Credit Risk Measurement Based On Bootstrap Methods And Applications

In the late of 20th century, with the rapid development of global economy and increasing volatility in financial markets, world's banks and financial institutions are facing more and more credit risks, More and more financial events were gradually exposed, such as the bankruptcy of Lehman Brothers's, the default of Russian National Bank in last century, the subprime mortgage crisis of USA and so on. It brought a heavy blow to the development of world economy and reflected many issues in credit risk management, exposing lots of flaws and shortcomings in existing theories and methods of credit risks management. So an urgent need for a more sophisticated risk identification, measurement and prevention techniques is required. Therefore, the credit risk measurement models are created and developed rapidly in this field. KMV credit risk measurement model, Credit Metrics model, Credit Portfolio View (CPV) model are the current mainstream models. But those mainstream risk measurement models are mostly based on assumptions which have been given the distribution of the sample, and which parameters based on the given assumption, At the same time, the selection of the sample also has some limitations, which requiring a larger sample than general. However, in practice, especially in the credit market, a large number of effective data is difficult to obtain, and effective historical data for reference is even less. So it is much difficult to obtain a large and effective sample on these models and presume to the distribution of model parameters. Or the error will be relatively large, even can not be calculated.This paper introduces a new risk measurement method which base on the Bootstrap sample, especially in the case of the small sample. The new Bootstrap sample will be closer to the overall real sample so that we can use traditional risk measurement models to calculate the credit risk.This paper is divided into four parts:The first part of this paper is Chapter 1 and Chapter 2. Chapter 1 introduces the purpose and significance of this paper, and illuminates the relevant research status, the idea of this study and the structure of this paper as well; In Chapter 2, the author makes a brief recommendation of credit risk, as well as theory of three classic basic credit risk measurement models, In the later part, the advantages and problems of three models will be explained one by one.Chapter 3 is the secend part. It briefly introduces the basic theory of Bootstrap methods and Stable Distribution, and theirs development, implementation process and so on.The third part is Chapter 4, It empirically analyzed the application of three main risk measure models which based on the Bootstrap sample in the credit risk market, and describes the basic idea and operation of three main risk measure models which base on the Bootstrap sample.The parameters of KMV model which based on the formula BSM-the expected probability of default (EDF) is solved by the empirical distribution of Bootstrap sample and estimating its confidence interval of empirical distribution; Credit Metrics model which based on the traditional VAR generally assumes that the value of debt subordinates normal distribution, while this paper assumed the value of the debt subordinates the general empirical distribution of Bootstrap sample, which is stable Distribution, the parameter of traditional Credit Portfolio View model-probability of default (Y) was calculated on the given normally distributed under the conditions of the small sample, But this paper is under the conditions of large sample which base on the Bootstrap sample to empirical analysis, finally, this paper make a comparison in the case of large with the small samples in parametric test and date accuracy.The fourth part is Chapter 5 which is a summary of this paper. It pointed out the problems remained to be resolved and the future research directions of this topic.