Back to search

Institutional Repository of Vilnius University

Baltijos šalių elektros kainų prognozavimas /

Abstract

dc:description

This thesis analyzes electricity prices in the Baltic countries and evaluates their forecasting possibilities using time series analysis methods. The aim of the research is to assess the suitability of univariate and multivariate models for electricity price forecasting and to determine which approach provides more accurate results. The study is based on electricity price data from Lithuania, Latvia, and Estonia for the period 2020 - 2024. First, descriptive statistical analysis was performed to examine the main characteristics of the data. Stationarity tests indicated that all price series are non-stationary in levels but become stationary after first differencing. Cointegration analysis revealed long-run relationships among the Baltic electricity prices, confirming that these markets are closely interconnected. For forecasting purposes, ARIMA and VAR models were applied. ARIMA models were estimated separately for each country, while the VAR model captured the interdependencies between the price series of different zones. Forecasts were generated using the rolling window method, which allows evaluating model performance under real-time conditions. The empirical results show that the VAR model outperforms the ARIMA models in terms of forecasting accuracy. This is confirmed by lower values of mean absolute error (MAE) and root mean squared error (RMSE) across all analyzed zones. Although both models exhibit forecasting errors, the multivariate approach better reflects electricity price dynamics by incorporating cross-border relationships. The study also demonstrates that electricity price forecasting remains a challenging task due to high volatility, extreme price fluctuations, and the influence of external factors. As a result, even advanced models cannot fully eliminate forecasting errors. The results not only confirm findings discussed in the literature but also extend them. In this thesis, electricity price forecasting in the Baltic countries is analyzed in a comprehensive way. Both univariate and multivariate time series methods are combined. In addition, a rolling window forecasting scheme is applied. Unlike many previous studies, the models are evaluated not only based on theoretical criteria but also using forecasting accuracy measures obtained under real-world conditions. Empirical findings further show that incorporating cross-border relationships (through the VAR model) significantly improves forecasting accuracy compared to separate ARIMA models. It is also found that the electricity price time series exhibits a complex structure. This structure combines a trend, multiple seasonal patterns, and a strong residual component. These results support the need for applying more advanced modeling approaches. In conclusion, the findings suggest that multivariate time series models are more suitable for forecasting electricity prices in the Baltic region. Future research could focus on incorporating additional explanatory variables or applying more advanced methods to further improve forecasting accuracy.

Degree

thesis:*
Grantor dc:publisher
Institutional Repository of Vilnius University
Year dc:date
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Žūkas, Motiejus,

Rights

dc:rights
Statement dc:rights
  • info:eu-repo/semantics/openAccess
Language dc:language
lit

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
Vilnius University
Base URL
epublications.vu.lt/oai
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Žūkas, Motiejus,. Baltijos šalių elektros kainų prognozavimas /. Institutional Repository of Vilnius University, 2026. https://repository.vu.lt/VU:ELABAETD308327345&prefLang=en_US