Crystal Balls Have Become Hazy – Applied Forecasting

A WSJ columnist this week examined why industry forecasts have been stumbling, and found that many rely on economic forecasts, which didn’t fully anticipate the recession. As a result, recent forecasts of hotel-industry activity, gas prices, air travel and communications-industry revenue all missed the mark.

To forecasting pros, these problems reflect broader shortcomings. "There are few examples of good industry forecasts," said J. Scott Armstrong, a professor of marketing at the Wharton School who studies forecasting. "It is difficult to find firms that use evidence-based methods to forecast and, without that, I can see no way that forecasting accuracy could improve. Nor do they often use proper validation procedures. When they do use such methods, there is no incentive to tell others about it."

Added Len Tashman, editor of the journal Foresight, "It's an ongoing campaign to upgrade forecasting practice."

Forecasts struggle mightily when there are external shocks. "The external factor (like the run-up in oil prices that occurred last year) are something that most models cannot cope with well," said forecasting researcher Jim Hoover. "The same could be said about the impact of the current recession on last year's projections of this year's sales for many firms."

The Energy Information Administration took note of that weakness. To cope, it is seeking to communicate to users of its forecasts that energy prices are volatile and have gotten harder to predict. Its forecast Tuesday emphasized this volatility. "The swing in oil prices last year prompted us to make analysis and presentation of price volatility and uncertainty a priority," said Tancred Lidderdale, a senior economist at the Energy Information Administration.