Volatility modeling and forecasting in energy market

Abstract

In this work, we test several GARCH models; linear, non-linear, univariate, and multivariate models are constructed to find which one best models volatility characteristics. Another important aspect of this study is the best models for forecasting, according to loss functions. In addition, volatility spillover effects and covolatility spillover effects are also studied in our series using univariate and multivariate models. Lastly, optimal hedge ratios are discussed, and their capabilities are tested using hedging effectiveness index. Through this paper, GARCH, EGARCH and GJR models are being used with three distributions (Normal, t-student and GED), while for multivariate models we use diagonal GARCH and DCC, with the latter being used for the optimal hedge ratios mostly.