Reference class forecasting
A method that starts from what happened to comparable cases before looking at the particulars of the present case, to correct the expert's optimism.
Reference class forecasting asks one to start with the outside view: what family of comparable cases does this one belong to, and what happened to that family? Only then does one adjust for what sets the present case apart. Daniel Kahneman and Amos Tversky proposed it to correct the planning fallacy, the tendency to judge a project from the inside and to ignore the failure rate of similar projects, and Bent Flyvbjerg put it into practice for large infrastructure projects. Actuarial science has long practiced it without the name: credibility theory weights an insured's own experience against that of its class, and the Bornhuetter-Ferguson method starts from an a priori loss ratio taken from comparables. For the underwriter facing a new risk, it protects against the persuasive story of a client who describes its case as unique.
In 2004, the UK Department for Transport adopted optimism bias uplifts calculated by Bent Flyvbjerg's team from comparable projects already delivered: the budget of a new rail project is adjusted according to the overruns observed in its class, before its particulars are even examined.
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.