| In the theory and practice of capital market, a measure research of investment risk has been the focus that the academics and practitioners concern. As a basis for risk management, if risk measurement is not accurate, it will lead to failure of risk management strategies. With the growth of financial market volatility and liquidity, people are paying more and more attention to risk management. A series of changes of risk measurement methods also have taken place, there is nominal value method, the sensitivity measurement method, fluctuation method, value at risk and so on, while B coefficients and VaR values are used the most widely.In the CAPM model, B coefficient is the portrayal of systemic risk, so the research to the B coefficients of the model is one of the focuses of this article. In recent years, study ofβcoefficient focused on stability and changes in their characteristics and so on. Much of the research shows that B coefficient of the stock market renders uncertainties in the characteristics, the influencing factors include the economic cycle, interest rates and other macro-level ones, the company’s financial and other micro-level ones, also include the company’s industry and other meso-level ones. However, because of stringent assumptions of the traditional CAPM model, when the model is describing the time-varyingβ, there will be a big limitation, which also brings about a certain degree of difficulty on giving an accurate measure of systemic risk. Therefore, the academic community starts a series of research improvements; the results included ICAPM, CCAPM, LAPM, EPT, and BCAPT and so on. These are longitudinal studies on the CAPM model, and could well explain some respects of the economic phenomena, but still there are their own deficiencies. The focus of horizontal studies is to explore the B coefficient estimation methods, including recursive estimation, rolling window regression, kalman filtering Beta methods, in which Michael D. Mckenzie and other people have studied the relationship among the Australian industry beta, the Australian market index and the world market index, using kalman filtering method. And they pointed out that the relationship between the domestic market and the industry beta was more related. To accurately measure industry’s systemic risk, this thesis estimates theβcoefficient of CAPM model in kalman filtering method, introduces state-space models, so that it can accurately estimate the value of time-varyingβ.Another research focus of this thesis is the VaR method on risk measurement. With the establishment of the information system JP Morgan Risk Metrics, VaR value is increasingly attracting more and more risk managers’attention. It can summarize all the components of market risk in a digital system, including the curve of risk, differential sub-line, and volatility risk and so on, while it is very simple and convenient. With the expansion of application fields, people are demanding more on VaR value, therefore, an accurate estimate of the value VaR will be more important. In general, there are parameter estimation method and non-parametric estimation one. The traditional VaR estimation method is assumed that the income obeys a certain distribution, estimates parameters according to earnings sequence, and thus calculates the VaR value. This method has two shortcomings, one is that assumed income distribution has been constant, which is inconsistent with stock market volatility; the other is that it overlooks the very nature of financial time series, such as earnings peak fat tail, the accumulation of volatility, long memory and so on. This also brings about certain degree of difficulty on accurately portraying of financial risks. Therefore, Engle firstly established ARCH model in 1982, and introduced it into the calculation of VaR value. Thereafter, scholars continued to improve the ARCH family models. For example, Bollerslev increased q themselves-returned items in the ARCH (p) model in 1986, promoted into the GARCH (p, q) model, overcome Multicollinearity caused by the high-order ARCH model. In the same year, in order to describe sustained high volatility of the stock market, Engle and Bollerslev proposed points GARCH (IGARCH) model. In order to capture the absolute value of earnings volatility proxy indicators such as the slow decay of autocorrelation features, Baillie, Bollerslev and Mikkelsen (BBM,1996) proposed score points GARCH (FIGARCH) model, which is the expansion of conditional mean ARFIMA model in terms of conditional variance, which can portray the accumulation and long memory of financial time series volatility rate and has been widely used in the VaR measure. This thesis will introduce FIGARCH model into the calculation of VaR value, which can characterize the risk characteristics of financial time series much better.While researching the measurement of the systemic risks, the paper testifies the characteristic of the systemic risks of top ten revitalization industry based on its data, and compares the result of these two measurements. The paper is divided into six chapters, each chapter and its contents are as follows:(1) Introduction; (2) The theory of systemic risk; (3) model selection and model transformation; (4)The empirical research series of measurement method; (5,6) as the end of this article, review the analysis results, summarize the text.The principal innovation of this article are:(1) a measure of systemic risk the industry as a research object, and focus on the estimated two metrics; (2) use the of top ten revitalization industry as a positive sample of ten.There are some deficiencies in researching the scale of systemic risks. For example, the paper will be more comprehensive and persuasive if the research of non-systemic risks is added. Furthermore, the research of risks in the industrial system is medium level which can be referred to in studying the risk management. However, it only plays the role in guiding investors, which will be more beneficial if the improvement is made in the future research. |
PDF Full Text Request