危机传染背景下资产组合风险模型测试精度比较研究

发布时间：2017-12-30 23:22

本文关键词：危机传染背景下资产组合风险模型测试精度比较研究　出处：《西南交通大学》2013年博士论文　论文类型：学位论文

【摘要】：从1929年的美国股市大崩溃开始,国际金融市场历经了数次危机传染。而美国次贷危机更因在短时间内便震撼国际金融市场,演变成全球性的经济危机且影响深远,而成为金融危机传染的代表性事件。国际金融市场危机传染的频繁发生,使得投资者面临着巨大的风险,同时也对金融风险管理提出了更高的要求。金融风险管理的基础与核心在于如何正确地度量风险,由于在金融实务中,投资者往往对资产组合进行投资而非单一资产,因而对于资产组合风险的度量更具现实意义。同单一资产的风险评估相比,投资组合的风险评价更为复杂,因为在组合风险的计量过程中,不仅需要考虑组合中单一资产收益的波动率模型,还必须考虑到组合中各个资产间的相依关系。然而金融资产间的相关性错综复杂,所以如何正确地选择和运用风险度量工具就成为组合风险评价中最重要的一个环节。本文以美国SP500指数、日本日经225指数、中国上证综指和香港恒生指数作为研究对象,将四类时变Copula-EVT模型作为核心研究方法,以美国次贷危机为代表的金融危机传染作为一条贯穿始终的研究主线,层层推进四个关键问题： ·危机传染是否显著地影响了股市间尾部极值风险传导的强度? ·股市间风险传导的方向是否发生明显的变化? ·危机爆发后,对于二元资产组合及多元资产组合的多头头寸与空头头寸,各类风险模型假定下的VaR风险价值模型和ES预期损失模型的测度精度是否显著改变?其变化状况是否有所不同? ·哪些因素将影响到VaR模型和ES模型的预测准确度?其影响的结果怎样? 为此,本文在各股指收益的标准残差序列的基础上,结合EVT极值理论,构建边缘分布,分别运用四类时变Copula函数构建各个资产组合的联合分布,采用拟合效果相对较好的时变t Copula-GJR-EVT模型,得到危机爆发前后股市间的极值动态相依系数；通过时变SJC-Copula-EVT模型获得股市间的上尾和下尾极值相关系数。运用格兰杰因果检验的方法,分析了金融危机传染对于股市间风险传导方向产生的影响。实证研究结果表明,危机的爆发,对于股市间极值风险传递的强度和风险传导的方向产生了很大的影响。次贷危机发生以后,国际股市间极值风险传染的程度普遍增强,其时变特征也非常明显。从股市风险传导的方向上看：次贷危机爆发以前,股指间的风险主要汇集于纽约股市,而三大亚洲股市,即中国沪市、香港股市和东京股市间不存在风险传导关系。次贷危机爆发后,股市间风险传导的途径和方向发生了明显的变化：不仅美国纽约股市与其他股市间形成了双向风险传导关系,亚洲股市间也显现出密切而复杂的风险传导格局,中国股市与国际股市的极值风险关联度显著增强。这些鲜明的特点,无疑将影响到投资组合风险模型的测度准确度。在此背景下,本文将四个股指收益进行组合,构造了二元资产组合及多元资产组合,基于四类时变Copula-EVT模型和DCC-GARCH模型,分别针对多头头寸和空头头寸,建立了VaR模型和ES模型,并运用Backtesting方法进行后验分析,对比研究了危机以后各类风险模型测度精度的变化状况。实证结果表明：第一,次贷危机爆发后,金融市场间极值风险正向相关的程度显著增强,分散化投资的作用在一定程度上被削弱,资产组合VaR风险价值模型的测度精度有所降低；然而在某些状况下,预期损失ES模型的测度精度却在危机后有一定程度的提高。第二,无论是VaR模型还是ES模型,基于时变Copula-EVT构建的风险模型,其测度精度在总体上高于DCC-GARCH风险模型。第三,边缘分布模型的选择,对于时变Copula-EVT风险模型的测度效果具有重要影响。第四,由不同类型的时变Copula函数构造的风险模型,对于资产组合风险的预测准确度有所不同。综合来看,危机爆发以后,时变SJC-Copula-EVT-VaR模型与时变tCopula-EVT-ES模型的测度精度均相对较高,这进一步表明,次贷危机对于资产组合风险模型的测度效果产生了巨大冲击,善于刻画变量间非对称性、厚尾性相依特征的模型显现出较强的测度优势。尽管如此,对于资产组合的风险测度,仍需根据资产组合的分布特征以及科学的对比研究来灵活地选择合适的风险模型。在经济全球化的今天,无论是从时间还是空间的角度,金融危机传染都日趋严重。美国次贷危机爆发以来,金融市场的运行环境更加错综复杂,金融风险极容易在各个市场之间相互传染。在金融危机频发的背景下进行投资组合,应特别注意防范组合投资风险。对于资产组合的风险评估,应尝试构建多个风险模型,选择测度准确度相对较高的模型进行风险评估,并将VaR模型与ES模型结合使用。此外,对投资组合的风险评估还应立足于动态的角度,因为尾部极值风险传导具有时变特性,所以在使用静态类风险评估方法时必须谨慎,以防错误评估资产组合的风险,同时,必须及时有效地采取相应的止损措施,以防范极端金融事件导致股市同时暴跌而对组合资产造成巨额亏损。
[Abstract]:From the beginning of 1929, the U.S. stock market collapse, the international financial market has experienced several crisis. While the U.S. subprime mortgage crisis because it shocked the international financial market in a short period of time, evolved into a global economic crisis and a far-reaching influence, and become the representative of financial crisis incidents. The frequent crisis of international financial market contagion, the investors face a huge risk, but also put forward higher requirements on financial risk management.
The basis and core of financial risk management is how to correctly measure the risk, because in the field of finance, investors tend to portfolio investment rather than a single asset and portfolio risk measurement for more practical significance. The risk assessment with single assets compared to the risk evaluation of portfolio is more complicated, because the measurement in the process of portfolio risk, not only need to consider a single asset return portfolio volatility in the model must also take into account the dependence between the individual asset portfolio. However, the correlation between financial assets complicated, so how to correctly select and use of risk measurement tools has become one of the most important link in the evaluation of portfolio risk.
In this paper, the SP500 index, Japan's Nikkei 225 index, Chinese Shanghai Composite Index and Hongkong's Hang Seng Index as the research object, the four time varying Copula-EVT model as the core research method, sub loan crisis in the United States as the representative of the financial crisis as a main line through all the four key problems on promoting layers:
Does the crisis contagion significantly affect the intensity of the tail extremum risk transmission between the stock markets?
Is there a significant change in the direction of risk conduction between stock markets?
After the outbreak of the crisis, did the measurement accuracy of VaR risk value model and ES expected loss model change significantly under the assumption of all kinds of risk models for two asset portfolios and multiple asset portfolios' multi position and short positions? Are there any differences in their accuracy?
What factors will affect the prediction accuracy of the VaR model and the ES model? What is the result of its impact?
