擔保債權憑證隱含違約相關性之研究─以台灣為例

Abstract

在擔保債權憑證中，投資群組違約相關性的歷史資料並無法代表未來的違約相關性，因此進化成以市場交易的信用交換違約指數以及批次證券報價求算隱含違約相關性。求取的方法分別有Compound Correlation以及Base Correlation。Base Correlation會改善Compound Correlation 的correlation smile以及多重解的現象，並且可以評價客製化批次證券，目前市場上最常用的方法是JPMorgan的Base Correlation。 而若是要評價的投資群組與指數成分債權不符，則可以使用黃裕烈 (2006) 的方式模擬出信用違約交換指數以及批次證券指數公平溢酬。但是套用台灣資料以及Base Correlation時卻發現其違約機率被低估。未來研究方向將致力於改善違約累積機率的分配。

Historical default correlation of reference portfolio lacks predictability for CDO pricing, so default correlation needs to be implied from CDS index and Tranche Index traded by market investors. There are two ways , one is 「Compound Correlation」 and the other one is 「Base Correlation」。Base Correlation can improve the phenomenon of correlation smile and multiple solutions for Compound Correlation。It also can price bespoke tranche。The most popular approach is JPMorgan’s Base Correlation。 If the reference portfolio is inconsistent with the index reference portfolio，the solution is to use黃裕烈(2006) model to simulate market fair spreads of index and every tranches。But when using Taiwanese data and Base Correlation，the default probability is underestimated。Future researches will be focused on fitting the cumulated default probability for Taiwanese market。