| The report of the 19 th National Congress put forward “holding the bottom line of systemic financial risks,” and took precautions to resolve major risks as the first of three major challenges,showing the importance of preventing systemic risks and maintaining financial security.Internet finance is an important innovation in the combination of network technology and financial activities.P2 P lending is an important part of it and has developed rapidly both at home and abroad in recent years.Although P2 P financing has the characteristic of “small and decentralized”,which is conducive to reducing its own risk,there may still be a wide range of risk effects due to multiple factors such as moral hazard,adverse selection,and information asymmetry.Under certain conditions,this risk will also have a systematic impact on the traditional financial system.How to prevent the contagion and diffusion of systemic risks among P2 P lending platforms has become a top priority in the development of the industry.Risk prevention involves not only the regulatory authorities,but more importantly the platform itself and the investors.Based on above,a model of contagion and diffusion of systematic risks on P2 P lending platforms is built from perspective of complex network theory.Key node identification,systematic risk measurement,risk contagion path and diffusion mechanism and regulatory of the platform is studied using the methods of data processing,model building,simulation and empirical research,aiming at to provide advices about risk control for regulatory authorities.First,this paper obtains relevant transaction data of 494 P2 P online loan platforms of online loan home by writing a crawler program,statistical indicators are extracted a network diagram of P2 P online loan platforms is built.The macro,medium,and micro statistical eigenvalues and structural characteristics of the network are analyzed by the degree distribution function of the entire network and macro statistical calculating,The study finds that each layer of the network tends to have an association relationship with a node at the core layer,this reflects not only the priority connection characteristics of scale-free networks,but also the Law of Attraction.There are many internal cohesive subgroups,which are closely connected inside and loosely connected outside,reflecting the “herd effect”.Because the average distance(path length)of the network is 1,which is relatively small,and the average aggregation coefficient is 0,which is also very small,the characteristic of small-world networks cannot be fully satisfied,of which the average aggregation coefficient is large and the average path length is short.Second,this paper applies the bipartite network theory to the P2 P lending platform network model.the risk contagion mechanism is explained and diffusion path of the P2 P lending platform market is studied.In regard to the single vertex network,formed by bipartite network and weighted projections,is used to partition the community structure.It is concluded that the bipartite network and single vertex network both have small-world characteristics,and the bipartite network have scale property.However,in debtor networks and creditor networks,only small-world characteristics can be found;the degree distribution of them do not conform to the power-law degree distribution,that is,a single vertex network is not a scale-free network.Results show that the rankings of commonly used indicators(degree centrality,eigenvector centrality,betweenness centrality,closeness centrality)and the analysis of critical point detection,which is based on creditor network,debtor network,and bipartite network,have both consistencies and inconsistencies.It will be more scientific and profound with the identification and comprehensive consideration of edge strength.Third,by constructing the bipartite network model and abstracting the mechanisms of adding edges,removing edges and reconnections between nodes,static evolution model and dynamic evolution model of the loan bipartite network aredesigned respectively,and systemic risk measurement is studied in this paper.Also,regarding the P2 P platform relationship as a complex network,the systemic risk caused by the P2 P platform contagion in the network structure is measured based on the relevant theories of complex network and matrix methods.The empirical results show that,The target shock caused a higher contagion severity,while the random shock caused a lower contagion severity.There is no significant difference between the single shock and systemic shock.The real-world P2 P platform network structure often presents a scale-free bipartite structure,which can better absorb the negative impact of random shocks,and is not sensitive to single shock and systemic shocks.Last,the systemic risk contagion mechanism of P2 P lending platforms is studied and the risk contagion mechanism of the bipartite network is analyzed.Basing on the principles of “too big to fail," “too many connections to be reversed," and “too strong connections to be reversed",important,vulnerable,strongly connected relationships of empirical platforms and loan are identified,contagious path and loss rate settings is analyzed,the different risk spillover effects under single shock and joint shock are showed through empirical analysis.And from the perspective of systemic risk supervision of the P2 P lending platform,the various links of the risk contagion chain analyzed,this paper gives regulatory advice on the supervision of the source,the path and the vulnerable platforms of contagion.Compared with the previous researches,the study of this paper mainly includes the following innovative points:1.Establishment of a P2 P lending platform two-mode networkPrevious researches on financial complex network mainly focused on general complex networks.This paper builds a directed,weighted,scale-free,and two-mode complex network.An empirical network is built using P2 P platform as a node and the complex lending relationship between P2 P platforms as an edge.Through the statistical analysis of network characteristics and the evolution of network structure,the internal potential relationships of the two types of P2 P lending platforms and theintensity of the relationship between different types of platforms are found,providing a new perspective for understanding the risk attributes of P2 P lending platforms.2.Empirical study on bipartite network construction and critical point detectionBased on the empirical bipartite network,the credit network and debt network can be obtained by projection.K-core analysis and community discovery in the complex network are introduced into the P2 P platform risk measurement research,which is helpful to detect critical points.The purpose is to find common properties or differences from multiple platforms,in the medium level,caused by a closer connection,which is more observable and conducive to follow-up researches.It provides references for investors to narrow the scope of choices,and for regulators to work efficiently.3.Evolution algorithms and mechanisms of network structureThis paper studies the evolution mechanism and systemic risk measurement of P2 P lending platform network structure.Based on empirical data,a bipartite network model is constructed to study the evolution mechanism of network structure,summarize static and dynamic evolution algorithms,and adopt matrix method to identify and measure systemic risks.4.Systemic risk measurement,risk contagion mechanism research and regulatory adviceFor the empirical bipartite network model,this paper analyzed the system-based risk measurement,the risk contagion path and diffusion mechanism using the matrix method.On the basis of systemic risk measurement and contagion path analysis,this paper analyzes each link of the contagion chain(the important node is the source of contagion,the vulnerable node is the target of contagion,and the strong connection between lending and borrowing is the route of transmission.)and gives specific regulatory advice. |
PDF Full Text Request