Risk management in electricity markets

Purpose-The purpose of this paper is to explore risk management models applied to electric power markets. Several Value-at-Risk (VaR) models are applied to day-ahead forward contract electric power price data to see which, if any, could be best used in practice. Design/methodology/approach-A time-varying parameter estimation procedure is used which gives all models the ability to track volatility clustering. Findings-The RiskMetrics model outperforms the GARCH model for 95 per cent VaR, whereas the GARCH model outperforms RiskMetrics for 99 per cent VaR. Both these models are better at handling volatility clustering than the Stable model. However, the Stable model was more accurate in detecting the numbers of daily returns beyond the VaR limits. The fact that the parsimonious RiskMetrics model performed well suggests that efforts to specify the model dynamics may be unnecessary in practice. Research limitations/implications-The present study provides a starting point for further research and suggests models that could be applied to electricity markets. Originality/value-Electricity markets are a challenge to risk modelers, as they typically exhibit non-Normal return distributions with time-varying volatility. Previous academic research in this area is rather scarce.