Font Size: a A A

Extreme Risk Measurement Of Stock Markets Based On ARFIMA-HYGARCH-EVT Model

Posted on:2014-03-04Degree:MasterType:Thesis
Country:ChinaCandidate:H FanFull Text:PDF
Abstract/Summary:PDF Full Text Request
With the development of financial innovation and financial derivative instruments,the trend of financial integration becomes more and more strong, and financial riskmanagement plays an increasingly important role in the stable operation of thefinancial markets. In recent years, VaR method has become increasingly prominent inthe field of financial risk management, and it has become a mainstream method inmeasuring market risk by the current field of finance. Many of the economiccharacteristics of financial markets can not be explained by the efficient markethypothesis. For example, the conditional yield of financial asset often shows thedistribution characteristics of skewness and fat tail, and the volatility of financialasset prices show the characteristics of long memory, asymmetry andheteroscedasticity, etc. This thesis uses GARCH family and its derivative models todescribe the fluctuations of the stock market and long memory characteristics, itmeans that it considers the continuing impact on future volatility by the currentinformation. As the volatility of the financial markets reflects the size of the dynamicfinancial risk, the study on financial volatility persistence is of great significance.More and more studies found that the distribution of the rate of financial returnshas the characteristics, such as excess kurtosis, fat tail and asymmetry, and theconditional volatility of financial assets has characteristics of long memory andheteroscedasticity. In order to solve the problem, firstly, we use theARFIMA-HYGARCH model to capture these typical characteristics and analyze therate of conditional return and conditional volatility of financial assets. Secondly, wemodel the tail distribution of the rate of the standardized return using extreme valuetheory. This thesis proposes a new method to select the threshold for the POT modelof EVT theory, kurtosis method, and measures the extreme risk of various markets.Finally, using the LRT andLRi ndratios of back-testing to conduct a joint test on theaccuracy and precision of the risk model. The empirical results show that both thereturn series and conditional volatility of Shanghai Stock Exchange Composite Indexand the NASDAQ Index have long memory characteristics to some degree. The ARFIMA-HYGARCH-EVT model introduced in this thesis does well in themeasuring precision of risk, which is superior than other risk models in which the taildistributions are normal distributions or student-t distributions. The model provides aquantitative tool to control risks and makes it a better risk prevention for the investors,the managers of risk management and the government to ensure the healthy growth ofthe capital markets.
Keywords/Search Tags:long memory characteristics, ARFIMA-HYGARCH model, kurtosis method, POT model, VaR method, extreme value theory
PDF Full Text Request
Related items