Research On Credit Behavior Evolution And Infection Effcet Of CCSME Based On Agent-based Computational Finance

With the development of economic globalization, SMEs in the global economy has become the most active part. SMEs not only support the progress and development of the world economy, but also play an irreplaceable role in the promotion of global innovation. However, obtaining financial resources of SMEs do not match its contribution, therefore, the financing of SMEs has been a study that the world are actively explored and researched on. China is in a critical historical period of strategic transformation, economic development has also entered into the "new normal" era. Although the government and financial institutions have introduced a number of policies and measures in order to improve the financing capacity of SMEs. However, the weak business and other aspects of risk management and control of SMEs, coupled with the financial crisis and the aftermath impact of the macroeconomic environment, results in the difficulty of SMEs in financing. In order to alleviate the financing problems of SMEs, Credit Community of Small and Medium-sized Enterprises(CCSME) financing models come into being. Through horizontal supervision, incentive and punishment, and guarantee mechanisms, CCSME financing model effectively reduces the information asymmetry between enterprises and banks. CCSME financing model not only can reduce financing costs for SMEs, but also provides quality projects for banks, which achieves a better win-win situation. The financing business of CCSME is carried out all around China. However, meanwhile the default events of CCSME are also frequently seen in newspapers. Why CCSME changes from the "safety lock" of prevention and control of credit risk into the "fuse" of credit risk infection, has become a hot issue that academia and industry concern.So far, many scholars have launched a series of meaningful explorations from the theoretical and empirical levels of CCSME, but most of the existing researches are limited to the analysis of farmers and other individual borrowers guaranteed loans’ problem. For the discussion of complex CCSME financing problem, the research is relatively rare and focuses too much on the analysis of feasibility. Due to the overall credit risk of CCSME arises mainly from its internal members’ credit behavior interaction effects between members. Therefore, internal members’ credit behavior interaction effects are the key to the overall credit risk of CCSME, while the literatures in this area are extremely rare. Based on this, this dissertation carries out in-depth studies on these issues, in order to explore the evolution of credit behavior and interaction mechanisms of the CCSME members. On that basis, the contagion effect of related credit risk, as well as the impact mechanism of related structure and size of CCSME and CCSME size on the overall credit risk is studied. This dissertation not only characterizes the internal members’ credit behavior and related credit risk, but also provides important theoretical basis for effectively reducing the overall credit risk and improving CCSME models.The main innovation and feature of this dissertation is that the research is closely connected with the actual situation of CCSME, and carried out especially on the mechanism of evolution of related credit risk and control measures of CCSME. From the perspective of behavioral science and complex systems, by using the combination methods of theoretical analysis and simulation, the internal members’ credit behavior and interaction and the evolution of related credit risk of CCSME members is deeply discussed from micro and macro dimensions. Finally, the corresponding suggestions and countermeasures for prevention and control of the credit risk outbreak of CCSME are provided. Specific innovations are as follows:First, from the perspective of behavioral science, the evolution mechanism of CCSME credit behavior is discussed. At present, the researches on the credit risk of CCSME always ignore the interaction and impact among member enterprises, and little attention is paid to the member enterprises’ bounded rationality of credit behaviors. Based on this, an evolutionary game model of CCSME members’ credit behavior is built. Meanwhile, the interaction and mutual influence between CCSME members is structure described and set. Further, member enterprises’ strategy selection preferences and beliefs in-depth are characterized, through Prospect Theory(PT) and Experience Weighted Atrraction(EWA) learning model. In that way, this dissertation discusses the mechanism of impact of CCSME credit behavior on the overall credit risk of CCSME under conditions of bounded rationality.Second, from the perspective of complex systems, the mechanism of evolution of related credit risk of CCSME is studies in this dissertation. Given the evolution of related credit risk of CCSME is a typical complex adaptive system. Therefore, the complex systems theory and research methods to study this issue in this dissertation are drawn on. Through using Cellular Automata(CA) model, the mechanism and process of the credit risk infection of CCSME is studied. Further, by embedding CA model into Fitness Landscape Theory(FLT) to characterize the interaction behavior of member enterprises of CCSME, CANK model is built for analyzing the impact of CCSME size and related structure on the overall credit risk of CCSME. Finally, based on complex network model of CCSME, elimination mechanisms paly a promoting role of improving the overall credit level of credit community is discussed, by introducing the extreme dynamics(BS) model.Third, the Agent-based(AB) modeling and Agent-based Computational Finance(ACF) methods are introduced into the researching of the evolution of members’ credit behavior and contagion effect of related credit risk of CCSME. From the interaction and influence of CCSME members, adopting a bottom-up modeling approach, the credit behavior of CCSME members is designed to construct a theoretical model and simulation. This method not only effectively overcomes the shortcomings of traditional methods, but also achieves complement with the traditional method and improves the ability to analyze complex financial system. Additional, Combination of theoretical models and build complex models are tried in this dissertation, such as: in Chapter VI, a composite CA model is constructed for a more realistic characterization of the related credit risk infection; while in chapter VII, the CANK model takes the advantages and characteristics of both two models. Combining with the simulation model organic, it can help to take the advantages and avoid the disadvantages.Therefore, the research of combining model and simulation to characterize and simulate complex systems, is innovative and meaningful.