Applying well-specified quantitative models in energy forecasting and hedging continues to be challenging. Even well-established approaches require careful adaptations. In this issue of The Journal of Energy Markets we see the practical importance of advanced modeling techniques across four distinct application areas. Looking at oil derivatives, the integration of regional spot oil markets, electricity spikes and the linkage of the carbon price to prices in the main energy markets, the value of adapting generalized autoregressive conditional heteroscedasticity (GARCH) volatility modeling, regime-switching, copula and conditional correlation models is empirically demonstrated.The first paper in the issue, "Hedging crude oil derivatives in GARCH-type models" by Tak Kuen Siu, Roy Nawar and Christian-Oliver Ewald, undertakes an extensive investigation into the empirical performance of hedging strategies based on Delta and Delta-Gamma for crude oil options in a GARCH model environment. Particular attention is paid to studying the impacts of the conditional heteroscedasticity and the conditional nonnormality of the GARCH innovations on the option prices and the performance of these hedging strategies. The value-at-risk and the expected shortfall are appropriately assessed on the terminal values. The hedging results show that GARCH with shifted gamma innovations systematically outperforms the benchmark models (namely, GARCH with normal innovations and the Black-Scholes-Merton model).
The issue's second paper, "Extreme dependence of China's and the world oil market: empirical evidence and implications" by Xiaoqian Wen, Yu Wei and Dengshi Huang, presents an international focus on the implications of China's increasing dependence on oil imports. The paper examines the extreme dependence between global oil markets and the Chinese oil market through the use of time-varying copulas. Looking at Daqing (China), WTI (the United States), Brent (North Sea), Dubai (Middle East) and Minas (Asia-Pacific) spot oil prices, the authors observe that the dynamic conditional correlation Student t copula effectively describes the dependence structure. The authors subdivide their sample period (2007-11) into three subperiods and find that extreme dependence increases more during periods of global economic recession.
Our third paper, "Models for short-term forecasting of spike occurrences in Australian electricity markets: a comparative study" by Michael Eichler, Oliver Grothe, Hans Manner and Dennis Tuerk, also looks at extreme observations but in the contextof intraday-frequency real-time electricity prices. The authors consider the problem of short-term, ie, half-hourly, forecasts of spike occurrence in the Australian electricity market and develop variations of a dynamic binary response model, extended to allow for regime-specific effects, as well as an autoregressive conditional hazard model. The proposed models use load forecasts and lagged log prices as exogenous variables, resulting in important recommendations for spike forecasting in practice.
Finally, in our fourth paper, "Carbon price volatility and financial risk management", Perry Sadorsky uses multivariate GARCH models to model conditional correlations between carbon prices, oil prices, natural gas prices and stock prices. Compared with the diagonal model or the dynamic conditional correlation model, the constant conditional correlation model is found to fit the data best and is used to generate hedge ratios and optimal portfolios. Carbon appears to be useful for hedging natural gas. Hedge ratios and optimal portfolio weights vary considerably over the sample period, however, indicating that financial positions should, as expected, be monitored frequently.
Overall, this issue of The Journal of Energy Markets presents many practical insights into using (empirically validated) advanced modeling techniques and demonstrates potential applications across a wide range of traded energy products.
Derek W. Bunn
London Business School
Extreme dependence between China’s oil market and the world oil market: empirical evidence and implications
Models for short-term forecasting of spike occurrences in Australian electricity markets: a comparative study