The Sunk Cost Fallacy and Risk-Taking Behaviour. Evidence from a computer game experiment

Abstract

We examine whether behavioural sunk costs are related to an increased willingness to make risky decisions. Rational agents’ decisions should not be contingent on sunk costs; however, foregoing research suggests that individuals in fact do react to such costs. Few studies have examined behavioural sunk costs and risk-taking with “real stakes”, which is an important topic of research as many projects require investments in time and effort. If it is the case that behavioural sunk costs influence risk-taking decisions, it will be a particularly important finding in the field of risk-taking in avalanche terrain, as these types of activities are associated with large behavioural sunk costs. Our analysis is based on data from an experiment held at the start of 2020, with participants (N=65) from the psychology faculty at the University of Tromsø – The Arctic University of Norway. We are unable to find evidence of sunk cost effects. We do, however, find that risk-taking falls with time spent playing the game. This finding can either represent a learning effect or perhaps a reversed sunk cost effect.