Determining Value at Risk Based On Copula and GARCH

Abstract

Copula model has become more popular in risk dependency structure modelling because copula able to
completely model dependency structure of multivariate distribution linearly or non-linearly. To estimate Value at Risk
(VaR) of a portfolio, one can model a marginal distribution of risk position and choose the right parametric copula to
model dependency ratio and goodness of fit test matter from the new model. The objective of this research is to answer
these questions. How to determine the value of VaR with Garch – Copula proxy. Data used are Central Asia Bank
(BBCA), Waskita (WSKT) and Semen Indonesia (SMGR) stocks collected from December 1st, 2014 to November 30th,
2016 which consist of 516 of trade days. The application method of Garch-Copula as follows: First ARIMA and Garch
modeling, then copula function is used to model the compound distribution. Based on the research result, VaR’s
computation with 90% level of confidence, stocks return of BBCA and WSKT in 90% level of confidence is
0,007013151. That means if an investor invested 1.000.000 rupiahs in BBCA and WSKT portfolio stocks the investor
would likely experience maximum loss at 7.013,15 rupiah in a day.
Keyword: Value at Risk, GARCH, Copula.