【作者】 张家平；

【导师】 任兆璋；

【作者基本信息】 华南理工大学 ， 金融工程与经济发展， 2010， 博士

【摘要】 债权担保证券(Collateralized Debt Obligation, CDO)作为结构性产品的典型代表,在21世纪初获得突飞猛进的发展,交易品种和交易量迅速增长。2007年美国次债危机和随后波及全球的金融危机,又把CDO等结构性产品市场打入谷底。金融危机的爆发反映出一个基本事实:研究机构和结构性产品市场的参与者,没有充分认识和准确度量结构性产品的风险。同时,我国作为金融发展相对滞后的新兴市场经济国家,从2005年开始积极探索和试点结构性产品的发展。迄今为止,主要商业银行都发行过CDO类产品。基于以上认识,CDO产品不会因为次贷危机退出历史舞台,而是在吸取次贷危机的教训、准确度量风险的基础上获得稳健发展。本文选择CDO产品风险评估作为研究主题,分析现有风险度量方法和信用评级存在的问题和缺陷,在中国金融发展背景下提出分析CDO风险的研究框架和风险度量模型。本论文研究的核心部分是尝试解决CDO风险评估的三个关键问题。第一,在运用股市数据计算资产收益率相关系数时,为避免与相关结构无关的信息对相关系数估计的影响,本文提出了FGARCH-ST、FGARCH-SGED和FGARCH-NIG三个模型来刻画单资产股票价格对数收益率序列的典型特征,然后采用Copula技术估计残差序列的相关系数。实证分析得出,FGARCH-NIG模型的结果最好。第二,为充分刻画尖峰肥尾和偏度等信用资产组合损失分布的典型特征,本文基于分布混合假设(MDH)和Vasicek模型构建了NIG-Copula因子模型、双NIG-Copula因子模型和三NIG-Copula因子模型。实证分析表明,与信用资产组合风险评估的标准模型——CreditMetrics相比,NIG-Copula因子模型不仅具有半解析、简洁的分析框架,而且能较好地反映信用资产组合损失分布的尖峰肥尾、偏度和尾部相关等特征,提高了极值VaR估计的准确性。第三,提出基于完整现金流分布的分券风险度量模型,充分利用完整现金流的所有信息,评估CDO分券的尾部风险。本文还考察了结构性产品的界定和国际主要信用评级机构对结构性产品的信用评级。结构性产品的迅猛发展导致产品种类繁多和命名的混乱,这对结构性产品的认识和阅读金融数据造成很大的困扰。通过分析产品结构和风险特性,本文给出结构性产品的统一定义和与主要发达国家结构性产品统计相一致的产品分类。作为对本研究的一个重要补充,本文分析了CDO产品信用评级存在的问题。根据CDO产品信用评级的特点,重点考察了信用评级的模型风险和评级套利,以及内在利益冲突、透明度和竞争状况等衍生问题,从另一个角度得出对CDO风险认识不足的原因。本文是在金融危机余波未尽、全球CDO产品市场陷入低谷,同时我国快速发展结构性产品的背景下,对CDO产品风险的一次探索性研究。利用后发优势,吸取发达国家CDO产品市场的经验教训,在充分认识和准确度量CDO风险的前提下稳健发展CDO市场是发展中国家金融发展的必由之路。本文提出的模型可进一步拓展到CDO的定价。更多还原

【Abstract】 As a stylized representative of structured products, Collateralized Debt Obligation (CDO) has developed rapidly since the beginning of 21 century, with a fast growth of trading scale and varieties. While after the outburst of the subprime crisis in 2007 and subsequent financial crisis, the development of CDO markets is brought to a standstill. The breakout of the crisis discloses a basic fact that research institutes and participants of structured products market, have not fully recognized and measured the risk of structured products. At the same time, China, as an emerging market country which is lagged in financial development, began to explore and launch a pilot project of structured products in 2005. Up to now, most commercial banks have issued some structured products. Basing on aforementioned view, we can conclude that CDO will not step down from the stage of history, but continue to develop stably basing on the lessons from the subprime crisis and measuring the risk correctly. This paper chooses to research on risk measurement of CDO and has constructed a framework of risk analysis and risk measuring models of CDO.The core of this paper tries to solve the three key problems of risk measuring of CDO. Firstly, In order to prevent information irrelative with correlation from influencing on the estimation of correlation when calculating asset return correlation from stock prices data, this paper put up with FGARCH-ST, FGARCH-SGED and FGARCH-NIG to capture the stylized characteristics of time series of log return of stock prices. And then estimate the correlation of residuals using Copula functions. Empirical analyses show that the results of FGARCH-NIG are best. Secondly, For the sake of capturing the stylized features of loss distribution of credit portfolio, such as fat-tail and skewedness, this paper constructs NIG-Copula factor model, double-NIG-Copula factor model and triple-NIG-Copula factor model basing on mixture- distribution-hypothesis and Vasicek model. Empirical analyses show that NIG-Copula factor model is not only semi analytical and compact framework, but also better to describe the fat tail, skewedness and tail correlation of the loss distribution, comparing with CreditMetrics, which is standard model to measure credit portfolio risk. These models contribute to estimating extreme VaR accurately. Thirdly, this paper puts forward a risk measuring model of tranches basing on analyzing full cash flow and measures the tail risk using all information about cash flow.This paper also explores the definition of structured products and credit ratings made by main international credit rating agencies. The fast development of structured products leads to a great variety of products and a confusing nomenclature. It is hard to understand these complex financial products and read relative statistical data. This paper gives a uniform definition by analyzing structure and risk characteristics of structured products. As an important complement of our research, this paper explores credit rating on CDO by main international credit rating agencies。Model risk, rating arbitrage, conflict of interest, and competition are analyzed mainly. We get the reason why the risk of CDO has not been understood fully from the rating angle.This paper is an exploring research on the risk of CDO under the background of the aftermath of financial crisis not calming down, global CDO markets keeping a standstill, while structured products developing fast in China. It is the only way to develop CDO market in lagged and developing countries on the precondition of utilizing advantages of backwardness fully, learning a lesson of CDO market in developed countries and measuring the risk of CDO correctly. The models proposed in this paper can be extended to pricing of CDO.更多还原