Optimization Of Hierarchical Structure Of CDO Based On Random Simulation

This paper studies the optimization of detachment point and attachment point in CDO hierarchy.In the past academic research on CDOs,most of the scholars were focused on the pricing of CDO coupons.However,in the hierarchical structure of CDOs,the products are almost sold in equal segments of length.So does the probability interval for each of the points best cover the risk it takes? The key to the optimization lies in the similarity of risk distribution and risk characteristics.A natural idea is to divide the probability with similar risk loss distribution into one coupon,so as to theoretically play the function of transferring risk and risk coverage of CDO to a greater extent.This is also the core idea and goal of this paper.In the optimization of CDO structure,there is a difference in which model is used to fit the risk structure and risk distribution in reality.The traditional factor Copula model was established on the assumption of joint normal distribution,but a large number of studies and the subprime crisis in the United States in 2008 showed that there was a correlation smile in the default process of CDO,which indicated that the traditional Copula model could not accurately describe the statistical characteristics of the default process of assets in the CDO asset pool.In order to better simulate the risk loss distribution as well as to the CDO modeling,we use the Monte Carlo stochastic simulation method,normal inverse gaussian(NIG)is used to describe the distribution features of the return on assets,the use of Vasicek process to describe the change of interest rate,and through a series of reasonable assumption,the default correlation of each asset in the asset pool smiling facts contained in the model,rather than mechanically defined relevance into an orthogonal public risk factors and specific risk factors.This paper proposes a method to find the best points by cutting the cumulative loss distribution function of total assets,which is more scientific and reasonable than subjectively setting attachment and departure points,and obtains the result with lower utility function.In the empirical analysis,according to the data of i Traxx and random simulation,the respective risk loss distributions under the two risk models were calculated.By optimizing the attachment points and separation points of each layer,the probability with the most similar risk characteristics was sold in the same layer and compared with the existing i Traxx.The results showed that the CDO could better cover and transfer risks after the optimization of the hierarchical structure.The effect of traditional factor Copula method will also be investigated.