Assessing the Stability of Regional Energy Supply Systems: Econometric Models and Forecasts

Abstract

This study investigates the econometric modeling and forecasting of energy consumption dynamics within the energy supply system. Per capita primary energy consumption was selected as the indicator of energy consumption in the energy supply system, and the analysis was conducted using annual data covering the period from 1990 to 2025. First, the descriptive statistical characteristics of the indicator were assessed, revealing an overall declining trend in energy consumption over the study period. The dynamics of energy consumption were modeled using the ARIMA(1,1,3) model, and the forecasting accuracy of the model was evaluated using the MAPE, MAE, and RMSE metrics. The estimated ARIMA(1,1,3) specification yields R² = 0.93, while the forecast evaluation reports MAPE = 3.14%, MAE = 597.96 kWh/person, and RMSE = 749.64 kWh/person. The baseline forecast indicates a decline in per capita primary energy consumption from 15,355.95 kWh/person in 2026 to 13,760.47 kWh/person in 2030, equivalent to approximately 86.3% of the 2025 level. The results demonstrate the usefulness of a transparent time-series benchmark for short- and medium-term energy planning while also highlighting the need to interpret forecasts conditionally on historical dynamics.