Parameter estimation and stress testing of portfolio credit risk based on transition probability distribution
In recent years,stress testing has played an increasingly important role in modern bank risk management,because of its properties to measure financial risk in extreme environ-ments.Accurately measuring the driving effect of systematic factors on portfolio risk is the key to effectively controlling tail risk and preventing financial crises.This paper studies the problem of parameter estimation and stress testing of portfolio credit risk,and establishes a Vasicek credit transfer model including systemic factors and multi-level correlation coefficients.After analyzed the influence of correlation on the credit transfer probability between ratings,the correlation coefficient conditions of rating monotonically consistent is given to ensure the monotonic con-sistency of the credit rating transfer probability.In this paper,a parameter estimation method based on the degradation probability distribution is constructed.Using the transition probability matrix,the rating lifetime distribution under the market equilibrium state is derived,and then the estimation algorithm of the correlation coefficient and the systemic factor is given.Compared with the existing algorithms,the proposed method weakens the influence of the fixed bias of the systematic factor,overcomes the dependence of the distribution of the systematic factor,so that improves the accuracy of the estimation while reducing the computational cost.Simulation re-sults show that the proposed method not only effectively improves the calculation speed,but also significantly outperforms the existing methods in estimating the systematic factor,correlation coefficient and portfolio loss.Besides,the extreme loss estimated by this method can fully cover the out-of-sample portfolio loss,it follows that the financial risks can be better predicted and prevented in stress testing,and the financial market stability can be maintained.
NETL
NSTL
万方数据
维普
