"Prediction is very difficult, especially about the future" -- attributed to Niels Bohr (physicist, 1885-1962)
Every buyer faces the question: "What's going to happen to the price?" The only correct answer is: "It will go up, or down, or, exceptionally, stay the same." If I were able reliably to forecast the price of rapeseed oil (or cocoa, steel or tomatoes), then I would not be writing this article -- I would be sunning myself in the Maldives.
To understand why accurate forecasting is very hard to do, let us consider something as apparently straightforward as forecasting the result of a Super Bowl XLVI. In theory, this should be easy: one might reasonably expect that, based on past history, New York should beat New England, but there are many reasons why this might not happen. The final score is the result of a series of extraordinarily complex interactions between 22 players, the officials, the crowd, the head coach, the weather, and so on. If the kicker fails to convert an extra point, if the referee misses a blatant foul, or if another player plays below par due to an injury, then an unexpected result may occur. If forecasting were easy, then 'sports betting' would not have held so many people in so much thrall for more than 100 years.
A football match lasts for just 60 minutes and exists within limited physical boundaries. Imagine, then, how much more difficult it is to forecast, say, the price of wheat in six or twelve months' time. There are many more factors to consider, and each factor is itself subject to unpredictable influences. The availability of many raw materials depends on the weather. The climate is an eco-system which is inherently chaotic: the proverbial flap of a butterfly's wings causing a hurricane on the other side of the world. Even leaving aside major climatic events such as hurricanes or El Niño, the simple fact that it might or might not rain tomorrow in a particular field will affect the crop in imperceptible ways. Even though the effects of rain, drought, wind, frost, etc. on a crop might be well understood in broad terms, it is not possible to calculate the impact on millions of individual plants separately.
Any mathematical model of real-world behavior is necessarily a simplification. Since the real world situation at any given moment is unique and unrepeatable, it is impossible to reflect all of the myriad influences on price to the required degree of accuracy. Even for non-food commodities, these factors -- supply and demand, sentiment, substitutability, to name a few -- are subject to further influences such as economics, fashion and politics, that are themselves irrational and unpredictable.
Add into the mix the various steps that occur in the supply chain: growing, harvesting, processing, packing, and freight. Each impacts the selling price, and it is easy to see that there can be no guarantee that a price forecasting model will be accurate.
Consider the following chart showing the price of an (unnamed) commodity from Jan 2007 to Jun 2008, with the annual trend-lines helpfully added. Before reading on, would you care to predict what happened in the second half of 2008?
Given the acceleration of the upward trend in the first half of 2008, you might reasonably suppose that the price would continue to rise; or you might think that the price has got ahead of itself and would drop back. Let's move on a further 6 months.
Oh dear! If in the middle of 2008 you had bet on the trend continuing, you would have lost a lot of money.
The conclusion is that you just can't predict football results, the lottery or commodity prices.
Nevertheless, if you understand how a commodity price fluctuation could affect your business, then you can look for ways to reduce exposure, whether by fixing a contract price, back-to-back contracts with a customer or by index-linking some of your markets. The key rule is that the safest price can in fact be better than the cheapest price. (And in case you hadn't guessed, the charts show the price of Crude Oil).