construct the DAI economic spatial weight matrix.Afterwards,based on this matrix,this dissertation constructes a series of spatial econometric models to capture and analyze the multi-dimensional,and mixed spatial effects of the first and second moment of financial market returns between cross-economy and-market of financial risk contagion.And this dissertation comboines the complex network theory to construct the high-order information spatial econometric model to analyze the contagion mechanisms and paths that financial risk contagions among financial markets and between the financial markets and real economy to deeply unearth the intrinsic laws and operation mechanisms of the spatial spillover of financial risks.The content of this article includes the following four parts:(1)This dissertation introduces the time-varying information weight based on two economic units by using the advantages of capturing information of the transfer entropy menthod to improve the classic financial risk contagion GARCH model,and then constructs the time-varying symbolic transfer entropy GARCH model.Based on this model,this dissertation studies the correlation changes between the stock markets and bond markets of nine major economies in the world of the impact of the Brexit.The results show that the estimation accuracy of the improved model has been improved at a certain degree,which ondicates that the improved GARCH model can more effectively capture the correlation information among financial markets to a certain extent.That is,the transfer entropy method proposed in this dissertation is valid,which can better capture the information among financial markets to some extent.It is further illustrated that transfer entropy method has certain practical significance in the application of practical problems in the financial field.(2)This dissertation expands the information weights of the two economic units introduced by the transfer entropy method to multiple economic units to improve the construction method of the traditionally symmetric spatial weight matrix in the spatial econometric theory.In this dissertation,a DAI spatial weight matrix is built to improve the ability to capture the spatial effects of financial risk contagion.Based on this,this dissertation constructs the panel data Spatial-SUR model.And nder the background of the black swan events in 2016,we study the long-term and time-varying spatial effects patial spillover effects of fianncial risk contagion of the first order moment of sub-financial markets returns in China.The results show that compared to the traditional spatial weight matrix,the estimation accuracy of the model based on the newly proposed DAI spatial weight matrix has been improved to some degree.In particular,through the study of dynamic spatial effects,it is found that the spatial effects of classical models cannot be analyzed under the condition that the spatial effects of classical models are not significant.While the improved Spatial-SUR model can capture more information and have a significant spatial effect.This reflects the advantages of the improved model and extends its scope of transfer entropy menthod.(3)Given the first moment DAI financial risk contagious Spatial-SUR model can not analyze the second moment information of fianncial risk contagion.In order to solve this problem,this dissertation uses the spatial econometrics method to improve the traditionally multivariate volatility model,and then constructs the second moment DAI financial risk contagion multivariate Spatial-BEKK-GARCH model.And then capture the spatial effects of the financial risk contagion between the second moment information in the financial markets,and mitigate the common dimension disaster problem in the traditional multivariate volatility model.With the background of the downgrades of the sovereign credit rating of Europe,this dissertation empirically studies the spatial volatility spillover effectS between the sencond moment information of the PIIGS.The results show that compared with the traditional spatial weight matrix,the model constructed based on the DAI spatial weight matrix captures more information,the estimation accuracy of the model is further improved,and with the improved model,the spatial effects of financial risk contagion among stock market volatilities can be analyzed.And also show that the structured process method adopted in this dissertation can effectively alleviate the common dimension disaster problem in the multivariate volatility model.(4)The development of electronic and informational methods of trading has provided us with a great deal of data and information.In order to make full use of these data and information,this dissertation uses the information theory and complex network theory to propose a method for constructing a time series complex network with direction and weights to effectively process large amounts of data.And then by combining the advantages of spatial econometrics,we construct a high-order information spatial econometric model based on the network attributes,which can analyze the spatial effects of financial risk contagion in the context of big data.In the process of modeling,this dissertation selects appropriate network attribute indicators to replace the original financial data and carries out parameter estimation to examine the contagion channel and spatial effects between the global financial sector and the financial sectors of various economies,and those of the financial sector impacting on the real industry sectors.The empirical results the new model can analyze the characteristics of financial risk contagion in a large amount of data.And compared with the classical econometric model,the accuracy of the high-order information spatial econometric model has been improved to some extent.