The joint modelling of energy forward curves

Analogous to yield curves in fixed income markets, commodity forward curves reflect market participants’ views on future price levels and are the main inputs for pricing, risk management, and project evaluations. This thesis focuses on a factor-estimation method incorporating the joint dynamics of multiple commodity forward curves for a term structure model. First, we introduce PCA on PCA as the main tool to formulate the factorvolatility functions with commonality, extending the Heath-Jarrow-Morton (1992) model for multi-commodity modelling. The proposed factor estimation method is intuitive and easy to implement as a direct extension of ordinary PCA. We demonstrate the estimation procedure of common eigenstructures and analyse the loadings to give economic interpretations to the identified common factors. Second, we apply our common factor model for the pricing of commodity spread derivatives. Our contribution includes the option pricing formula when one of the underlying assets is denominated in foreign currency units, which has not been considered carefully in previous commodity literature. Third, we analyse hedge ratios and their effectiveness with and without assuming common factors across multiple forward curves. Empirical studies on each topic are carried out for European energy commodities or marine fuel products. We find that the existence of common factors lowers the option prices and improves the minimum-variance hedge where appropriate. These results have important implications for commodity producers, traders and financial institutions that trade highly dependent underlying assets: neglecting common factors may result in economic losses caused by mispricing financial contracts or mistreating inherent risks.