The Study Of Financial Market Risk Measurement Based On Copula Theory And GPD Model

Nowadays, the relationships between the energy and financial markets are gettingincreasingly close and complex because of the rapid development of the energy andfinancial markets,deepening of energy innovation,economic globalization and financialinternationalization. The2007-2009financial crisis has shed light on the importance ofcontagion and systemic risk, and revealed the lack of adequate indicators for measuringand monitoring them. The recent european debt crisis has also raised many questionsregarding the meaning of negative impact on the global economy, meanwhile it presentsnew challenges to the energy and financial risk management as well.This paper firstprovides an overview of developments, methodologies, and applications of various risk(VaR), then key methodologies of VaR estimation and evaluation are discussed andcompared. A approach with extreme value theory (EVT) technique provides aframework to characterize the behavior in the tails of a distribution.This dissertationfocuses on observations beyond a particular threshold are considered excess and can beshown in the limit to follow a Generalized Pareto Distribution (GPD). In GPDestimation a likelihood function is maximized over the excess observations. ApplyingEVT to the energy and financial risk management can make up for the lack of VaRmethodology, therefore could estimate the energy and financial extreme events causedby the extreme risk more accurately.In recent years, correlation analysis has drawn more and more attention in themodern financial analysis.Linear correlation coefficients and conditional correlationsboth can be misleading or reveal little about the underlying nature of the dependence.Instead of relying on correlation or conditional correlation measures, we then rely onthe Copula approach to detect how the dependence between markets changes over time.Tail dependence functions are a more general tool for analyzing extremal dependence.We demonstrate that the joint generating function method which describes the tailcorrelation by combining the coefficient of tail dependence and slowly changingfunction is better than the common tail correlation coefficient. We construct a series ofMonte-Carlo simulation and bootstrapping tests to estimate the statistical significance ofthe tail index difference. Most of the previous literatures have so far focused onbivariate Copula models, here we introduce multiple t Copula which simultaneouslyallows fully-general correlation structures in the bulk of a multivariate distribution and an arbitrarily high degree of dependence in the left tails. This is ideally suited formodeling financial assets which may display moderate correlation in normal times, butwhich experience simultaneous left tail events,such as during a financial crisis. Welastly emphasis on the time-varying Gaussian Copula,Rotated Gumbel Copula andSymmetrized Joe-Clayton (SJC) Copula, and briefly summarize Copula’s applications infinancial markets.In addition,the linkage between the dynamic Copula and the volatilitymodel (GARCH&SV) is examined in a non-linear regression framework. Themeasurement technology of the energy and financial assets, such as equities, derivatives,and foreign exchanges,are critical for decisions concerning portfolio allocation, riskmanagement and derivative pricing.The main work and innovations of the present thesis are as follows：Firstly, this study integrated some foreign exchange market factors into aneffective method of risk measurement on Chinese multiple foreign exchange reserve bythe Copula-ASV-GPD model. Currently, researchers usually analyze the reservecurrency structure of foreign exchange problem from the perspective of asset portfoliomodel, Heller Knight model and Dooley model. Based on the characteristic of skewed,fat tail, asymmetric, fluctution heteroscedasticity and other typical fact characteristics offoreign exchange price sequences, there are obviously several drawbacks when mostdomestic scholars adopta combination of Copula and GARCH model to measure therisk inmultiple foreign exchange reserve portfolio. Thus, in this study, a novelcombination of ASV model, which is better than GARCH model, and GPD model wasemployed to depict the single exchange rate asset return volatility and tailcharacteristics. Then the t Copula function was applied to treat with non-linearstructures among assets. Futhermore, the risk of portfolio was measured by Monte Carlosimulation.The second innovation of this thesis lies in its application of Copula intoanalyzing the dependence structure between TWII and KOSPI stock markets. For thejoint distribution of price sequences in the emerging Asian financial markets, most ofprevious literatures made the hypothesis that price sequence follows a normaldistribution and linear correlation. And in the light of foreign empirical study focus onmature Western market, there are less studies focus on emerging Asian market usingCopula. Although the " Taiwan Experience "and " Hanjiang Miracle " have been praisedas a model of economic development for developing countries and regions, there havebeen little studies on Taiwan’s weighted index and Korea composite stock price index correlation based on Copula. By employing GPD model as a marginal distributionfunction, this thesis introduced de Haan and Bootstrap method for quantifying thresholdselection, and then three Copula methods were applied to investigate the relationshipbetween TWII and KOSPI. Moreover, the fitting effect was also evaluated between thesingle and dual Copula parameter. In addition, the conditional probability was estimatedif the two market encountering extreme market risk. This study also investigated theapplications of the dual parameter Archimedean Copula function in building a jointdistribution, and finally predictted the proxima luce market portfolio risk value by therolling time window method.The third innovation lies in its application of Copula into analyzing stock indexderivatives risk measurement. For the stock index option derivatives problem, whichhas not launch in our country yet, previous domestic literatures mainly focus onnumerical methods and its introduction. However, the risk measurement of financialderivatives market, especially American basket of stock index option, are extremelylimited reported. This paper aim to measure the risk of American stock option portfolioby multivariate t Copula contrast. Firstly we apply the least Square Mento Carlosimulation (LSM) approach for approximating the value of American basket put optionsby simulation. And then using the Latin hypercube sampling technology based on twosamples, to comparative analyze American basket option including the world majorstock, such as SSE Composite Index.The four innovation lies in its application of Copula to analyze the dynamic taildependence structure between ZXI and CSCS share markets. Currently in China,time-varying Copula is still at its starting stage. The existing literatures on dynamicCopula risk measurement mainly use SJC Copula method, without consideringtime-varying Copula combined with EVT. This study aim to study the dependence ofequity markets, by using of extreme value theory, GJR-SKST distribution andtime-varying Copula.The ARMA-GJR-SKST-GPD model was used to examine themarginal distributions, while time-varying Symmetrized Joe-Clayton (SJC) Copulamodel are employed to analyze the joint distributions. The combination of both modelsproved to be useful in determining the tail dependence and calculating VaR of ZXI andCSCS portfolio through Monte Carlo simulation. Furthermore, the main findings of thisstudy illustrate that two different standard devations series are more likely to correlate with each ot her during market upturns than downturns, which has not been documentedin previous literature.