The Research Of Inter-Industry Dependent Structure During The Stock Market Crash

With the development of economic globalization and financial integration, financial markets growing interdependence of various sectors, financial risk is also increasingly complicated and diversified. On June 12,2015, the china securities regulatory commission has issued internal notice to brokers, suspended the otc endowment new port access, and brokers began to clear otc match. It is rapidly after the Shanghai composite index hit a 5178-point and adjusted, to July 9,2015 stock index fell to 3373 points, fought a bitter "crash". So far China has erupted nine crashes, every time crash lead to huge losses to investors, and been compared with the foreign stock market crash, crash in China has its own features, such as short cycles,large average amplitude. A crash is different from other general share price volatility is that it has the sudden and destructive, correlation, the feature of uncertainty. Crash is easy to cause the financial crisis, the economic crisis, and even serious catastrophic consequences of social unrest. In contemporary, with the in-depth development of economic, financial innovation, industry stocks in the stock market is more and more has the characteristics of oneness, correlation, especially when the financial crisis this extreme risk happens. Also it is because the stock in extreme cases has a strong correlation, so easy to trigger a comprehensive stock market boom collapsed. So the correlation between stock and its related structure is more and more realistic. Because of the financial sector stocks in the financial risk of extreme cases of strong correlation, on the one hand, in the field of risk management, the yield loss of the financial industry would quickly spread to other industries, has a strong contagion effect, triggering system risk, so the stock market industry correlation can provide advices and guarantees for risk measurement and management. On the other hand, in the field of investment, according to the Markowitz’s theory, investors in the selection and combination, assets need to consider the correlation between the assets, asked to choose smaller correlation of assets. So the correlation between stock market industries also can guide investors to asset allocation. Study of assets structure can also be applied to many aspects, such as asset pricing, insurance pricing etc., Here No Longer Say.In this paper, based on the reference of forefathers’ research, we have qualitatively and quantitatively analyzed the latest crash. Especially the quantitative analysis of the block, at present in view of the quantitative analysis is the extreme risk, but a crash as a kind of extreme risk, this aspect of the quantitative analysis is relatively less, especially for the crash. About the crash, we only during a crash, linkage effect between industry groups showed strong quantitative analysis. For measuring assets correlation can be divided into linear correlation and non-linear correlation, so this article on the relevance of this crash model considering the linear correlation between industry groups and considering the nonlinear correlation. Data using csi 300 index of four big industry index of mixed frequency. Mixed frequency is that index according to the sampling data with different frequency. In mixed frequency according to the main of this paper includes five minutes in the daily stock market returns and yield in the day the two frequencies. Five minutes of stock returns is a kind of high frequency data, daytime yield is low frequency data. In terms of linear correlation model, the high frequency data has the advantageous superiority, because of the high frequency data is days, it is daytime data contains more information than the low frequency, thus it depict correlation coefficient matrix, more precise. So we consider using high frequency return on assets data, to depict the linear correlation, including model selection is HAR-type model, and according to the thought of GARCH-DCC model, estimate the realized variance covariance matrix. Also considering the linear correlation cannot fully measure of asset correlation, and nonlinear correlation measurement based on copulas connect theory research in recent years, so we consider filtering out the linear correlation of the assets and using copulas related theory to study the nonlinear correlation of standardized residuals. Especially for the combination of multiple assets, we need to use higher dimensional copula connect function, due to the difficulty of higher dimensional copula connect function estimation, we consider the vine copula structure theory is used to decompose copula connect function. Finally, we considered these correlations may be varying over time, so we considered on the basis of the vine copulas establishing time-varying copula model. On the choice of time-varying driver model, this paper choose the GAS score (generalized autoregressive model) model. In econometric model research is a high frequency of data and combining the vine time-varying copula, so this is first studied as an innovative point in this paper.Finally the empirical results show that based on telecommunications, finance, energy and consumption, the linear correlation between the four industry has significantly increased during the crash, and nonlinear correlation of financial and consumer significantly increased; do not know whether to increase, such as telecommunications and consumption, there are constant, such as telecoms and energy.