| Industrial structural adjustment is an important strategical task for current Chinese economy development, and it would impact on bank credit structure especially credit industrial structure deeply. Irrational credit structure, such as being over-concentrated or mainly distributing among negative-influence industries in industrial structural adjustment, would bring banks systematic risk. This thesis, from bank’s perspective, focuses on credit structure and credit risk in the industrial structural adjustment environment. The research follows3questions step by step:how is credit structure? Why and how does it change with industrial structure? What is the influence of industrial structural adjustment on credit risk? Accordingly,3topics are researched:the unbalance of credit structure and industrial heterogeneity analysis; the action mechanism and influence of industrial structure on credit structure analysis; credit risk of industry portfolio measurement based on time-varying Copula function.This thesis provides the basis for banking decision makers to estimate future credit structure and credit risk change. It allows banks to take actions in advance to optimize credit structure and prevent systematic risk.The main work and conclusions are as below.(1) Static and dynamic anaylsis of Chinese credit structure unbalance, and research of industrial heterogeneity factors causing above unbalance.In this part, the full view of Chinese credit structure is given, which characterizes being unbalanced among industries. Currently credit loan is mainly distributed among basic establishment and heavy industries, fitting current industrial structure well. Absolute and relative industrial credit size measures are proposed to analyze static unbalance. HHI index, σ-convergence and absolute β-convergence tests show the dynamic trend of future credit structure is still unbalanced, while the extent of unbalance narrows down slowly.In the framework of supply-demand market mechanism, the unbalance is theoretically revealed to come from industrial heterogeneity. The ownership of industry during transition period weakens the effect of market on industrial allocation of loan.Panel empirical research shows those heterogeneity factors include industrial capital intensity, industrial competition level, industrial earning capability, industrial guarantee capability and industrial ownership. (2) The action mechanism and the influence of industrial structure on credit structure are studied quantitatively.In empirical research, the relationship of industrial structure and credit structure is’studied with data of19industries from2005to2012by establishing a dynamic panel model and estimating parameters with system GMM.The result shows previous industrial structure accumulation influences existing industrial allocation of credit structure; the change of industrial structure influences industrial allocation of incremental credit. The former is more important, and for different industry, influence period is different. A varying intercept and a varying coefficient fixed-effects panel model are established to study the quantitative correlation between the two structures. The result shows industrial structure has positive effect on industrial distribution of credit structure generally. At the same time, credit is more weighted in encouraged industries than in declining industries, and flows to the former more.(3) A credit risk measurement model of industry portfolio is developedIndustrial structural adjustment causes the change of credit structure and the latter further causes the change of bank overall credit risk. A credit risk measurement model of industry portfolio is established in this thesis by two steps. First step is to get distribution function of one industry’s credit risk. Inspired by Merton’s model, I use publicly available industrial average debt ratio data to calculate industrial EDF. Empirical research shows industrial credit risk distribution presents characteristics of autocorrelation, fat-tail and clustering of volatility. According to AIC, BIC and LL indexes, AR(1)-GJR(1,1)-Skewed-t model can describe this distribution well.Second step is to get joint distribution function of more industries’credit risk. There are two problems:how to get the correlation of credit risk between industries, and seek suitable joint distribution function. Inspired by Credit Metrics model, I use the correlation between two industrial stock yield rates to derive the correlation between two industrial credit statuses. Time-varying Copula functions are introduced to describe the non-linear correlative joint function of industries’ credit risk. In empirical research,4time-varying Copula functions (normal, t, Clayton and SJC) and5static Copula functions (normal, t, Gumbel, Clayton and SJC) are used to fit the real distribution. AIC, BIC and LL indexes show time-varying t-Copula with symmetrical tail fits best. Monte Carlo simulation is applied to calculate EDF of industrial credit portfolio with this model.So the model established in this thesis has the marginal distribution of AR(1)-GJR(1,1)-Skewed-t, and joint distribution described by time-varying t-Copula function. Based on this model and the research result of other characters, an example is given about how to calculate the change of credit risk caused by industrial structural adjustment.The main creativity of the paper is as following.(1) Credit industrial structure nationwide is studied with empirical method first time in this thesis. It is found that the allocation of credit among industries is not only statically unbalanced but also dynamically unbalanced. Absolute size measure and relative size measure are created to describe the unbalance. HHI index, σ-convergence and absolute β-convergence tests are introduced to analyze the dynamic unbalance trend. It is proved that the unbalance of credit allocation among industries mainly comes from industrial heterogeneity by theoretical analysis and empirical method (fixed-effects panel model). Main relative industrial heterogeneity factors are settled down. The study provides the basis for bankers to grasp the future situation of credit industrial structure, and deeply understand the contributing factors.(2) The action mechanism and influence of industrial structure on credit structure are studied by empirical method. Current available researches mainly focus on "how can credit support industrial structural upgrade", while the reverse research is scarce. By establishing a dynamic panel model and making system GMM estimation, how the credit structure is changed with industrial structure is uncovered. By establishing a varying intercept/varying coefficient fixed-effects panel model, the quantified correlation between the two structures is analyzed.(3) A credit risk measurement model of industry portfolio is developed. Among the process of solving the following three difficulties, the model is developed finally. First inspired by Merton’s asset value model, using publicly available industrial average debt ratio data, I explored a method to calculate industrial credit risk represented by industrial EDF (Expected Default Frequency). Then inspired by Credit Metrics model, I use the correlation between two industrial stock yield rates to derive the correlation between two industrial credit statuses, solving the un-observable difficulty of industrial credit. Last by introducing time-varying Copula function, which describes the dynamic non-linear correlativity, we find a method to describe the industry joint distribution of expected default frequency. Monte Carlo simulation is applied to calculate expected default frequency of industrial credit portfolio.
|
PDF Full Text Request