本文延伸Hull and White (2010)之跨池因子繫聯結構模型中違約相關性之描述，藉由納入Normal Inverse Gaussian分配並允許其帶有狀態轉換之特性，我們探究信用結構式商品清償順位結構中，影響次順位信用保護層(subordination level)之因素。我們以房屋抵押擔保貸款債權憑證(MBS CDO)為例，分析資產違約相關性、資產池微粒化程度、跨池違約相關性等結構性變數如何影響分券評等之合理性及風險特徵。本文的研究結果呼應Azzalini and Capitanio(2003)中所提及採用Gaussian因子繫聯結構模型之於評價信用結構商品的缺失。我們發現增進信用資產池損失分配的之厚尾性描述，得以改善高估或低估分券信用價差的情況。 / By incorporating the Normal Inverse Gaussian distribution and allowing for regime shifts in the correlation structure of the multi-pool factor copula of Hull and White (2010), in this thesis we explorer the factors constituenting the subordination levels of credit-structured products. Using MBS CDOs as an example, we examine how model-embedded variables, such as default correlation, reference-portfolio granularity, and cross-pool correlation, affect the risk profiles of MBS CDO tranches. Our numerical results echo the findings of Azzalini and Capitanio(2003) in that correlation structure obtained under the Gaussian factor copula model may be inadequate in capturing the fact-tailed characteristic of the reference-pool loss distribution, thus can result in over/under-estimation of CDO tranche spreads.