FINANCIAL MATHEMATICS SEMINAR

Title: USING EXTREME VALUE THEORY AND GARCHCOPULA METHOD TO MEASURE RISK OF PORTFOLIO

Abstract: One of the most popular tools in risk management is Value-at-Risk (VaR). This tool provides us a powerful measure to estimate the portfolio’s worst loss over a given time horizon and at a given confidence level. In most research, the distributions of financial assets are assumed to be the light tail, e.g., normal distribution. Presently, it is widely recognized that the returns from financial assets are not normally distributed. These returns consist of heavy tails (fat tails) on the distributions, i.e. extreme events occur with higher frequencies. In this research, we compute VaR for heavy-tailed
distribution of a portfolio by utilizing a method that combines Copula theory and Extreme Value Theory, and also GARCH models. This method is applied to a portfolio of stock investments consisting of CTG, MSN, VIC, and VNM. The portfolio VaR is then estimated by using the Monte Carlo Simulation approach. Backtesting methods are used to determine the goodness of fit of an approach. We infer from the results that the GARCH-EVT-Copula approach gives better performance. Especially, the GARCH-EVT-t-Student Copula and Clayton’s Copula outperform all other GARCHEVT-Copulas as well as conventional methods like Historical Simulation and Variance Covariance methods.

Speakers: Nguyen Ngoc Phung; Nguyen Hung Quang Khai; Tran Hoang Phi.