The Weight of the Past: Sunk Costs, Escalation of Commitment, and the Limits of the Evidence

A forward-looking decision-maker should ignore costs that cannot be recovered. A large be-havioral literature nonetheless reports that prior, irrecoverable investments of money, time,or effort push people to continue with courses of action they would not otherwise choose.This paper reviews that literature with attention to both its strongest findings and its weakestpoints. We separate the sunk cost effect (Arkes & Blumer, 1985) from escalation of commit-ment, which adds personal responsibility for the original decision (Staw, 1976); summarizelaboratory, professional-sports, and entrepreneurial evidence (Staw & Hoang, 1995; Camerer& Weber, 1999; Åstebro, Jeffrey, & Adomdza, 2007); and report a meta-analytic estimate of amoderate average effect (Roth, Robbert, & Straus, 2015). We then set out the evidence thatcomplicates the textbook account: incentivized and field studies that find small, erratic, or nosunk-cost effects (Friedman et al., 2007; Ashraf, Berry, & Shapiro, 2010); failed replications ofa foundational escalation experiment and of a de-escalation technique (Białek, We¸grzyn, &Meyers, 2021; Schulz-Hardt et al., 2021); and the argument that responding to sunk costs cansometimes be rational (McAfee, Mialon, & Mialon, 2010). We conclude with decision-processsafeguards, distinguishing those with direct experimental support from those that are plausi-ble but untested or contested, as ways of narrowing the resulting Rationality Gap.