Empirical Study Of Commercial Bank Risk Contagion Effect In China

The Third Plenary Session of the 18th CPC Central Committee in 2013 made systematic deployment to deepen reforms comprehensively. In the "2014 Government Work Report", Premier Li stressed the importance of the comprehensive deepening reform. He proposed to focus on economic reform and deepen financial reform. People’s Bank of China once again pointed out that in 2015 PBOC will continue to take measures, such as implement a prudent monetary policy, strengthen and improve macro-prudential management, and use various tools portfolio flexibly, to maintain adequate liquidity in the banking system, to guide moderate growth of monetary credit and social financing scale, and to take comprehensive measures to avoid regional systemic financial risks. All these signals clearly show the determination and perseverance of our country to promote the financial reform, and the emphasis on the prevention of financial risks.Currently, under the background that the recovery of global economic slows down and the international financial environment experiences ongoing turmoil, China is facing the challenge of keeping away financial risks. Statistics released by the Chinese Banking Regulatory Commission show that, non-performing loan ratio of banking financial institutions at the end of 2014 is 1.64%, non-performing loan ratio of commercial banks at the end of 2014 is 1.29%, both figures are higher than those of 2013, so financial regulators and commercial banks should keep high vigilance to the outbreaks of group risk. In recent years, our country continues to promote the process of financial marketization, the bankruptcy law of commercial banks is being promulgated. At that time, the commercial banks will bear the consequences of bankruptcies on their own due to the failure in risk prevention. Therefore, the main task of the regulatory authorities is to maintain the ecological stability of the entire banking system, and the country is no longer "pay" for the risks of every commercial bank.At the forthcoming of the bankruptcy law, our commercial banks should put more attention to the risk of default, which can push them to the brink of bankruptcy at any time. But few scholars did research on the default behavior of commercial banks, most of them only focused on the default behavior of lenders. For this dissertation, the author uses the theory of backward stochastic differential equations (BSDE) to construct the commercial bank’s default probability model, using stochastic dynamic Copula Methods to portray the infectious feature of risk between commercial banks, including the default behaviors and the infectious problems into the research framework of risks, using empirical analysis to depict the behavior of default and risk infection, and exploring the root causes of the phenomenon.The logical starting point is:Researches which study the default behavior of lenders have become common place, and few researches covered the default behavior of commercial banks themselves. The main purpose of this dissertation is to discuss the issue of commercial banks in default risk and infectious default risk, and therefore it would be developed in accordance with the technical route of "Phenomenon Description-Literature Review-Model-Case Study". The author builds the research foundation by characterizing the probability of default by commercial banks, and regards the probability of default as a commercial bank risk indicator. By investigating the individual probability of default, the joint probability of default, and the conditional probability of default, the author found that Knight uncertainty is an important factor in the probability of default by commercial banks. Meanwhile, this dissertation determines infectious problems of the risk between commercial banks from the perspective of nonlinear dynamic study, pointing out that stochastic dynamic Copula model not only can solve the deficiencies of the existing methods, but also dynamically characterize the degree of correlation among the risk of commercial banks. To undertake the construction of earlier theoretical models, namely default probability models and stochastic dynamic Copula models, the author did some empirical analysis about the infection of the risk of our 14 listed commercial banks, and made judgment about our systemically important banks.Method logical starting point:This paper regards the measure of the probability of default of commercial banks as a starting point, using theory of BSDE to get the analytic solution expressions about the maximum probability of default and the minimum probability of default under the condition of Knight uncertainty. Based on the two-dimensional normal distribution assumption, we extend a single individual probability of default to a two-dimensional framework of the joint probability of default, and obtain analytic solutions expressions of the conditional probability of default, then, we use numerical simulation analysis of the factors that affect the probability of default to do some simulation study.Because the strong assumption of two-dimensional normal distribution is difficult to meet, and you cannot measure non-linear relationships between variables, the author uses Copula method to improve these issues. And in order to describe the dependent structure with the characteristics of changing with time, we use stochastic dynamic Copula Function to depict it. Finally, the data of 14 listed commercial banks are used in empirical research on commercial bank risk contagion problems, and respond to risk indicators selection, model selection and interpretation of practical questions.The basic structure of the complete text is as follows:(A) Literature review. According to different research objects, the author sorts and summarizes some related literature about risk measurement, risk infectivity, BSDE theory, and Copula modeling. We found some existing deficiencies in the current study:(1) the risk measurement process ignores the impact of uncertainty; (2) Copula modeling process ignores the time-varying characteristics of dependent parameters.