本文在違約事件為條件式獨立的假設下，運用遞迴法則(recursive method)建構雙層保護合成型擔保債權憑證之損失分配，進而分析憑證之風險特徵。研究結果發現雙層保護合成型擔保債權憑證雖然標榜以雙重信用違約保護層及更大風險分散程度之債權群組(reference pool)配置來兼顧利潤與風險的平衡，但實際上卻是高槓桿程度的商品，其雙層保護結構設計，擴大了模型底層參數變動對上層損失分配之影響效果。本文針對雙層保護合成型擔保債權憑證，提供信用增強、違約相關性與資產重疊度之敏感度分析。數值結果顯現權益一權益分券(equity on equity tranche)至次償一先償分券(mezzanine on senior tranche)等六種分券的名目本金加總，雖然只佔其商品資本結構之極小比例，但卻承擔了部分之總信用風險。因此市場參與者必須仔細區分「風險金額移轉數目」與「內含風險移轉程度」之差異。

In this paper we show that a synthetic CDO of CDO(subscript s), though claiming to offer enhanced credit protections and higher yields to investors through its double-layered structure and a well-diversified reference pool, is in fact highly leveraged. The double-layered product structure directly amplifies the resulted expected-loss of the master CDO due to default entities in the underlying pool of inner-CDO(subscript s). We conduct sensitivity analysis of tranche spreads with respect to the credit subordination levels, the default correlations, and the extent of asset overlap among inner CDO(subscript s). Our numerical results reveal that the notional sum of equity to mezzanine sub-tranches, though being only one-tenth of the total notional, in fact withstands 80-90% of the total credit risk.