Modelling energy markets and pricing energy derivatives

The main objective of this thesis is to provide an empirical assessment of the popular methodologies for modelling the underlying spot price dynamics in energy markets. After a brief introduction in the alternative forms of derivation that may be used for speculative and risk management purposes in energy markets, we assess the performance of the standard Black's framework in modelling energy prices. For the first time in the literature we use a powerful and realistic data set which covers oil, gas and electricity markets and tests the appropriateness of the Geometric Brownian Motion process to explain the observed dynamics of the spot prices in these markets. We also provide spreadsheet based computer algorithms to price popular energy derivatives based on the Geometric Brownian Motion specifications. In Chapter-3 we try to accommodate observed stylised facts in the spot price behaviour, namely mean reversion and jumps. For the first time in the literature we test a jump diffusion model, and a mean reversion jump diffusion model against our broad data set and compare the findings to the Black's Geometric Brownian Motion specifications. In Chapter-4 we use a forward curve approach as an alternative-modelling framework to the spot price models. Based upon an almost proprietary data set of historical forward curves, we determine the number of independent factors that are needed to model the forward curve's dynamic evolution. After carrying out principal component analysis on historical forward data a threefactor-model emerges as the most appropriate for energy markets in general. The first factor being the volatility (level effect), the second the smile and the third sesonality. Finally in Chapter-5 of the thesis we compare the ability of spot models (Jump Diffusion and Mean Reversion Jump Diffusion Model) and forward curve based models to price WTI options. The results show that the Jump Diffusion Model is the best model as the option prices given are very accurate in comparison with the other models and closest to the market observed options prices.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

City University London

Year dc:date.issued

2003

Author dc:creator

• Gkinis, S.

• HG Finance