The Coal Price Forecasting Model Based On The Co-integration Theory And ANN

Coal, as an important energy, is the lifeline of China s economic development and social progress. Coal will remain China s major energy source for a long time to come. Coal price is an important indicator of the development of the coal industry. Predicating the coal price in a scientific way can help to grasp the development trend and volatility of the coal industry, so as to provide the basis for coal enterprises and the government to formulate development strategies and control policies.Given that factors influencing coal prices are complex and comparably independent, this paper analyzed those factors and their forming mechanisms. The existing coal price predicating methods often lack validity and precision in long-term predications. In view of this problem, this paper dug the effective factors influencing the coal price, and revealed the long-standing nonlinear equilibrium relationship between different variables by using the co-integration theory. Besides, this paper also established a coal price predication model based on the co-integration model and the artificial neural network by making use of the inherent advantage of the artificial neural network in nonlinear function approximation. By conducting bi-variate and multivariate co-integration analysis, neural network training and ADF unit root test on coal prices and economic variables, the author found the results were consistent with the requirements of stability test. The nonlinear function resulted from the training reflected the co-integration relationship between the coal price and relevant economic factors. Using this predication model, the author predicated China s coal prices over the next five years, and analyzed the action mechanism of different factors on the coal price. In the end, some suggestions are proposed to enable coal enterprises and governments to be more effective in adjusting strategies and making policies.