Font Size: a A A

The Analysis Of Portfolio Of Foreign Exchange Based On Copula Theory

Posted on:2014-06-25Degree:MasterType:Thesis
Country:ChinaCandidate:Y S WangFull Text:PDF
GTID:2309330482471563Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
As to investors, the investment risk may be reduced effectively when they invest in a variety of financial assets with lower correlation, so how to choose the optimal portfolio has become the focus of investors. The traditional portfolio risk management models have many deficiencies, such as the methods of risk measurement, the hypothesis of normality and linear correlation. In order to overcome these faults, we should not only apply more scientific methods of risk measurement, but also construct a more reasonable joint distribution, both of these plays a crucial role to the measurement of the portfolio risk and the selection of the optimal investment strategy.At the point of risk measurement, VaR has become a widely used measurement method and also is the focus of the financial risk management; on the other hand, the Copula theory supplies a convenient and scientific way to construct a joint distribution of the portfolio. And the joint distribution can base on the assumptions of non-normality and non-linear correlation of the financial assets. So this method can overcome many disadvantages of traditional risk management model.The main content of the paper is the analysis and measurement of portfolio risk with multiple financial assets. The paper bases on Copula theory, and utilizes VaR as the risk measurement, applies Copula function, GARCH model, SV model, VaR and Monte Carlo method to solve the non-normality and non-linear correlation problems of multiple financial assets, so it provide a new method for the selection of the optimal portfolio and the measurement of investment risk. The first part of study of the paper relate to a portfolio with four major foreign exchange assets in China’s foreign exchange market. First, we make a comparison between GARCH model and SV model to establish a model of marginal distribution for the return of risky asset; then apply the PC algorithm to estimate the DAG, which represents the dependence structure among variables. And a new VaR prediction model is constructed by PCCs, namely Gaussian DAG Copula, which can describe the portfolio dependence structure better. Based on the Gaussian DAG Copula, we apply the Monte Carlo methodology to study VaR of a portfolio of four major foreign currencies in China, and the validity of the prediction model is proved by back testing. At last, we give different investment weight to four foreign exchange assets and estimate VaR in different confidence level base on Gaussian DAG Copula model. The second part of study of the paper relate to a portfolio with seven major foreign exchange assets in China’s foreign exchange market. A mixed C-vines copula model based on Pair Copula Constructions method was used to construct the practical distribution of multiple asset returns and joint distribution function of dependency for studying portfolio. The model, which allows the variables to be ordered according to their influence and chooses the bestfamilies of copula functions for every Pair Copula can describe the dependence structure of the multiple asset returns better. Based on the mixed C-vines copula models, we apply the Monte Carlo methodology to study VaR of a portfolio of seven foreign currencies in China, and the validity of the model is proved by empirical analysis. Finally, we also give different investment weight to seven foreign exchange assets and estimate VaR in different confidence level base on mixed C-vines Copula model. The two parts of the study show that the established model can describe the correlation structure better, and more scientific and accurate in risk measurement.
Keywords/Search Tags:Portfolio, Pair Copula, Gaussian DAG Copula, mixed C-vines Copula, VaR
PDF Full Text Request
Related items