| Collateralized Debt Obligation (CDOs) are a type of structured asset-backed security (ABS) whose value and payments are derived from a portfolio of fixed-income underlying assets, that is, CDOs securitizes underlying assets into an assets pool to re-split into different tranches to reallocate the credit risk for different types of risk investors as a derivative securities product, based on mortgage debt credit. As one of core issues in Collateralized Debt Obligations, CDOs pricing is how to determine debtor's default probability and correlations between different underlying assets and thus to get the cumulative default loss distribution of assets pool. Since the 2007 subprime mortgage crisis originated by the U.S, the Gaussian factor copula model, as the market standard model, is no longer able to meet actual needs. It is imminent to seek the better Collateralized Debt Obligations pricing model reflecting the market characteristics.This paper summarizes main Collateralized Debt Obligations pricing models and its research review at home and abroad. Then, focusing on dynamical two-parameter levy subordinator model, this paper develops its extensions in two ways.Firstly, this dissertation expresses the recovery rate as a function of the indicator of economic developments, constructs a Collateralized Debt Obligations pricing model based on stochastic recovery. After the numerical simulations, it is shown that the model with stochastic recovery can reflect the real situation of the Collateralized Debt Obligations market.Secondly, this paper constructs the other extension models through introducing inverse Gaussian subordinator and gamma subordinator, and gives the cumulative loss distribution of assets pool under Fast Fourier Transform techniques. Under different criterions, It is shown that these two model can reflect risk of different tranches, and the inverse Gaussian subordinator model can reflect the situation that sometimes the second tranche credit spread is much larger than the third tranche in the Collateralized Debt Obligations market. Except that, this paper respectively discusses the relationship between the subordinator model and factor copula model, intensity model. It is shown that default probability in subordinator model is the same as that in subordinator model when firm-specific is not be considered and in some ways, the subordinator model is the limit of the intensity model for the mean reversion rate to become to infinity. |