Friday 03 Apr 2020 | 00:06 | SYDNEY
Friday 03 Apr 2020 | 00:06 | SYDNEY

Failure to forecast


Mark Thirlwell

26 November 2008 16:33

Yesterday’s Financial Times has a long piece pointing out that this has been a particularly bad year for economic forecasters. While I enjoyed the story, the only news here is that forecasting failures are still news. As Tim Harford noted in the same newspaper back in August, people often chuckle about the (lack of) forecasting skills of economists, but once the sniggers are over, they are back asking for more forecasts.

So why do people continue to pay so much attention to economic forecasts, despite repeated evidence that the forecasters will get their numbers wrong? Doesn’t a record of continuous forecasting failure mean that the whole process has no real value? I think one reasonable defence of this particular black art (aside from the pecuniary one that goes, ‘they keep paying me to do it and who am I to turn down good money?’) is that the real value of forecasting exercises is to be found in the overall analysis and the discussion of risks that accompany the projections, rather than the projections themselves. 

If so, why not skip the numerical forecasts altogether, and stick to some broad, descriptive scenarios? Well, at a practical level, its actually pretty difficult to talk about what you think is going on in an economy without throwing at least a few numbers around (if you say that you think growth is going to slow by a lot, for example, the immediate question that comes back is, 'how much is a lot?').

Also, doing the numbers should serve as a consistency check to make sure that those interesting stories you are telling about the world actually add up.

Finally, I should emphasise that failure to forecast is not just a shortcoming of economists. As Sam commented back in December last year, it turns out that most experts just aren’t that good at prediction.