Forecasting spot price volatility using the short-term forward curve

Abstract

We use high frequency real time spot prices and day-ahead forward prices from the Pennsylvania–New Jersey–Maryland wholesale electricity market to calculate, describe, and forecast spot price volatility. We introduce the concept of forward realized volatility calculated from day-ahead forward prices. Forward realized volatility improves forecasts of spot price volatility – in the sense of higher R2s and significantly lower forecast errors – when compared with forecasts based solely upon historical volatility. The largest forecast improvements obtained when the change in forward realized volatility is large in magnitude. Splitting total volatility into its continuous and jump components is crucial for forecasting volatility at weekly and monthly horizons.