How predictable are prices of agricultural commodities? The possibilities and constraints of forecasting wheat prices

Wheat price forecasts are very important for traders, farmers and politicians as well. However, only quite accurate price predictions can guide these groups towards making the best decisions. Therefore the well-known wheat price projections of both the OECD and the FAPRI from 1996 on are tested for their predictive accuracy using Theil's inequality coefficient. Despite the fact that both models could not foresee the price peak which occurred in February 2008, their predictions offer more accurate values than a naive prediction of no price change. Nevertheless, precise price forecasts cannot be expected by the models of the OECD and the FAPRI since some short-run effects such as inappropriate weather are not predictable. Thus, our own econometric model is developed taking the previous price development, the stocks-to-use-ratio and the crude oil price into account. In comparison to the projections of both institutions the model, with rather simple assumptions, was able to generate forecasts more accurately. In a simulation study which takes different crude oil price levels and stochastic effects of the world wheat consumption and the average yields per hectare into account, the possible wheat price range is shown as large. Therefore, price predictions can only inform about general longrun trends.