Facoltà di scienze economiche

The electricity price modelling and derivatives pricing in the Nord Pool market

Volpe, Valeria ; Barone-Adesi, Giovanni (Dir.)

Thèse de doctorat : Università della Svizzera italiana, 2009 ; 2009ECO008.

An in-depth analysis of Nord Pool electricity price has been outlined. This market is the oldest and one of the largest European electricity exchange, allowing to observe a consistently long time series. The main contributions of this work are: (a) A wavelet-based algorithm, named Rα-WTMM, to automatically detect spikes on a time series. The WTMM algorithm, used in signal processing to detect... Plus

Ajouter à la liste personnelle
    Summary
    An in-depth analysis of Nord Pool electricity price has been outlined. This market is the oldest and one of the largest European electricity exchange, allowing to observe a consistently long time series. The main contributions of this work are: (a) A wavelet-based algorithm, named Rα-WTMM, to automatically detect spikes on a time series. The WTMM algorithm, used in signal processing to detect singularities, has been modified in order to selectively identify the spike locations. Spikes are one of the most distinguishing features of the electricity commodity and they cannot be ignored when dealing with electricity modelling. (b) A spot price model with a regime-switching configuration, between a diffusion and a non-parametric component defined by a double bootstrap procedure, has been choosen. The adopted model enables to reproduce the main features of the observed price trajectory, such as price-dependent volatility, mean reversion and spikes, without adding any jump component; it achieves a high level of freedom in the possible trajectories shapes, although with very few parameters: for instance both positive and negative spikes are allowed, and multiple and different length spikes are also contemplated. Moreover, thanks to the nice probabilistic properties of the chosen bootstrap, a sensible Markovianity is preserved. (c) The pricing of American options, starting from Monte Carlo Least Square procedure, has been developed and applied to the proposed regime-switching model. By introducing an Ornstein-Uhlenbeck Brownian Bridge, continuous sample paths, needed to perform backward induction, have been consistently generated. Further, to reduce the high memory storage required by dynamic programming, the underlying process has been simulated in a backward time-moving direction, generalizing this backward simulation technique to any random processes. (d) An extensive empirical analysis of the Futures term structure aimed at understanding the futures risk premium behaviour has been provided. A Futures pricing formula based on reservoirs optimal allocation, given the predominance of the hydropower generated electricity, has been derived. Each Futures contract has been defined as a linear combination of an Asian option and an American option, both written on the electricity spot price. The market data support this hypothesis, suggesting interesting economic interpretations.