Determinants of electricity demand in Sri Lanka, univariate and multivariate time series approach

Abstract

Studying energy consumption problems has become an important topic of research in recent decades. Efficient energy distribution planning necessitates accurate forecasts of future demand in order to achieve a balance between energy supply and demand. This study was conducted by focusing on developing a demand model for forecasting electricity demand in Sri Lanka. A univariate and multivariate model were focused on forecasting the electricity demand. Economic variables, including gross domestic product, foreign direct investment, inflation rate, population, average unit price, and number of consumer accounts, were used for the multivariate analysis. Data relevant to economic variables and electricity demand was collected from 1969 to 2020. Under the univariate analysis, naive, drift, and mean models were used as benchmark models, and exponential smoothing (ETS) and the ARIMA model were used further for the analysis. Under the multivariate analysis, the Granger causality test was used to identify the nature of the relationship between each economic variable and the electricity demand in Sri Lanka. A VAR modeling approach was used to build up a relationship model between the electricity demand and the selected economic variables under the Granger causality test. Forecast error was used to select the best model. According to the univariate analysis, as per the forecast error calculations, ARIMA(0,2,1) was found to be the best univariate model. The Granger Casualty test under the multivariate analysis indicated that foreign direct investment and average unit price Granger cause the electricity consumption in Sri Lanka. Based on an overall analysis of forecast errors, the VAR model demonstrates superior statistical fitness compared to both univariate and other multivariate analysis for predicting electricity demand in Sri Lanka.