Michael Lynch: Are There Oil Price Superforecasters?

At an academic seminar once, an economist remarked that oil prices were stochastic in nature. The student next to me asked what that meant, and I explained stochastic means random, and economists use it to look smarter. He then asked me if oil prices moved randomly and I explained that when an economist says something is random, it means he can’t explain or predict it.

The recent work by Phillip Tetlock and Dan Gardner, like others before them, has found that people who are relative novices tend to do as well or better as experts at a variety of things, from psychological diagnoses to economic forecasting. Of course, in the stock market, those relative novices are known as small investors and they are thought to always be buying high and selling low. On the other hand, there are many in industry who have done the same thing: snapping up shale leases at the height of first US gas prices and then oil prices, for example, which has left a number of small oil companies in dire financial straits.

And there are many instances where dart throwers have bet expert investors and/or the market, or octopi have predicted World Cup winners, and of course the famous one hundred chimpanzees in a room with typewriters turning out Shakespeare in 37 years. (One of my columns: two chimps, one week, tops.) But most of these examples represent first and foremost cherry-picking of the data (if the octopus gets it wrong, you don’t hear about it) or just randomness, as when a particular throw of the darts ‘gets lucky.’

BUT: and it is a big but (no snide comments, please), there are other explanations for why experts could underperform, especially in economic forecasting. Sometimes experts are more sizzle than steak. An oil company CEO is a manager, not a forecaster or economist (generally), and often has only a superficial knowledge of industry economics. Plus, all too many who publish books and articles are relative novices.

Peak oil is a great example of this, where numerous books and even some academic articles have been written by bankers, physicists, oceanographers, computer scientists and so forth. Many of them have made very basic, often critical, mistakes, usually repeating something from another writer whom they presumed to be expert. David Goodstein, a Caltech plasma physicist, argued that once production of a nonrenewable resource declines, it cannot recover, which is not only nonsense but easily refuted. (Go to BP.com and see that oil, gas and coal production have all fluctuated, both globally and by country.)

In 1994, I published an article which looked at long-term oil market forecasting, and found that between the Iranian Oil Crisis of 1979 and the oil price collapse of 1986, the vast majority of oil price forecasts predicted ever higher prices. Even as prices fell in the early 1980s, the expert community insisted that they would recover, and soon, right up to the point where prices collapsed in 1986. Sound familiar? Think of how the drop from $100 a barrel last year saw numerous articles contending that recovery was inevitable.

An interesting aspect of this record is that it refutes the rational expectations theory to which many economists subscribe. The contention is that while any given forecast might be wrong, on average forecasts will prove to be correct. Oil price forecasts proved to be laughably wrong: I only found one that was actually too low, and I suspect it was backdated from after the 1986 oil price collapse.

The reason behind this mistake was not bad luck but bad theory. The experts believed that resource depletion ensured ever-rising prices, even though this was demonstrably false. The more mathematically-inclined argued that it could be proven that nonrenewable resources had to rise exponentially, the so-called Hotelling Principle. Not only did this contradict historical behavior, the math was actually wrong but when leading resource economists, such as my mentor, M. A. Adelman, explained this, they were largely ignored.

So, in theory, non-experts could beat the experts at oil price forecasting if they produce random forecasts while the experts’ forecasts are based on bad theories that leave them inclined not to be unlucky, but down right wrong. In the short-term, oil prices move unpredictably for a variety of reasons, but longer term, there is much less uncertainty—except in the minds of most experts.

Forbes