Copula 函數型 CDO 評價模型中各種相關性測度之比較

Abstract

我們比較了在各種相關性測度法下, 根據傳統 copula 函數為基礎的CDO評價結果. 傳統上是使用各資產報酬的線性相關係數作為衡量CDO 資產池中各項資產之間的相關性準則. 在本篇論文中, 我們嘗試使用信用違約交換利差(CDS spread) 的變動率來替換資產報酬率. 並且, 我們比較了若干相關性衡量公式, 如 Kendall’s tau 和 Spearman’s rho, 實證結果指出, 在某些時期它們的確可以得出更貼近市場報價的評價結果.我們也比較了使用不同時期資料所造成評價結果的差異, 結果指出在2007 年, 各項資產之間的相關性結構可能有明顯的改變.

We compare several types of correlation measurements used as inputs of the traditional Gaussian copula pricing model for CDO tranches. Using the linear correlation coefficient of stock returns to measure the correlation between each entity within the CDO is the traditional approach. Here, instead of using stock returns, we try to measure the correlation via using CDS spreads. We compare the pricing results among different initial inputs, such as different time horizons of data and different correlation formula. Our result indicates that instead of using the linear correlation coefficient, in certain periods, using other dependence measurements such as Kendall’s tau and Spearman’s rho can perform better pricing results. Also, the pricing results carry high explanatory power for various correlation measurements. Furthermore, we find that according to our pricing results, the correlation structure among the entities may have changed significantly during the 2007 subprime crisis.