In deregulated markets, electricity prices are typically characterized by four key features: seasonality, mean-reversion, the possibility of large downward or upward unexpected spikes, and volatility clustering. We propose a time-series price model with skewed and leptokurtic shocks, which displays all four features above. Importantly, the model fundamentally relies on a continuous and monotone transformation of a one-dimensional normal random variable, which is of considerable interest when the electricity price model is only a part of a larger problem whose solution requires the use of numerical integration and/or simulation techniques. Using a maximum likelihood approach, we compare the proposed model with other specifications of the electricity price dynamics. The estimation is done with data from Nord Pool Spot, NYISO and the United States EIA. The results reveal that the proposed model provides an adequate fit and summarizes well peak period electricity price data.
Published February 2015 , 20 pages