Forecasting day-ahead electricity prices in Sweden : Has the forecasting accuracy decreased?

Sammanfattning: Sweden is currently transitioning towards having 100% electricity generation from renewable energy sources by 2040. To reach this goal, Sweden will ramp up the generation from wind power while simultaneously phasing out nuclear power. Replacing nuclear power with an intermittent production source such as wind power has been proven to increase the variability of electricity prices. The purpose of this study has been to investigate if the increasing electricity generation through wind power in Sweden has decreased the accuracy of price forecasts provided by ARIMA models. Using an automated algorithm in R, optimal ARIMA models were chosen to forecast on-peak and off-peak hours for both winter and summer periods in 2015. These forecasts were then compared to forecasts provided by ARIMA models calibrated on data from 2020. The results from the empirical analysis showed that the overall forecast errors are twice as large for the 2020 forecasts indicating that increasing electricity generation from wind power has decreased the forecasting accuracy of price-only statistical models.