交易對手信用風險簡單的說就是在店頭市場(OTC)進行衍生性金融商品交易的交易對手之間的信用風險。考慮到全球店頭市場衍生性金融商品的規模之大,交易對手信用風險無疑十分重要。但是在2007年金融海嘯之前,高信用評等之金融機構、主權實體或有抵押物的交易對手的交易對手信用風險被大大低估甚至忽略了。不幸的是,金融海嘯證明了這些看似十分安全的交易對手恰恰承擔了最大的交易對手信用風險,例如黎曼兄弟銀行(Lehman Brothers),美國國際集團(AIG),貝爾斯登(Bear Sterns)及希臘政府等。 信用估值調整(CVA)是交易對手信用風險的價格。對於參與店頭市場的衍生品交易的銀行來說,它正在變得愈發重要。從經濟的角度來看,CVA可以度量交易對手信用風險之大小,因此銀行有動機計算CVA。從監管的角度來看,第三版巴塞爾資本協定要求銀行針對其CVA曝險計提資本,該要求自2013年起生效。尤其是對於獲准以內部模型法計量市場風險的銀行來說,他們可以自建模型的方法計算CVA的風險價值,即CVA VaR,並以此作為CVA風險資本計提的基準,而這一方法相較標準法可節約銀行資本。 本文提出了一種基於跳躍-擴散CIR(Jump-diffusion CIR, JCIR)出險率的CVA VaR計算方法。此方法合乎巴塞爾III對交易對手信用風險的要求,且相比通常作為標杆法的歷史模擬法,有較好表現。
Counterparty credit risk is basically the credit risk between over-the-counter (OTC) derivatives counterparties. Given the huge size of global OTC derivatives markets, counterparty credit risk has always be important. However, for many years before the 2007 crisis, counterparty credit risk with high rated institutions, sovereigns and collateral posting counterparties were largely underestimated or even ignored. Unfortunately, the recent crises showed that these are often the entities that represent the greatest counterparty credit risk -- think about Lehman Brothers, AIG, Bear Sterns and Greece etc. Credit Value Adjustment (CVA) is a price of cost in counterparty credit risk. It has become increasingly important for banks engaged in OTC derivatives trading. Economically, since CVA quantifies counterparty credit risk as a single, measureable, P&L number, banks are motivated in calculating CVA in order to accurately measure and properly manage their counterparty credit risk assumed. From the regulatory point of view, a capital charge for banks against CVA variability under Basel III framework comes into effect since 2013. The banks with the approval of internal model method (IMM) approach for market risk management can calculate their economic capital requirement based on the value at risk (VaR) of CVA, which can be lower than the capital charge based on the standardized approach. In this thesis, I propose a method for CVA VaR calculation based on the Jump-diffusion CIR (JCIR) hazard rate model. It is in accordance with the Basel III framework and it performs better than the historical simulation method, which is usually set as the benchmark method.