Abstract
High price volatility is a long-standing characteristic of world oil markets and, more recently, of natural gas and electricity markets. However, there is no widely accepted answer to what the best models and measures of price volatility are because of the complexity of distribution of energy prices. Complex distribution patterns and volatility clustering of energy prices have motivated considerable research in energy finance. Such studies propose dealing with the non-normality of energy prices by incorporating models of time-varying conditional volatility or using stochastic models. Several GARCH models have been developed and successfully applied to modeling energy prices. They represent a significant improvement over models of unconditionally normally distributed energy returns. However, such models may be further improved by incorporating the Pareto stable distributed error term. The article compares the performance of normal GARCH models with the statistical properties of unconditional distribution models of energy returns. We then present the results of estimation of energy GARCH based on the stable distributed error term and compare the performance of normal GARCH and stable GARCH.