Designing a Oil Market Model and Comparing Crude Oil Price Forecasts

The price of crude oil is one of the factors affecting economic indicators. Therefore, the prediction of oil prices and the accuracy of the applied methods have always been discussed by economists. In this study, the effect of all effective variables on the supply and demand of crude oil based on McAvoy's competitive theory is investigated, and the supply and demand are estimated using the system of simultaneous equations and conventional statistical methods. Then, using algebraic operations and the assumption of equality of oil supply and demand in the long term, the long-term potential of oil supply and demand is extracted with respect to each of the variables in the model. Based on the results, the world's gross domestic product (GDP) has the greatest impact on oil prices with a demand potential of 0.6039, and the world's military and security tensions have the least impact with a demand potential of –0.0110. After estimating the model, the prediction accuracy of three combined mothod is compared with conventional and single-variable methods of neural network and ARIMA. These three combined methods are: (a) neural network and system of simultaneous equations, (b) ARIMA and system of simultaneous equations, (c) neural network and ARIMA and system of simultaneous equations. The results showed that the combined method of ARIMA and simultaneous equation system provides better reslts for 5-year forecasts while the combined method of neural network and ARIMA and simultaneous equation system shows better results for 10-year forecasts.