Analysis of the Fundamental Predictability of Prices in the British Balancing Market

Abstract

This research analyses the non-linear and complex effects of drivers of system imbalance prices in the GB electricity market. Unlike day-ahead prices, the balancing settlement prices are comparatively under-researched, yet their importance is growing with greater market risks. The fundamental drivers of these prices are analysed over 2016-2019. The result of a nonlinear modelling approach reveals that system imbalance price exhibits a regime-switching behaviour, driven by weather and demand forecast errors, as well as other system effects. Surprisingly, balancing prices are predictable out of sample and a regime switching specification is more accurate than a linear model for prediction.

Bayesian Predictive Distributions for Imbalance Prices With Time-Varying Factor Impacts

“…This process repeats under each SP. This episodic nature of the SP process led [21] and others to consider the predictability of prices in the balancing market. Our focus is to consider how the marginal effects of the various exogenous factors that influence prices in each SP may evolve over time.…”

“…The data-previously analyzed by [21]-were obtained from the Balancing Mechanism Reporting Service (BMRS) provided by ELEXON 1 . The sampling period is from 1st July 2016 to 30th June 2019, providing 3 years of historical data for developing and testing the statistical model.…”

“…For the entire sample period, Panel A and B provide the time series and histogram plots of the price data. As already noted by [21] the series has some spikes and occasional negative prices and was therefore trimmed slightly in that analysis. There are pros and cons for such trimming, but in order to directly compare results, we followed the same data pre-processing before fitting the model: all observations larger than 140 were replaced with 140 and all negative prices were replaced with zeros.…”

“…Nevertheless, the market participants (generators, retailers and speculators), who may be exposed to imbalance prices on their own trading positions, can benefit substantially from improved predictions. Thus, in the British imbalance prices, [21] recognized that there may be different price formation processes depending upon whether the system operator is seeking to increase or reduce generation in the system, and as a consequence, they found it beneficial to use a Markov-switching model to predict imbalance prices. In contrast, also on British balancing prices, [22] used machine learning techniques to explore the relevant of various real time variables with promising implications for forecasting.…”

“…Our methodological approach to density forecasting with time varying parameters has its origins in the Bayesian dynamic linear model (DLM) of [24] and [25]. We adapt this to model the British electricity imbalance prices in the dataset used by [21], as previously published in these Transactions. This provides a basis for direct comparison with those previously-published regime switching results.…”