21 May 2009

sometimes it pays to be wrong

In a recent newsletter, the financial analyst Bob Prechter, makes a great observation about the perils of always relying on statistical probabilities. A given event, or occurrence, may be rare, but that that is no reason not to plan or prepare for it. That 200 year flood or hurricane may be rare, but when they do occur the consequences are disastrous. When the consequences of a given event are huge, it is far better to plan for the worst, even though the chances of dire circumstances occurring anytime soon are slim, rather than do nothing until the inevitable actually does occur which destroys you.

In Bob’s example, he talks about a restaurant which terrorists have attacked several times over the years. Sure, the restaurant patrons might only be killed one out every 500 days, but on that day when the terrorists come, it’s game over. It might be worth while avoiding the restaurant for 50 days of every year that have particularly higher odds of attack (due to special commemorative dates, etc), even though most of those days will wind up seeing no violence at all.

Prechter goes on to explain that economists often fall into this trap of playing the odds. Economic contractions are rare, so it is usually a safer call to predict expansion and growth. However, when contractions do occur they are often severe, and can be utterly devastating to businesses (and investors). What use is an economist who is right 99% of the time (predicting blue skies and growth), if they can’t be relied on to give insights as to when a disastrous economic upheaval may occur? It would be preferable to have an economist err in crying wolf 20% of the time if they manage to advise people correctly about avoiding the next wipe out.

No comments:

Post a Comment