The Overconfidence Problem in Forecasting

It may be neither troubling nor surprising that C.F.O.'s can't accurately predict the stock market's path. If they could, they'd be running hedge funds and making billions. What is troubling, though, is that as a group, many of these executives apparently don't realize that they lack forecasting ability.

Some economists have questioned whether experimental findings are relevant in competitive markets. They suggest that students, who often serve as guinea pigs in such tests, are overconfident, but that the top managers in large companies are well calibrated. A recent paper, however, reveals that this hopeful view is itself overconfident.

In that paper, three financial economists — Itzhak Ben-David of Ohio State University and John R. Graham and Campbell R. Harvey of Duke — found that chief financial officers of major American corporations are not very good at forecasting the future. The authors' investigation used a quarterly survey of C.F.O.'s that Duke has been running since 2001.

Among other things, the C.F.O.'s were asked about their expectations for the return of the Standard & Poor's 500-stock index for the next year — both their best guess and their 80 percent confidence limit. This means that in the example above, there would be a 10 percent chance that the return would be higher than the upper bound, and a 10 percent chance that it would be less than the lower one.

Read more…