| dc.description.abstract | Facing the losses and the harmful agents that investors must suffer. The task that we
identify, measure and control risks in order to prevent and minimize these risks alway
important. The topic \Estimating Value at Risk (VaR) of portfolio by GarchCopula method" introduces VaR as a tool to predict in advance the market loss value
of the portfolio and assets, including GARCH-Copula moldel that bring high accuracy
compared with traditional VaR methods, helping organizations and investors can forecast
the level of portfolio losses and perform hedging.
VaR is a very common measure of loss, having a central role in risk management, is a
simple measure but difficult to estimate. In this thesis, we consider estimation of VaR
on portfolio include three stocks listed on the Vietnam stock market by using GARCHCopula approach. In our analysis, GARCH type models are used to filter the margins
while the copulas are used to link these margins together into a multivariate distribution.
Specifically, GARCH model and GJR-GARCH model are covered with normal, student-t,
and skew-t distributed innovations. As for the copula, we used two Elliptical copulas
(Gaussian and Student-t) and three Archimedean copulas (Clayton, Gumbel and Frank)
in our analysis.
Empirical results indicate that GARCH-Copula approach can be successfully applied to
estimate VaR. To more specific, we illustrate that GJR-GARCH model with skew-t distribution provide the best fit for the margins. As for modeling the dependence structure it
turns out that Student-t copula has the best performance among the five estimated copulas. Finally, estimating VaR of portfolio by GARCH-copula method at three confidence
levels.
Key words: Value at Risk, GARCH model, GJR-GARCH model, Gaussian
copula, Student-t copula, Gumbel copula, Clayton copula, Frank copula | en_US |
