전력 수요 시계열 분석과 장기예측

Abstract

This study considers the time series characteristics in residential electricity demand function and forecasts long-run electricity demand in Korea. It is widely known that electricity demand and its determinants are unit root processes, so that estimating and forecasting the electricity demand function based on cointegration relationship are common in empirical analyses. We apply unit root tests on monthly time series data for the period 2006:01-2022:12. Korea has experienced the rapid development and social changes over this period, which provides convincing evidence of the possibility of structural changes. The unit root test of Dickey, Fuller (1979) tends to be biased in favor of non-rejection of the unit root null hypothesis if the process considered is trend-stationary with a slope change. What we found is that electricity demand and its determinants are trend-stationary processes allowing for a slope change, whereby there is no need to invoke either cointegration regression model or error-correction model under the unit root assumption. We apply a regression model with a structural change to estimate the residential electricity demand function and forecast the long-run electricity demand for the period 2023-2050.