在這篇論文中，為了觀察高與低授信品質資產間不同的相關係數結構，我們將擔保債務憑證內的投資組合分成兩群，分別有不同的違約機率。首先，在單因子模型下，控制平均相關係數水準，我們發現低授信品質資產間的相關係數較高授信品質資產間的相關係數來的高。第二，在隨機因子負荷量模型下，我們可以從市場上擔保債務憑證的價格反推回隱含的相關係數結構。我們使用四組不同的相關係數結構假設並且使用敏感度分析，結果顯示低授信品質資產的相關係數在經濟出現極端狀況時會較高。 In this paper, to examine different correlation structures between high and low quality names in a CDO, we separate the portfolio into two groups with different hazard rates. First, under one-factor model, the results show that correlations are higher among low quality names than those among high quality names when controlling the average correlation level. Second, under random factor loadings model, we can calibrate the correlation structures from market spreads. We use four assumptions of correlation structures and conduct a sensitivity analysis. It shows that among low quality names, correlations are higher when extreme economy states occur.
