Study On Credit Risk Measurement Models

Credit risk is the distribution of financial losses due to unexpected changes in the credit quality of a counterparty in a financial agreement. It pervades virtually all financial transactions. And how to measure and manage credit risk arouses great concern from scholars and practitioners in this field. As compared with western countries, there is an enormous distance between China and the developed countries in the area of study and management on credit risk. For our national financial institutions, while introducing advanced credit risk management systems from abroad, they should also carry out basal academic study on credit risk measurement to develop credit measurement and management systems adaptive to the status quo of our country.The paper begins with a summary of credit risk measurement models and evaluation methods based on the survey of the key conceptions about financial risk and credit risk measurement, and analyzes the principles of credit risk, then discusses the classical modeling paradigms involving structural models including Merton and first-passage-time models and reduced form approaches, and then summarizes some results about rating transitions and LGD(Loss Given Default). After these analyses, we give the general formula to price defaultable bonds and loans. Finally, we sketch some aspects in measuring credit risk concerning the correlation of defaultThis thesis mainly probes some academic models, and proves classical and fundamental Merton model and first-passage-time model, and then makes an analysis of empirical comparison on traditional models forecasting default rate with findings that the probit model taking account of the heteroskedasticity has the stronger explanatory power than discrimant and homoskedastic models. We calculate the rating transition frequencies matrix with datasets from a state-owned commercial bank, and make a study on LGD, to find out factors to determine LGD include debt agreement characteristics, firm-specific characteristics, industry effects and macroeconomic factors. Meanwhile we give several pricing paradigms on risk of defaultable bonds and loans, and present the program written in Matlab language to simulate the VaR of loan portfolios, in the end we skillfully obtain the portfolio set constrained by VaR.