Modelling Electricity Prices Based on Meteorological Variables and Fossil Fuel Prices: A Time Series Approach

Sammanfattning

This thesis aims to describe how different weather parameters and commodity prices affect real electricity prices in Sweden's electricity price area SE3. The motivation for this analysis is that the Swedish electricity market has been subject to significant price fluctuations during the recent years, which is explained partly by variations in weather conditions and commodity prices. This project has been carried out using time series analysis, specifically an ARMAX-model, which models both the correlation between exogenous variables and the electricity price, as well as the autocorrelation between the electricity price in different time steps. The exogenous variables used are wind speed, temperature, solar irradiance, precipitation, and coal, oil and gas price. The data set used consists of daily data from each day during 2023 and 2024. The resulting model shows that wind speed, temperature and coal price have significant effects on the electricity price, with coal acting as a buffer at low levels of renewable energy production. The best model after variable transformation consisted of these three explanatory variables, and has an adjusted $R^2$ value of $0.6954$. These results have contributed to a better understanding of the structure of the Swedish electricity market, and result in a descriptive model. This model should be used as a basis for further work on the subject, in order to construct a more general and robust predictive model.