COVID-19 and the Information Content of Electricity Futures

This paper examines the change in expected future electric power prices as market participants became aware of COVID-19. Using an extension of the framework developed by Fama and French (1987), we recover future spot price expectations based on information embedded in prices of futures contracts traded before and after the emergence of COVID-19. Our estimates show that market participants reduced their estimates of future electric power prices upon the emergence of COVID-19. The expected prices for spring 2020 embedded in futures contracts traded in March 2020 were close to the realized spot prices in the early part of the pandemic. Still, prices expected later in 2020 were lower than eventually realized. A notable feature of our methodology is a decomposition of price prediction errors into two components: one due to ex-post weather shocks and another due to other factors, including COVID-19. We find that weather anomalies explain a large portion of the variability in prediction errors, particularly for the winter months during which weather shocks account for as much as 30% of the error in price predictions. We use the elastic net machine learning/shrinkage estimation method to choose covariates for our price predictions and for our error decomposition