cdo与cdo2评价.ppt

* * * * * * * * * * * * Factor Copula Probability Bucketing Factor Copula Probability Bucketing數值方法 Factor Copula 模型設定 (Hull White 2004) Factor Copula Next Step 利用Hull and White (2004) probability bucketing方法估算離散型之CDO條件違約損失分配，如此再透過Gaussian Quadrature積分技巧之運用，即可估得損失機率分配 Probability Bucketing Model Setting Choose buckets{0,b0},{b0,b1},…,{bK-1, ∞} for the loss distribution. {0,b0} is the 0-th bucket;{bn-1,bn} is the n-th bucket(1≦n ≦K-1); {bK-1, ∞} is the K-th bucket. pn: the condtional probability that the loss by time T will be in the n-th bucket An: the mean loss conditional that the loss is int the n-th bucket (0≦n ≦K)? ? Initial condition: p0=1 , pn=0 for n0 A0=0 , An=0.5(bn-1+bn) for 0nK , AK=bK Probability Bucketing Suppose there are N debt instruments Lj: the loss given default from the j-th debt by time T αj: the probability of a default from the j-th debt by time T u(n): the bucket including An+Lj Rule Calculating pn and An iteratively by first assuming that there are no debt, then there is only one debt, then assuming that there are only two debt, and so on. Probability Bucketing Suppose we have calculated the pn and An when the first j-1 debt are considered. Consider bucket from K to 0 , An+Lj Every bucket’s mean loss adds αjLj andExpected loss = when u(n)=n p*n, p*u(n), A*n, A*u(n) are the values before the probability shift is considered Probability Bucketing when u(n)n 機率 由Pn轉到Pu(n) Example 假設CDO內有三個信用資產，資產大小皆為10，且每個資產違約機率皆為0.5，違約後之剩餘價值回收率為0.5，試問此一內含三個信用資產的投資組合損失分配為何 機率倒桶法 (Probability Bucketing method) A0=0 P0=1 A1=2.5 P1=0 A2=7.I5 P2=0 A3=12.5 P3=0 A4=15 P4=0 0 5 10 15 Example Example Reference 擔保債權憑證(CDO)之評價與分析PPT-廖四郎　教授 擔保債權憑證之評價－Copula 分析法 -廖四郎　教授 Valuation of a CDO and an n-th to Default CDS Without Monte Carlo Simulation – John Hull and Alan White (2004) * * * * * * * * * * * * * * * * * * * * * Copula方法簡介  Copula方法簡介 Copula原理及其運用 Copula函數類型 變數相關性之衡量 違約機率 CDO分券之評價模式 CDO評價流程---Copula法 計算出分券[C,D]信用價差 1