Forecasting electricity prices: The impact of fundamentals and time-varying coefficients

This paper investigates the day-ahead forecasting performance of fundamental price models for electricity spot prices, intended to capture: (i) the impacts of economic, technical, strategic and risk factors on intra-day prices; and (ii) the dynamics of these effects over time. A time-varying parameter (TVP) regression model allows for a continuously adaptive price structure, due to agent learning, regulatory and market structure changes. A regime-switching regression model allows for discontinuities in pricing due to temporal irregularities and scarcity effects. The models that invoke market fundamentals and time-varying coefficients exhibit the best predictive performance among various alternatives, in the British market.