Research On The Identification And Early Warning Of The Financial Contagion Pattern In The Global Stock Market

Financial contagion is one of the most important factors that disrupt the economic order and undermine the structure of the economic and financial system.Considering the rapid spread and wide impact of financial risks,they may cause great damage to the financial system at the country level or even the global level.Moreover,considering the global economic situation is getting more complex and volatile,it is of great significance to propose more targeted preventions to ensure the stability of the financial system.Such a context places higher demands on the study of the dynamics of the financial risks propagations(FRPs): on the one hand,there is a need to accurately identify the processes of the FRPs with different patterns;on the other hand,there is a need to timely and accurately predict the financial risk propagation patterns(FRPPs)that may occur in the future.However,there are some limitations in the related researches.To solve this problem,this thesis uses the global stock market as a sample and adopts the hybrid and multidiscipline approaches to identify the FRPPs to distinguish different types of the processes of the FRPs,and constructs a data-driven early warning index(EWI)based on the correlations between topological features of finnacial networks and the time-varying features of the propagation patterns.The contribution and innovation of this thesis are reflected in the following areas.(1)A dynamic financial contagion model(DFCM)of the global stock market has been constructed and the dynamic propagation features of financial risks have been explored.Based on the integration of BEKK-GARCH,Delta-CoVaR and SIS-Cascade models,the direction,degree and mechanism of financial risks are clarified,and the DFCM of the global stock market is proposed.Using this model,a dynamic simulation of the FRP is realized,which quantifies the speed and scope of the FRP.The study found that the FRP has a critical effect: the general volatility is easier to propagate than the extreme volatility;In addition,the critical value of the FRP has a time-varying feature: the extreme volatility is easier to propagate when the market is at risk than stable;Moreover,the topological properties of the financial network have stronger impact on the FRP than the systematic risk properties of the global stock market.(2)A quantitative approach based on financial risk response time calculation(FRRTC)was used to identify the FRPPs.Based on the simulation results of the DFCM,three FRPPs are identified on the basis of a combination of the downscaling,clustering,FRRTC and association rule analysis methods.It was found that the dynamic features,network structure features,overall risk features and time distribution features of the three patterns were different.Among them,the composite propagation(CP)has the highest risk level,a larger network size,and corresponds to spike and tail truncation features,which occurred frequently after 2007;the degree-limited propagation(DeLP)has the second highest risk level,a higher network centrality,and corresponds to dwarf-peak features,which were concentrated before 2007;the distance-limited propagation(DisLP)has the lowest risk level,a longer network diameter,and corresponds to flat and weak features,which occurred more evenly.(3)An early warning model based on the motif transformation matrix(MTM)was constructed for the three propagation patterns,and the effectiveness of the early warning model was tested.Since the network structure corresponding to different propagation patterns is found to be quite different,this thesis constructs an EWI from the perspective of analyzing network topology changes.Further studies have found that the responses of network motifs are more sensitive when the network structure changes,therefore,this thesis constructs an EWI for the three propagation patterns based on the statistics of the evolution of the motif transformations.It was found that the EWI proposed in this thesis is valid and robust.Among them,the T-M index has a better warning effect on the CP,with an advance warning period of 9 working days;the M12.M12 index has a better warning effect on the DeLP,with an advance warning period of more than 200 working days;the M5.M5 index has a better warning effect on the DisLP,with an advance warning period of more than 40 working days.