Simon Spencer

Abstract

This thesis reports three research studies into aspects of the term structure of energy and related commodities.

The first study considers what we can learn about a physical commodity by studying its hedging characteristics. I use a hedging study to shed light on important properties of ethanol (a developing market) and corn (a mature market). My three primary contributions are empirical, with implications for all storable commodities. I identify important differences between regularly cited data sets for spot ethanol prices and clearly explain these differences in terms of the data collection methodology. The data selection implications for hedge effectiveness are found to be substantial. Having provided clarity on the data, I find consistent evidence to support the simple is better hypothesis in relation to futures hedging models. Finally, I caution against complacency, as my methodology reveals how extreme events can lead to biases which reduce the hedge effectiveness at the very times when effective hedges are most needed.

The second study investigates the relationships between different types of OPEC announcements and term structure variables (level, slope and curvature) for WTI crude oil futures. I find that agreements to increase (decrease) pro- duction are positively (negatively) associated with changes in oil price levels in futures markets. This finding is consistent with recent research. For the first time in the literature, I evaluate announcements which signify ‘no change’ in production quotas. In particular, a ‘no change’ announcement may be due to a unanimous agreement to maintain quotas unchanged, or it may be due to a failure to agree on a change in quotas. I find the term structure effects (on slope and curvature) of the two types of announcement to be statistically significantly different from each other. My results suggest that markets are more optimistic following agreements to maintain production, and more cautious following a failure to agree.

The third study demonstrates that Nelson-Siegel parameters can predict WTI futures holding period returns for crude oil futures better than forward spread or principal components. The decay factor in the Nelson-Siegel model, which provides a measure of market expected mean reversion, is significant both in sample and in an out of sample forecasting exercise. This predictability is not diminished by augmenting the predictive regressions with macroeconomic indicators or commodity specific predictors. In addition, the possibility that the decay factor is proxying for curvature is rejected. In this formulation, hedging pressure does not have the expected role in determining returns, reinforcing questions over the conventional proxies for hedging pressure. The predictability attributable to the Nelson-Siegel factors does not appear to represent a risk premium and can therefore be tentatively regarded as an anomaly.