Roots of Crisis: Why Economic Forecasts Failed?

Yesterday, Feb 27, it was announced that in the last quarter the U.S. economy shrank at a 6.2 percent annual pace. According to the median estimate of 74 economists surveyed by Bloomberg, GDP was projected to contract at a 5.4 percent annual pace; forecasts ranged from 3.8% to 6%. Moreover, the 2.4% deviation from the previous forecast was almost five times as large as the average adjustment.

Couple more examples...
- Early in 2008, the Bank of Canada forecast the Canada’s economy would grow by 2.8% in 2009 and predicted $200-a-barrel oil. Impact of subprime mortgages was widely expected to be quite temporary.
- Canada lost 129,000 jobs in January 2009; way more than the consensus of economists which called for only 40,000 losses.

How did economists miss their targets by a mile?
Well, they admit that they are not that good at forecasting “turning points” like recessions. They missed this recession; moreover, the early 80s and early 90s recessions were not very well predicted either.
However, it looks like they are not very good at forecasting on quarter scales either. So, what is the problem? And why did they miss the change in consumer spending/saving ratio? There are a couple of reasonable explanations:

Herding behavior
Herding behavior is one of the reasons:

"There’s a considerable reward for being the outlier and being right, but I think
there’s a severe punishment if you’re an outlier and you’re wrong because then
you really look bad. It’s not like everybody missed it. You alone missed it."– Don Drummond, Chief Economist at TD Bank.

Focus on Growth
Such as our economic reporting is focused primarily on growth the most of attention is typically being paid to GDP and socially important indicators: inflation and unemployment. Other important variables (consumer spending, government spending, export, and import) are being monitored, but usually they are a part of more professional discussions.

Mainstream Economic Models
Nevertheless, the major problem of economic forecasting is that mainstream economic models are extremely complex. I analyzed problems associated with this complexity in an article about mainstream economic forecasting. The two major consequences of this complexity are:
- mainstream economic models do not have predictive power for turning points and they are inadequate for unusual economic situations (like current)
- with the huge number of parameters and variables it is hard to see the whole picture.

So, what is the point of a weather forecast which by definition can’t predict a hurricane, tornado or major snowstorm? Probably nobody would pay a buck for such service. Fortunately, we have weather forecast services which are able to predict major weather patterns. Also, we managed to build a tsunami forecast service, which will probably save thousands of lives in the future. There is a downside of their imperfect forecasts – sometimes they issue false positive (no issue, but forecast identified one) warnings. But, it is perfectly fine. It is much better than false negative (there is an issue, but forecast did not identify one) warnings which were constantly issued by our lead economists. False positives save lives, false negatives lead to casualties, bankruptcies and foreclosures.