I am reading (courtesy of Ezra Klein) William Easterly's wildly enthusiastic review of Nobelist economist Daniel Kahneman’s Thinking, Fast and Slow, about the workings of our unconscious and conscious thinking processes. I encounter a quick-summed conclusion that looks a little dicey to me:
Even worse, we don’t know what we don’t know. In one experiment, chief financial officers of corporations were asked to forecast the return on the Standard & Poor’s index over the following year, giving one number they were 90 per cent sure was too high and another they were 90 per cent sure was too low. The true number was outside their intervals 67 per cent of the time.Wait, I think...what year? 2008, perhaps? To assess the 90% confidence level, wouldn't you want, say, 10 years? And do I need to buy the book to check this out?
Nah... Google experiment cfos forcast S&P next year 90 percent. And lo:
Managerial Overconfidence and Corporate Policies (April 2006)Turns out all's well:
...we collect five years worth of forecasts by Chief Financial Officers (CFOs) about stock market returns together with confidence intervals. CFOs are miscalibrated, on average: realized returns are within respondents' 80% confidence range only 39% of the time.The fit's not perfect (90% v. 80%, etc.), but the research is pretty plainly by the same people, or at least a descendant/predecessor on the same line of inquiry, and in any case dispatches my skepticism. This leads me to confront my own bias, which was toward finding a flaw, as in this post. You could call it predatory reading, a reflex triggered or neutralized by all kinds of confirmation biases as I work my way down a page. Not to mention the miracle of internet search and access -- its wow factor may be getting long in the tooth, but it's still operative!
Also on display: the plusses and minuses of Internet-induced ADD. I had another project in mind this morning; first, I flip to Twitter, thence from Klein to Easterly to here. Worth it?
Now, to finish that review...