A Comparison of the Predictive Ability of VAR, ARIMA and Artificial Neural Network (ANN) Models: OPEC’s Oil Demand

Abstract

Awareness of the future oil demand is essential for OPEC member countries to determine priorities and policy selection for achieving economic growth and development. In this study, demand for OPEC’s oil, using time-series models Including Vector Autoregressive (VAR), and Autoregressive Integrated Moving Average (ARIMA) models and an alternative model, artificial neural network (ANN) (using monthly data from 2001:1-2010:10), is predicted. To measure the ability of predictive power of the models, three criteria are used: Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The results show that VAR pattern with the error rate of 6% for the sum of squared error, mean absolute error of 19% and 5% of the average of the absolute value is the most appropriate forecast for OPEC’s oil demand. Based on VAR model, it is predicted that demand for oil is growing over all the months in the year 2012. Also, the projected demand in 2015 shows that the demand for OPEC’s oil has a rising trend but in 2014 this trend will be slower.