Abstract
Anchoring describes the tendency for people's estimates of unknown quantities to be affected by numbers (or information) they have recently seen (or vividly remember). In the oil & gas industry, anchoring is regarded as a potential problem resulting from, for example, the sparsity of data that leads to the common practice of using analogues. "Base Case" or "Reference Case" models can also act as anchors. Much of the anchoring research, however, ignores the question of what estimates look like in the absence of anchoring values. Here we discuss previous research on anchoring and examine the conditions under which anchoring provides a benefit to decision making and whether these occur within the oil industry. We present the results of three studies, the first showing that, while anchoring results in greater deviations from the true value on average, it can also restrict the range of estimates so as to actually reduce the expected error in any single estimate.
The second study follows up on recent research showing benefits from elicitation techniques that cause a decision maker to repeatedly probe their knowledge. Thus, it tests the hypothesis that a decision maker who has seen multiple anchors may produce better estimates. Finally, in study 3, we demonstrate that people with greater knowledge are, in fact, less susceptible to anchors.
We conclude that, while anchoring can cause significant bias in estimates, the judicious use of anchoring values will not necessarily result in worse estimates. Instead, we identify circumstances where providing one or more anchoring values can produce a net benefit in terms of accuracy. Given the use of analogues and "bases cases" within the industry, understanding how anchors affect estimates is essential to improving estimates. This paper demonstrates that anchoring need not always be regarded as a bias to be overcome - rather, sometimes it is a tool that can be used to improve estimates.