Modeling of electricity spot prices for derivative valuation : capturing volatility clustering and price jumps

Stylized features of the electricity market are non-storeability, strong seasonality, high volatility and extreme events where the electricity spot price exhibits spiky behaviour and volatility clustering. Extreme events appear when e.g. a power plant is unexpectedly closed down, or temperature drops significantly, and the spot price volatility may exceed mean price by several orders of magnitude. In this thesis, we build on the conjecture that the electricity spot price is composed of two mean-reverting processes, a Gaussian process driven component and a jump component. The Gaussian component represents the regular, relatively small-scale changes while the jump part represents the occurrence of extreme events. We identify these extreme events as a daily spot difference outside two standard deviations of the mean. We compare the Ornstein-Uhlenbeck process and geometric Brownian motion for modeling the Gaussian component, and the self-exciting process and compound Poisson process for the jump component. Through visual analysis and Monte Carlo simulations, results indicate that the combination of an Ornstein-Uhlenbeck and self-exciting process most accurately represents electricity spot price dynamics, particularly in capturing mean reversion and volatility clustering. The findings also highlight the limitations of geometric Brownian motion in this context and suggest that volatility clustering does not affect the price of derivatives with a one year maturity, in a significant manner.

Author dc:creator

• Kári Georgsson 2000-

Contributors dc:contributor

• Háskólinn í Reykjavík

• Fjármálaverkfræði
• Meistaraprófsritgerðir
• Raforka
• Verðlag
• Financial engineering
• Electricity
• Spot prices

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/1946/48706

OAI identifier oai:identifier

oai:skemman.is:1946/48706