Selecting the best project: what the ranking may overestimate
A management team compares several investments and selects the one with the highest estimated value. The useful question is precise: does that estimate remain a reasonable expectation once the project has been selected?
What the literature establishes
James E. Smith and Robert L. Winkler (2006, Management Science) formalise the “optimizer’s curse”: selecting the highest estimate from several uncertain evaluations can lead to overestimating the selected project's value. In their framework, estimates that are unbiased before selection do not guarantee an unbiased estimate for the winner. A ranking can therefore select both a good project and a particularly favourable estimation error.
The related mechanism of the winner’s curse
Richard H. Thaler (1988, Journal of Economic Perspectives) examines a related problem in auctions for an asset whose value is common but uncertain. Winning the auction may indicate that the bidder produced one of the highest estimates, rather than discovered an exceptionally attractive opportunity. The situations differ, but share a mechanism: selection provides information about the possible error in the winning evaluation.
The premature conclusion
One might conclude that managers should avoid the top-ranked project or apply a uniform discount to every forecast. These studies justify neither rule. The approach developed by James E. Smith and Robert L. Winkler (2006, Management Science) instead involves revising estimates to account for their uncertainty and the available information about possible values. (our executive and employee training programmes)
What internal data alone cannot show
The finding of James E. Smith and Robert L. Winkler (2006, Management Science) is a modelling result, not a measurement of how frequently this phenomenon occurs in Neuchâtel companies. Returns below forecast may also result from a market change or an execution problem. If only selected projects are documented, distinguishing a general forecasting error from a selection effect becomes difficult.
A practical check in Neuchâtel
At a Neuchâtel company working in microtechnology, watchmaking or precision microelectronics, SHR’s “Decision-making” programme can use an equipment investment comparison as an exercise. Before the decision, record each option’s estimated annual benefit, uncertainty range and volume assumptions; then have someone reassess the selected option without knowing its ranking or the stated benefit. The verifiable measure is the difference, in francs per year, between the original estimate and this second assessment, followed by the difference between each estimate and the benefit observed at a review date set before the decision. Repeated across several decisions, this monitoring can flag recurring overestimation without, on its own, establishing its cause. To go further: explore the Decision-making training in Neuchâtel, or browse our executive and employee training programmes in Switzerland.
In pictures: Decision-making in Neuchâtel
- decision-making
- Neuchâtel
- research
- training neuchâtel