重隨機假設下之動態違約相關性描述

Abstract

本研究於重隨機假設下，藉由Hull and White(2008)非同質性(non-homogeneous)波松抵達的立足點，考量Das, Duffie, Kapadia, and Saita (2007)在實證上所作之違約傳染性論述，建構資產違約相關性描述之動態模型。本研究的動態描述允許違約事件的發生可同時由系統風險因素及非系統風險因素所造成，且具有彼此傳染的效果。本研究進而考量一具多重信用資產之資產池，其違約發生的頻率可分解為市場因素所驅動及非市場因素所驅動。且允許不同時點之波松抵達之跳躍幅度受上一觀察時點之信用事件發生的密集度所影響。本研究之研究成果預期可廣泛應用於信用衍生性商品之評價，本研究希望藉由參數校準結果，反應目前市場對於違約相關性之預期另外，本研究亦可針對模型中各參數進行敏感度分析，區分當參數變化時，對於各商品分券信用價差之影響。

In this research, we aim to provide a dynamics description of default correlation under the doubly stochastic assumption. We consider the argument of Das, Duffie, Kapadia, and Saita (2007) on the role of contagion in default correlation, and extend the approach adapted by Hull and While (2008) where non-homogeneous poisson jumps are used to model the arrivals of credit events. We allow for the driving risk factors for the default frequency to be both systematic and idiosyncratic, and in order to account for default contagion, we allow the magnitudes of the jumps to be a function of observed default frequencies. Our results are expected to be widely adaptable to the pricing and risk management of mult-name credit derivatives. We also aim to provide calibration procedures that model under development are capable upon.