Modeling spike and drops dependence in european electricity markets

Abstract:
Electricity and gas prices are characterized by sharp increases or decreases in their price, followed a few days later by an equally large decreases or increases. These phenomena are called spikes or drops and are causing headaches for system developers and planners. One of the explanations for the spikes are sudden increases in demand (eg cold weather) and/or decreases in supply (eg emergency shut down of a power plant) and vice versa for drops.



Recent research has suggested that spikes and drops can be accurately described by statistical (Hidden Markov) models. This paper fits several Hidden Markov models, including some novel ones, to European electricity (EEX, PowerNext, APX Power UK & NL, Nordpool System & Sweden) and gas (TTF and NBP virtual hubs) markets in order to analyze the regime switching nature of the prices.



We then proceed by studying the smoothing distribution of the regimes, ie. the switching behavior of the regimes, once all data is available for analysis. The smoothed regimes are used for describing the joint spike and drop dependence.



We find that some market react similarly, while others (especially the in Nordpool area that are stabilized by a large percentage of hydro power) are almost independent of the rest of Europe. The frequency of spike and drops, and their dependence with other markets are positively correlated with the amount of wind power in the electricity system. It can be argued that the dependence results in this paper gives an indication of what future dependence (for more integrated markets) will be.



The results in this paper are of interest for financial risk management and for design of spinning reserves as more markets are being integrated and more wind power is used. The integration of markets should reduce the size of the spinning reserve needed while the increasing amount of wind (and solar) power will increase the size needed, partly because of the volatile nature of wind power generation but also because of the spatial dependence of the wind, effecting several markets simultaneously.