The feeling of throwing good money after bad: The role of affective reaction in the sunk-cost fallacy.

Abstract

Continuing investing in a failing plan (i.e., the sunk-cost fallacy) is a common error that people are inclined to make when making decisions. It is impossible to get resources back that already have been invested. Hence, economic theory implies that decision makers’ decisions should only be guided by future gains and losses. According to the literature, the sunk-cost fallacy is driven by negative affect. Previous studies focused on negative incidental affect. We investigated, in contrast, whether the sunk-cost fallacy is caused by integral affect elicited by the specific decision context. Study 1 demonstrated a positive relationship between affective reaction and the sunk-cost fallacy. Study 2 replicated the finding in Study 1 in a within-subjects design, and demonstrated a full mediation of type of scenario (invest vs. non-invest) on the sunk-cost effect, mediated by integral affective reaction. A mediation using a within-subjects design additionally demonstrated that the effect is mediated by integral emotional responses experienced in relation to each scenario, and not by incidental emotional states that are unrelated to the scenarios. Study 3 replicated findings in the previous studies, and demonstrated that the relation between the sunk-cost fallacy and affect is moderated by justification. Participants who justified their decision were more resistant to the sunk-cost fallacy, and showed less negative affect elicited by the scenarios, than participants who did not justify their decision. Study 4 provided supporting evidence for our hypothesis by hindering conscious deliberation, and promoting reliance on affect, via cognitive load. The results showed that the relation between affect and the sunk-cost fallacy was stronger for participants under high cognitive load, than under low-load. The paper discussed how this research leads to new ways to protect against the sunk-cost fallacy in the discussion.