Forecasting the price of two species of fishery products in southern of Iran with emphasis on new econometric methods

As one of the sub-sectors of agriculture and natural resources, Iranian fisheries has a significant contribution in the economy of the country. Recently, planning and investment in the fisheries sector have been difficult duo to the sharp fluctuations in price of products. This paper aims at determine and choosing the most appropriate fishery product price prediction model using autoregressive integrated moving average, time-delayed artificial neural network and combined pattern of the two above-mentioned methods. The data used in this research is related to the wholesale price of two products of the fishery (Seer fish and Black Pomfret fish), from April 2001 to September 2018. It was found that the ARIMA model showed a weak performance in predicting the price of both products in comparison with the artificial neural network method. And also, the hybrid method was more effective in forecasting the price of products than the other two methods. In conclusion, it is necessary to use nonlinear methods to forecast the prices of fishery products. Also, hybrid model can be used in long-term planning due to improved performance prediction with increasing forecast horizon.