The Betfair Prof / Leighton Vaughan Williams / 16 October 2009 / Leave a comment

It is an incontrovertible fact of life that well over 90% of all US oranges used in frozen concentrated orange juice are grown in central Florida. The reason is that oranges grown in central Florida produce better concentrated orange juice than oranges grown in California, the other major source of oranges in the US.

Clearly this makes the weather in central Florida of critical importance to the market price of orange juice. One would expect, therefore, that the weather forecast for central Florida should be a key factor influencing the price at which orange juice futures trade. That's one thing. But what now if we phrase the issue in reverse? Can we use these orange price futures to actually forecast the weather? And if so, how well do forecasts so obtained compare with the forecasts issued by the official National Weather Service?

A study carried out by Professor Richard Roll of the Graduate School of Management at the University of California sought to find out. In the resulting research paper, called "Orange Juice and Futures", published in the American Economic Review in December 1984, Roll found what he termed a "statistically significant relation between OJ returns and subsequent errors in temperature forecasts issued by the National Weather Service for the central Florida region."

In particular, if the closing orange juice futures price is higher than its opening price, we would obtain a more accurate prediction of the weather by adjusting downward the National Weather Service's weather forecast. This is because a higher futures price implies a smaller crop and colder weather. Correspondingly, if the orange juice futures price is lower than its opening price, we should adjust upwards.

A front-page article in Florida's St. Petersburg Times, on Monday, October 17, 1983, puts this in some perspective. "The National Weather Service spends $280 million each year predicting the nation's weather," it runs, "and it uses some of the niftiest gadgets ... to do it. But it may be overlooking the price of orange juice."

In an interview with the paper, Roll noted how in 1981, when the temperature in central Florida fell into the 20s (Fahrenheit) for four consecutive nights, the price of orange juice futures rose 40%. "They [the traders] have their own meteorologists", he explained. "They have a bigger incentive than the Weather Bureau itself to predict it correctly." In particular, if traders can anticipate falling temperatures they can buy before the price goes up and sell after it does.

An important point to note here is that nobody is saying that the official weather forecasting service is of no value. Indeed, the market might have been a lot less accurate if these forecasts didn't exist. What Roll's study does suggest, however, is that the commodity markets add some information above and beyond this, much as opinion polls can help forecast election outcomes, without representing the totality of information available to those trading the election betting markets.

Whoever would have thought the economics of orange juice could be so interesting!