Modeling electricity prices: jump diffusion and regime switching

Abstract

In this paper we address the issue of modeling spot electricity prices. After summarizing the stylized facts about spot electricity prices, we review a number of models proposed in the literature. Afterwards we fit a jump diffusion and a regime switching model to spot prices from the Nordic power exchange and discuss the pros and cons of each one.Electricity price; Jump diffusion; Regime switching; Seasonality;

Modeling regime switching in day-ahead market prices using Markov model

“…9 , 23% of cases ARIMA method outperforms and 77% cases HW method beats its competitors. This behaviour is also consistent with (11). In addition, both Fig.…”

“…9. Entering daily absolute error series for all forecasting models as input data to Markov switch model we can get transition probability matrix (11). 1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 171 181 191 201 211 221 231 241 251 261 271 281 291 301 311 321 331 341 351 361 Absolute Error …”

The effect of missing data on wind resource estimation

“…These parameters can be estimated through nonlinear regression methods. Weron et al (2004) uses wavelet decomposition, breaking the time series into small and large frequencies, and then estimating the resulting cycles using a sine function, while Cartea and Figueroa (2005) uses Fourier transforms to estimate the annual cycle of the deregulated UK market. The above papers do not model daily patterns and opt to use a single day as the data resolution level, aggregating the hourly observations accordingly.…”

“…Weron et al (2004) shows that, by choosing an appropriate method to remove spikes from the observed data, the remaining observations are much better suited to the assumption of normally distributed increments.…”

Fitting semiparametric Markov regime-switching models to electricity spot prices

“…Rambharat et al (2005) and Misiorek et al (2006) extended the TAR model by an exogenous variable (temperature recorded at the same time as the maximum price of the day and load). For MS models Weron et al (2004) and Bierbrauer et al (2007) compared the model introduced by Huisman and De Jong (2003) using different distributional assumptions for the spike process, while Misiorek et al (2006) compared the regime-switching model with other model classes evaluating their short-term forecasting power. Extensions were for example given by Kosater and Mosler (2006) who introduced a third regime to account for the occurrence of negative spikes and by Mount et al (2006) who allowed the transition probabilities to depend on the load and/or the implicit reserve margin.…”