(B) Risk measurement of commercial banks and research of affecting factors. In the context of the literature review, we undertake the analysis of theoretical research. Firstly, we use debt options theory and KMV model to build bank default probability measurement model. Then, we use the backward stochastic differential equation (BSDE) theory to get the robust pricing model of bank default probability and the explicit solution expression under Knight uncertainty. Finally, we use numerical simulation to trace various factors which affect the probability of default, and get the law of changing trend in the probability of default. Through numerical simulation study, we found that:(1) Knight uncertainty has nonlinear and asymmetric effects to the probability of default for the banks; (2) Some macroeconomic variables (such as interest rates, volatility and term of loan), all have nonlinear and asymmetric effects to the probability of default for the banks.(C) Infection mechanisms of the risk of commercial banks and metrical research. For the measurement of individual risk of default alone is insufficient to reflect the breadth and depth of the spread of risk, it is necessary to study the infectivity of the risk. In this dissertation, we use Copula theory to solve the above problems, the emphasis is to use Sklar theorem to construct stochastic dynamic Copula model, and use EIS estimation method and two-step estimation method to estimate the dependent parameter vector. Because the building process of stochastic dynamic Copula model involve the process of setting the latent variables, this paper uses AR (1) process to reflect the time-varying characteristics of the latent variable. By constructing the reflection relationship Ψ(-) between Copula dependent parameters θt and latent variable λt, we obtain the progressive distribution of Copula dependent parameters θt, then, we show the complete analytic framework of model construction, dependent parameter estimation, hypothesis testing, and prediction.(D) Empirical studies about risk infection of commercial banks. Based on the theoretical models built in the third and fourth chapters, we calculate the data of 14 listed commercial banks at first and get the probability of default under the Knight uncertain environment. Secondly, we use stochastic dynamic Copula function to calculate the dependent sequence of the probability of default between all commercial banks. Finally, we use the dependency sequence we got to verify whether the effect of risk infection take place before and after commercial banks have experienced risk, and make judgment on some important commercial banks by calculating the changing range of dependent parameters before and after risk combined with the asset scale of commercial banks.The study found that, (1) the effect of risk infection between commercial banks exists, (2) the author found that large state-owned bank (including ICBC, BC, CBC and CTB) belong to systemically important banks, CITIC Bank and Minsheng Bank belong to potential systemically important banks, while other commercial banks in our sample all belong to non-systemically important banks, (3) through robustness test, the author found that the probability of default is appropriate to be seen as a risk indicators of commercial banks, and is better than the loss rate.This dissertation regards model construction as the main method, it studies the risk infection problem among commercial banks. The main contribution of this paper are:(A) The innovation of selecting the risk indicators. Existing reference mostly use stock loss rate as an indicator of risk measurement to study the effect of risk infection, there is no reference using the data of the probability of default to verify.(B) The innovation of research perspective. In the quantitative study of risk, we use backward stochastic differential equations (BSDE) theory to study the effect that Knight uncertainty brings to the default probability of commercial banks, and we also use stochastic dynamic Copula function to deeply discuss the risk infection problem caused by the default behavior. We extend the individual default risk to multi-frame analysis of default infection. And the modern quantitative analysis frame of the risk based on this theory achieves the conversion from single risk measurement to dependent risk measurement.(C) The innovation of research methods. We combine the backward stochastic differential equations (BSDE) theory with Copula theory to build default probability model of China’s commercial banks in Knight uncertain environment, and use stochastic dynamic Copula model to calculate the dependent coefficient of the default probability between commercial banks. We also emphasize the influence of the risk infection effect brought by Knight uncertainty, probability of default, and random fluctuations.