考慮信用風險下之多期風險值模型

Abstract

全球金融環境自由化之後，風險管理的重要性為大眾所重視。債券擔保憑證(CBO)是近年來成長迅速的資產證券化商品，由於其資產標的日益複雜，故評估其風險成為一重要課題。風險值(VaR)是目前最廣為運用的風險管理工具，但鑑於風險值多只考慮市場風險，未將信用風險同時納入考量。公司債投資組合需要同時考慮市場風險以及信用風險，且目前模型無法掌握風險值未來的變化。因此本研究希望發展考慮信用風險下的多期風險值模型。 本模型概念為利用債券投資組合信用資訊所提供的預期回收比率(Expected Recovery Rate)將有違約風險的現金流量轉成無違約風險現金流量後，利用RiskMetrics及整合債券法進行風險值之衡量。 實證分析運用局部評價法中的一階常態法、二階常態法及完全評價法中的歷史模擬法衡量公司債投資組合的風險值。並運用公債投資組合測試本研究所提出的整合債券法，將其結果與RiskMetrics作比較。 本研究的結論如下：運用整合債券法以及RiskMetrics進行公債投資組合的風險值衡量，發現整合債券法的估計相對較準確。

Risk management has experienced a revolution recently. It started by the new method for measuring financial market risk--Value at Risk (VaR) that was developed in response to the financial disasters of the early 1990s. The original VaR method focuses only on measuring market risk and cannot trace the variation of VaR in the future. In order to improve it, this study builds a multi-period VaR model being able to incorporate credit risk. Knowing default probabilities, recovery rates, and default correlation for all the assets in a bond portfolio, we could use expected recovery rate to transform risky cash flows into default-free equivalent cash flows. Moreover, RiskMetrics and Aggregated Bond Method could be employed to obtain the portfolio VaR. After comparing results from these two methods, we conclude that Aggregated Bond Method can better catch the risk better than RiskMetrics.