Therefore, based on the standard error sequence of the stock index on the EVT combined with the extreme value theory, constructing marginal distribution and joint distribution using the four class construction of each portfolio Copula function, the fitting result of time-varying t Copula-GJR-EVT model, obtained before and after the stock market crisis between the extreme dynamic dependence coefficient; through time varying SJC-Copula-EVT model to obtain the stock market between the upper and lower tail extreme correlation coefficient. By using the method of Grainger causality test, analysis of the financial crisis contagion effect on stock market risk conduction direction. The empirical results show that the outbreak of the crisis, the strength and the extreme risk transfer between stock market risk conduction in the direction of the great the effects of the subprime mortgage crisis. Later, the international stock market risk contagion degree is enhanced, the time variation characteristics are also very obvious. From the stock market risk conduction direction: before the outbreak of the subprime crisis, the risk of the stock index mainly collects in the New York stock market, and the three major Asian stock markets, namely Chinese Shanghai, no risk conduction relationship between Hongkong stock market and Tokyo stock market. After the outbreak of the subprime crisis, change ways and direction of stock market risk conduction not only the New York stock market and other markets are formed between the two-way relationship between risk conduction, Asian stock markets also showed the risk conduction pattern close and complex, China stock market and international stock market related extreme risk is significantly increased.
These distinctive features, will undoubtedly affect the measure accuracy of portfolio risk model. In this context, the four stock index returns are combined to construct two yuan asset portfolio and multi asset portfolio, four time varying Copula-EVT model and DCC-GARCH model based on the address of the long and short positions, the establishment of VaR model and ES model, and analyzed the test by using the Backtesting method, a comparative study of the changes of all kinds of crisis risk model measurement accuracy. The empirical results show that: first, after the outbreak of the subprime crisis, a positive extreme risk related financial city significantly enhanced, diversification of investment effect is weakened to a certain extent, measure the accuracy of portfolio VaR risk value model is reduced; however, in some cases, measurement accuracy of ES model is the expected loss in the wake of the crisis to a certain extent Increase. Second, either VaR model or ES model, based on time-varying risk model of construction of Copula-EVT, the measurement accuracy is generally higher than that in the DCC-GARCH risk model. Third, the marginal distribution model, measurement model for the time-varying Copula-EVT risk model has important influence. Fourth, the risk from different types of time-varying model the Copula function is constructed, the prediction accuracy for portfolio risk is different. In general, after the crisis, the time-varying measurement accuracy of SJC-Copula-EVT-VaR model and time-varying tCopula-EVT-ES model are relatively high, this further indicates that the subprime crisis has had a huge impact on the effect of portfolio risk measure model, good asymmetric variables of heavy tailed dependent feature model shows strong advantage measure. Even so, the risk measure of portfolio assets, according to the needs The distribution characteristics of the combination and the scientific comparison study to choose the appropriate risk model flexibly.
In the economic globalization today, whether it is from the angle of time and space, the contagion of financial crisis are becoming more serious. Since the outbreak of the subprime crisis, the operating environment of financial market more perplexing, financial contagion in interaction between various markets easily. Investment portfolio in the financial crisis under the background of frequent, should pay special attention to prevent the combination investment risk. Risk assessment for the portfolio, we should try to construct a multi risk model, selection of measurement accuracy higher relative risk assessment model, and the combination of VaR model and ES model. In addition, the risk assessment of the portfolio also should be based on the dynamic perspective, because sometimes the varying characteristics with tail extreme risk conduction, so in assessing the risk of using a static class method must be careful to prevent error risk, portfolio assessment at the same time, must be timely and effectively take corresponding The stop damage measures to prevent the stock market plunge at the same time to prevent the extreme financial incidents caused huge losses to the portfolio.

【学位授予单位】：西南交通大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：F224;F830.91

【参考文献】