Fitting Alternative Autoregressive and Moving Average Models to Nigeria Crude Oil Prices (Published)
The need to compare the efficiency between the Autoregressive Integrated moving average (ARIMA (p,d,q)) models when modelling Crude oil price is the motivation behind this research. The research focuses on different orders of Autoregressive Integrated moving average models. The trend Analysis of the original series were plotted and was observed that crude oil prices were not stationary. The data were transformed by taking a natural log and the series becomes stationary after first differenced. The ACF and PACF of the stationary time series were also plotted which were the basis for the suggested ARIMA models. Error variances for the suggested ARIMA (p,d,q) models were derived and estimated as the basis for model performance comparison. Empirically, Crude Oil Price data spanning from January 2006 to July 2023 were used for the analysis. Findings from the study has revealed that, ARIMA (2,1,1) with the least error variance outperformed the other suggested models. The study further stated the estimated models for forecast of the future value of the crude oil price. The study recommends the use of error variance as a criterion for best model suggestion and ARIMA (2,1,1) was selected as the best model for modelling Nigeria Crude oil price.
Keywords: Q)., autoregressive integrated moving average (ARIMA) model, d, error variance, error variance of ARIMA (p, stationary time series