| Economic stability and development is an important issue related to national prosperity and social prosperity, and for the prediction and assessment of potential risks has also been the focus of academic attention and financial regulators. With the progress of development of the Internet and computer technology, especially since the 1970 s, the global financial system has changed dramatically. According to the annual report released by the OECD forecast in 2014, the world’s developed countries are ready to quit during the 2008 financial crisis, the launch of the economic stimulus plan will cause global financial markets, and will face "enormous turbulent times." Meanwhile, the financial sector innovation is increasing dramatically, and financial derivatives growth is explosively increasing. Also, in the financial sector transaction, no matter the transaction speed or volume, has increased unprecedentedly. The uncertainty and complexity of the financial markets has reached unprecedented levels as these factors are intertwined. Meanwhile, with the enhanced integration of financial markets trends, fluctuations between different financial bodies are mutual infection, also, dependence between financial markets are unprecedented enhanced. In one hand, technological advances and global financial integration trends are promoting a great deal of social progress and convenience to the people’s economic and daily life. On the other hand, technological advances and global financial integration trends has also put higher requirements on the financial risk management and financial risk regulation. Risk management means a financial institution or business in accurate identification and measurement of market risk on the basis of competitive advantage and in accordance with its risk appetite, using a variety of tools and techniques of risk aversion and prevention of metastasis(diversification, hedging, insurance) and retention(risk pricing and risk capital configuration) process. The VaR method based on rigorous due to statistical theory based on the integration of a variety of risky assets into a single number, with a clear and understandable advantages result, now the method has become the mainstream of financial market risk measurement methods. In addition, because the portfolio can effectively spread the risk, the risk of the investment portfolio performed relative to the single-asset investment has a broader application of risk, portfolio of research studies more meaningful than a single asset. But the traditional VaR calculation method to the efficient market hypothesis(Efficient Markets Hypothesis, EMH) as a precondition is calculated based on compliance with the linear correlation between the measurement of financial assets. However, there are a large number of studies have shown that there is a time series of financial revenue side, fat tail characteristics, the presence of earnings volatility clustering and leverage, etc., and internal investment portfolio does not meet the simple linear correlation. This suggests that the traditional VaR method does not accurately and efficiently calculate portfolio risk, we must find a more appropriate model for the distribution of financial returns and correlations were fitted.Based on these facts above, we select the closing index data with time period of April 1, 2005 to April 2, 2013 in the Chinese CSI 300 Index(HS300), US S & P index(SP500), Japan’s Nikkei 225(NKY), United Kingdom The FTSE 100 index(FTSE) to build a portfolio. Copula function using combination Garch model portfolio risk measure conducted. Firstly, after analysis, model building GJR-Skewt to capture the portfolio yield biased sequence features, thick tail characteristics, as well as the leverage effect, re-use T-copula function can portray thick tail distribution fitting combination of assets relationship between the risk of the portfolio for the future conduct of-sample dynamic prediction(out-of-sample) of. Then, we extract the effective parameters based on historical data, the use of Monte Carlo simulation(Monte Carlo simulation, MC), generates a pseudo-random number between compliance with the relevant parameters, and then reverse by GJR-Skewt model residuals obtained distribution a few days, and the samples were outside the VaR forecast. Meanwhile, in order to improve the prediction accuracy of VaR, this paper uses one day ahead prediction method(one-day-ahead forecast) rolling time window method. Again, this article use the same algorithm to calculate the volatility model several other groups(ie: Garch-Guassian, GJR-Guassian, GARCH-skewt) as the marginal distribution VaR prediction results when, as a comparative study. Finally, the results of in-depth empirical analysis and results of the analysis carried out further policy recommendations. This article expand the narrative from the following four parts to: The first part, describe the general situation in financial markets, the volatility of the financial markets come and heightened vulnerability, dependency relationships in different markets reinforce the status quo, and then mentioned the importance of financial risk management. On this basis, the paper selected portfolio risk is calculated for the study, describes the theoretical and practical significance of this study. The second part of the research is literature review. Many research results at home and abroad on portfolio risk management to sort out, to extract the development of portfolio risk management theory and empirical research, in order to achieve the investment portfolio risk management research has a clear context and cognition. The third part is the theoretical reliability analysis. Detailed in the insurance value of the theoretical and computational methods, Copula function and the corresponding marginal distribution function theory Garch introduction, Copula methods in Monte Carlo simulation method to calculate VaR and specific implementation steps, the last is to test the model. The fourth part is the empirical analysis section. We select a different country stock index return volatility typical composition of the portfolio empirical data, and empirical results combined with the reality of the financial risk to the in-depth analysis of the market. Select a different distribution models Garch and GJR same test, the test results were compared to obtain useful conclusions. Finally, the reliability of the model reliability tests to verify the above mathematical model is valid. The fifth part is policy recommendations. On the basis of theoretical analysis and empirical results, settled on the actual situation of China’s financial markets, financial regulators and financial market participants propose appropriate policy recommendations.The results show that: first, from the time series, the CSI 300 Index(HS300), US S & P index(SP500), Japan’s Nikkei 225(NKY), the UK FTSE 100 index(FTSE) returns fluctuation map obviously when it can be seen from the figure the CSI 300 Index(HS300), US S & P index(SP500), Japan’s Nikkei 225(NKY), fluctuations in the performance of the UK FTSE 100 index(FTSE) yields a day degeneration, sudden and clustering features, do not satisfy the normal distribution assumption, but there is a significant "Rush fat tail" features, and leverage. Secondly, a single asset fitting model GJR-skewt model is more effectively fit these characteristics samples stock index compared with the other modelslike GARCH-skewt etc.Again, under the assumption of a normal distribution in the residuals, leverage can portray GJR model does not clearly show better than ordinary Garch VaR model predictive ability, suggesting that fat tail adequately describe the conditions and income characteristics of the VaR biased The prediction accuracy has a more significant impact, therefore, be able to describe the condition returns fat tail and biased characteristic GARCH-skewt model would be a relatively good model; Garch-Guassian model predicted VaR and expected number of failures failures is too small compared to the number, indicating a greater degree of risk in this model and the actual departure from the market, there is a tendency to underestimate the risks. |
PDF Full Text Request