Forecasting monthly cotton price
This paper examines the predictive performance of structural and vector autorregressive models for forecasting monthly cotton prices. Two distinct time periods were selected for testing: a period of major policy shock, and a period of more normal market conditions. The study also investigates a composite approach, using vector autoregressions to determine the future values of exogenous variables of the structural model. Multi-dimensional testing procedures were adopted to evaluate the accuracy of forecasts. Simulation results demonstrate the superior performance of the structural model in handling major policy changes, while the time series approach shows greater accuracy in forecasting normal price movement. Although the composite approach failed to show improvement in forecasting accuracy, a joint specification of the structural model and the time series properties of exogenous variables may merit further investigations.