Extreme Value Study On Financial Returns Time Series

In modern economic activities,the factors influencing financial activities are complicated. In financial market,investors and financial institutions are faced with risks for extraordinary volatility of assets prices which arises from the effects that economic and monetary policies and reforms on political regime in the countries have on. How to forecast reasonably the risks for yield and loss of assets is therefore what the investors and national financial supervision and management institutions focus on. Financial assets returns as an important index for prices variation has received concern from many specialists and scholars in financial and statistical industries,and a lot of research in theory and application are about financial returns.Financial Asset price volatility.especially the large change for stock price such as rise and drop wildly is extreme returns in terms of returns. In view of probability distribution,studies on occurrence laws for extreme returns is to explore the tail behavior of returns probability distributions. Extreme value theory and methods can only deal with the tail behavior for returns series,and it primarily uses generalized Pareto distributions to describe the fat tails of financial returns. Furthermore,only the data exceeding the threshold are used in statistical extreme value analysis,and most returns data can't be exploited enough,so the volatility and autocorrelation are unable to be described better.Financial practical analysis shows that not only has the asset returns fat tails,but also the volatility is clustered,i.e. large volatility is followed by large ones and small volatility is followed by small volatility,while response of volatility to the shocks of good news and bad news are asymmetrical,that is the bad news shocks have larger effects than the good ones. All these features can't be measured by extreme value theory and methods. It is well-known that the financial asset price volatility arises largely from the investors's prediction change which is attributed to shock effects of market news. Just as Lon-gin(1997) said, extreme movements are associated with both little tremors like market adjustments or corrections during ordinary periods,and also earthquake-like stock market cra.shes.bond market collapses or foreign exchange crises observed during extraordinary periods, therefore it is insufficient to explore the returns.tails for probability distributions only by applying the extreme valuetheory. We need to study returns series volatility,and from the volatility models analyze the statistical rules for the occurrence of extreme returns.The research literatures on finance returns presented some models which describe the returns volatility,and among these Engle's ARCH,Bollerslev's GARCH,Nelson's EGARCH and TGARCH by Glosten,Jagannathan and Run-kle are famous. These models in the related literatures are usually used to fit the volatility features for financial returns such as exchange rate and stock index returns,and few are exploited to explore the relationship between volatility and extreme returns,while these two are closely related with each other.This paper studies the tails behavior of returns from the returns volatility-models,and further explore the occurrence rules for extreme returns through analyzing the effects of good news and bad news shock on volatility. The method here combines extreme value theory with non-linear time series models,and discusses tail features for the marginal distribution by the time series characteristic. It has not only practical significance,but also is an exploratory research in theoretical research on financial time series.The paper considers the volatility and extremes of financial returns series as research objects,and presents a statistical modelling method by combining extreme value theory with non-linear time series models, residuals series are obtained by fitting returns volatility models in order to compare the tails behavior of innovations and returns. Practical analysis shows that the modelling method in the article is able to...