MODELING AND FORECASTING ELECTRICITY CONSUMPTION IN NIGERIA USING ARIMA AND ARIMAX TIME SERIES MODELS

Abstract

This study compared the extrapolation strengths of two models:  Autoregressive Integrated Moving Average (ARIMA) and Autoregressive Integrated Moving Average with an Exogenous Variable (ARIMAX) in the forecast of Nigeria's electricity consumption. Annual data on power generation and consumption from the Central Bank of Nigeria statistical bulletin for 2006 and 2016 over a 51-year period (1970-2020) was used. Industrial and residential electricity consumptions were examined for possible unit roots (non-stationarity) using the Augmented Dickey-Fuller test approach. The ADF test result showed that the time series achieved a stationary state for the variables under consideration at first difference.  Akaike Information Criterion (AIC) and Root Mean Square Error (RMSE) were used to assess the performance of each models. Comparing the ARIMA and ARIMAX forecast models, ARIMA(0, 1, 1) emerged for modelling and forecasting industrial electricity consumption in Nigeria while ARIMAX (1, 1, 1) with installation capacity as exogenous variable was suitable for modelling and forecasting residential electricity consumption in Nigeria. The study recommended that for optimal residential electricity consumption in Nigeria, installation capacity and the total power generation in Nigeria should be enhanced.