Research On The Risk Contagion And Control Between Firms And Banks Based On The Multiple Credit Networks

The modern financial system is characterized by a high level of connectivity. Thus, one agent’s default may lead to additional agents defaulting who have direct or indirect links with it; this is known as the domino phenomenon. As what the scholars’research shown, a network can be formed based on the links between agents. In such a network structure, one agent’s default is dependant not only on its own financial condition, but also those of neighboring agents who have links (such as lending-borrowing relations) with it. The neighboring agents’financial conditions are also relevant to the financial conditions of their own neighboring agents and so on. Similarly, one agent’s default not only negatively affects the net worth of its neighboring agents who have direct links with it, but also makes the negative effective of the default propagated to its neighboring agents’neighboring agents through the impact on its neighboring agents resulting from its default. Firms and banks are two kinds of important agents; the links between the two kinds of agents are getting increasingly connected. The firm agent’s default will result in bad debt. If the bank agent cannot absorb the bad debt resulting from the defaulted firm agents, it will default. As an extremely important economic agent, the default of the bank agent will bring in an extremely negative effect. Thus, it is necessary to research the risk contagion between firm and bank agents. Based on this point, inspired by the existing research, a model of endogenous multiple credit networks of firm-bank agents is established and the research on risk contagion and control is done based on the established model.First, a model of endogenous multiple credit networks integrating firm-bank agents is constructed based on the trade credit link between firm agents, the bank loan link between firm-bank agents and the interbank credit link between bank agents and the amount of the credit demand and the establishment of the link between agents are both endogenous. By means of simulations, the model is capable of showing some stylistic features existing in the real world:the firm size distribution can be well fitted with a power-law when the value of the size is above a specific threshold; the bank size distribution is lognormal with a power-law tail; the bank in-degrees of the interbank credit network as well as the firm-bank credit network fall into two-power-law distributions.Second, the risk contagion between firm-bank agents is respectively researched from the perspective of the credit network structure and the perspective of the agent behavior based on the model of endogenous multiple credit networks. For the credit network structure, the impact of the changes of the number M of the randomly selected partners and the parameter λ of the probability of switching to a new partner on the risk contagion between the firm-bank agents is researched. The research shows M and λ have different mechanisms of action for the agents and the research variables (cumulative bankruptcy numbers of the bank, upstream and downstream firm agents, aggregate output, mean values and standard deviations of bank credit and bad debt) take on different change paths with the changes of these two parameters. However, for the larger values of these two parameters, their changes will both lead to larger values of the corresponding research variables, larger risk contagion and aggregate output obtained because of the increasing economic fragility. Furthermore, the impact of the agent behavior on the risk contagion is respectively researched from the perspective of firm agent behavior and the perspective of bank agent behavior. For the firm agent behavior, the research shows the related research variables have different change tendencies with the increases of the firm output parameters φ and β. The larger values of φ produce larger values of the research variables; while the larger values of β lead to smaller values of the research variables (except mean values of bank credit). For the bank agent behavior, with the increase of the risk aversion coefficient Ψ, the cumulative bankruptcy numbers of the bank agents take on an upward tendency, while those of the upstream firm agents take on a downward tendency after fluctuating at high levels and those of the downstream firm agents take on an upward tendency after fluctuating at low levels. With the constant increase of the up limit of the dividend rate d, the cumulative bankruptcy numbers of the bank agents show an obvious upward tendency, while those of both downstream and upstream firm agents present fluctuation states with no clear tendencies. Higher values of the lower limit vmin of the deposit ratio produce larger cumulative bankruptcy numbers for both bank and downstream firm agents and smaller ones for the upstream firm agents, but the impact is limited, i.e. the cumulative bankruptcy numbers show a tendency toward stabilization when vmin is above some threshold. With the constant increase of the comprehensive parameter α, the cumulative bankruptcy numbers of the bank agents show an increasing tendency, those of the upstream firm agents first increase and then decrease, while those of the downstream firm agents fist increase obviously and then slowly. The changes of the above parameters also have different impacts on other research variables.Finally, the risk control based on the model of endogenous multiple credit networks is researched. The research shows increasing the deposit-reserve ratio is not a better method to control the risk contagion between the firm-bank agents. The increase of the deposit-reserve ratio limits the risky investment ratio, which incurs the decreasing defaults of bank agents. However, the credit supply of banking system is decreased because of the increase of the deposit-reserve ratio which may result in the increase of the cumulative bankruptcy numbers of agents because of the shortage of the liquidity. At the meantime, the macroeconomic output does not necessarily increase. While the default of the important agent such as the one with the largest net worth and the one with the biggest in-degree has a larger negative impact on the related agents in the artificial firm-bank system. Specifically, for both of bank and downstream firm agents, the cumulative bankruptcy numbers of the agents resulting from the default of the agent with the largest net worth are higher compared with the situation incurred by the default of the agent with the biggest in-degree, but the opposite situation appears for the upstream firm agents. This shows that the measurements that only prevent the defaults of the "too big to fail" bank agents are not enough; the bank agents with bigger in-degrees should also be noticed. Furthermore, the agents with bigger net worth and in-degrees at the meantime should be paid more attention. The research also shows that the debt-transparency benefits the decrease of the contagion between the firm-bank agents; the liquidity provision by the central bank enhances the stabilization of the firm-bank system with the increase of the aggregate output. Besides, different levels of the bank agent’s participation have the similar impacts on the contagion in the firm-bank system.In sum, a model of endogenous multiple credit networks of firm-bank agents is established based on the links between agents, the risk contagion of the firm-bank agents is respectively researched from the perspective of the credit network structure and the perspective of the agent behavior based on the constructed model and the risk control under the network perspective is also researched. This research studies deeply and systematically risk contagion and control between the firm-bank agents in theory. It provides feasible models and methods for the supervisors to refer to. Thereby it has positive theoretical values and practical significances.