Multivariate GARCH models

In this paper we examine the usefulness of multivariate GARCH models to estimate Value-at-Risk (VaR) and spillover effect using a portfolio of returns in the OPEC and WTI oil spot market. In this procedure first we estimate conditional covariance matrix using multivariate GARCH models, results show that in multivariate GARCH models, although CCC model estimate variance matrix well with utilize more complete information of correlation matrix. Also we detect extreme risk spillover effect between the two oil markets from existence covariance between variable. The tests showed the importance of time varying correlation in risk portfolio management. The estimated Value-at-Risk represents the superiority of CCC to other models. The distributional assumption has large impact on VaR estimation. These results are valuable for anyone who needs to evaluate and forecast the risk situation in international crude oil markets.