Forecasting factors for increasing the level of international competitiveness of electric power enterprises in conditions of uncertainty

Abstract

The purpose of the study is to generalize scientific and methodological principles and forecast the factors of increasing the level of international competitiveness of Ukrainian electric power enterprises in conditions of uncertainty. To achieve the set goal, the following methods were used during the study: review of scientific literature, correlation-regression analysis, analysis and forecasting of time series. Economic and mathematical modelling and systematic analysis of the factors of increasing the level of international competitiveness of Ukrainian electric power enterprises in conditions of uncertainty were also carried out. According to the results of the study, it was found that the main such factors are the level of spot prices on the DAM (day-ahead market), the volume of electricity demand (declared purchase volume), the volume of electricity sales (as an indicator of trends in its consumption), the volume of electricity production in neighbouring countries (for example, Poland). The study showed that the time series of hourly values of the first three factors and the quarter-hourly values of the last factor, when analysed on an annual scale, demonstrate a high degree of volatility and variation, which makes it impossible to use regression models for their forecasting. Despite this, the selected factors demonstrate pronounced seasonality. The results of the study showed that the selected factors have a low value of relative entropy, and the latter was calculated as the negative logarithm of the product of the autocorrelation coefficients, which for the first and seasonal lags demonstrated a high value. When calculating the autocorrelation coefficients, it was found that the predicted factors have the first degree of uncertainty – sufficiently predicted future. That is why the SARIMA(p,d,q)(P,D,Q)s model was used for forecasting under uncertainty, which takes into account seasonality and autocorrelation. As a result of building forecasts using this model, the following MAPE values were obtained for each factor: 12.3%, 5.8%, 10.1% and 2.5%, respectively. The main theoretical and methodological contribution of the study is the justification of the feasibility of using the SARIMA model as optimal for forecasting factors for increasing the level of international competitiveness of electric power enterprises under conditions of uncertainty, given their daily seasonality and the presence of a generally close autocorrelation with the first three lag variables. The scientific novelty of the study is that for the first time a formula has been proposed for calculating the relative entropy of a time series as the negative binary logarithm of the product of the autocorrelation coefficients with lag variables in order to use it as an indicator for choosing optimal methods, approaches and models for forecasting under conditions of uncertainty.