The Efficient Combinatorial Forecasters for Supply and Demand of Gasoline in Iran: Combination of Multivariate State Space Model and Artificial Neural Networks

The transportation sector is one of the most basic infrastructures in the country to achieve economic growth and development، and petroleum products are most important production factors in the sector. Due to importance of the transportation sector''s demand of petroleum products such as gasoline، government intervention in the production and distribution، growing consumption and imports of gasoline، in any countries energy planning is inevitable. Therefore، it is essential that policymakers understand the process of production، consumption and importing these products. In this regard، this paper utilizes two approach، multivariate state-space and artificial neural networks models and using data of production، consumption and gasoline imports in the sample (1978-2008)، to estimation of quantities of production، consumption and gasoline imports. Then، the article predicted production، consumption and imports of gasoline for short-term (2009-2011). The results of the paper suggest that artificial neural network model in forecasting in sample; have better performance than the Multivariate State Space model. Finally، the paper presented forecasting of production، consumption and import of gasoline; with combination of forecasts approach، that this forecast is efficient than two above mentioned methods.