Forecasting the recession


18 August 2009


Forecasters are under the pump with a recession that many didn’t see coming. As I don’t do any macroeconomic forecasting, I can sit back and smile smugly at some of my colleagues while I work on simpler problems such as forecasting in epidemiology, demography and energy demand.

Some of those colleagues are cited in the Wall Street Journal today. The following quotation is interesting:

The spate of cloudy crystal balls highlighted an uncomfortable reality about telling the future: It is hardest when it is most important.

Initially it sounds profound—just when you need to forecast, the data conspires against you and makes it difficult. But in hindsight I don’t think it is like that at all.

When it is easy to forecast (e.g., when there is a steady increasing trend and little volatility), no-one is thinking about the forecasting because it is obvious what is going to happen. And so forecasting doesn’t seem important because it doesn’t get much attention. But when there is a lot of volatility, then people look to forecasting for answers, just when it is hard to do it accurately. Consequently, it is hardest when people are thinking about it, because they only think about it when it is hard.

That said, macroeconomic forecasting has a bad name for a good reason. Far too many confident forecasts are made without discussion of the uncertainty. If only every forecaster produced prediction intervals every time they made a forecast, the users would realise that macroeconomic forecasts are little better than shooting blindfolded.

See my talk on Forecasting and the importance of being uncertain where I argue for mandatory prediction intervals for every point forecast.