Overconfidence
The tendency of experts to give ranges that are too narrow, which leads to underestimating the tail of a risk as soon as judgment replaces data.
Overconfidence is the gap between the certainty an expert displays and the accuracy of the expert's judgments. Its best-measured form concerns ranges: asked for a range containing the true value with ninety or ninety-eight percent probability, the expert almost always gives one that is too narrow, and the true value falls outside it far more often than announced. Marc Alpert and Howard Raiffa showed this in work published in 1982, and the effect has been found among physicians, engineers and financiers. In insurance, it strikes where data is lacking and expert judgment sets the parameters: the frequency of an emerging peril, the severity of a cyber scenario, the correlation between two lines. A range too narrow on a parameter becomes a tail too thin in the model, hence capital and price that are too low. The remedy lies in scoring past judgments, which makes overconfidence visible.
A Duke University survey asked chief financial officers of large US companies every quarter, from 2001 to 2011, for an 80 percent range for the coming year's return on the S&P 500 index. According to the study published in 2013 in the Quarterly Journal of Economics, the realized return fell within their range only 36 percent of the time.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.