倒帳相關性:比較研究

Abstract

正確估計倒帳相關性在風險管理及信用衍生性商品訂價中非常重要。本研究研討倒帳相關性的外在成因與影響。以及用來建構倒帳相關性的兩種類型”Structural form model”及”reduced form model”。並就代表性倒帳模式探討倒帳相關性隨著時間的變化之情況。不論是歷史統計資料或者是zhou(structural form model)及Gaussian Copula(reduced form model)，其倒帳相關性相對於時間的變動都有在很短的時間內很小的特性，甚至可視為隨機(互相獨立)。其後，快速上升到一個高峰後會趨於平緩且緩慢下降。倒帳相關性和所使用的資料的相關性呈正向關係並且倒帳相關性通常小於資產相關性。另外，研究發現Zhou模式在長時間下其資產相關性對倒帳相關性的影響不明顯。而在Gaussian Copula之下，資產相關性對倒帳相關性影響顯著且隨著時間增加擴大差距。

Modeling default correlation correctly is an important task in risk management and pricing credit derivatives. We discuss why default correlation exists and its implication .Covering the two primary types of models that describe default process,“structural form model”and“reduced form model”.Under different model, we demonstrate how default correlation change over time under different credit quality and asset correlation. We find that, in both models, the default correlations over a short horizon are very small. Over the long run they increase and then slowly decrease with time. Default correlation and the asset correlation have the same sign. Finally we find that under different asset correlation, The default correlations in Zhou(2001) converge over a long horizon, but Gaussian copula’s(reduced form) default correlations still diverge.