Realized volatility and the influence of market measures on predictability: Analysis of Nord Pool forward electricity data

This is the first paper to utilize intra-daily high-frequency data and to apply known market measures for the prediction of volatility in the Nord Pool electricity forward market. The work is based on recent methods of separating realized volatility into two components: continuous and jump volatilities. In addition, the link between future price volatility and current observable economic variables is examined. The measures-trading volume, time-to-maturity, asymmetric effect from negative shocks, and intra-week seasonality-are assessed to identify improvements in day-ahead predictions. The model where the total variation is separated into its continuous and jump components is compared with the simpler heterogeneous autoregressive model of realized variation both in- and out-of-sample. The results show a strong degree of persistence in realized volatility, and significant impacts from the mentioned market measures when predicting Nord Pool forward price volatility. Hence, there is a clear preference for models accounting for the systematic impact of market measures to improve volatility assessment for tomorrow. Moreover, separating the total variation into continuous and jump components seems potentially useful when predicting day-ahead volatility.