Sunk-cost Effect and Biased Beliefs

How to cite

QIAO, ZHONGHENG. 2024. "Sunk-cost Effect and Biased Beliefs." AEA RCT Registry. AEARCTR-0012836. https://doi.org/10.1257/rct.12836-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

ZHONGHENG QIAO*

* Primary investigator

Registry ID
AEARCTR-0012836
DOI
https://doi.org/10.1257/rct.12836-1.0
First registered
Last updated
Status
In development
Trial dates
2024-02-09 – 2024-12-13
Countries
Private
JEL codes
D81, D83, D91
Registry record
http://www.socialscienceregistry.org/trials/12836

Abstract

Although the sunk cost effect is widely discussed in different research settings, the behavioral mechanism underlying this phenomenon remains elusive. In this project, I plan to conduct a lab experiment to explore subjects’ beliefs and decisions in a two-stage investment task. I aim to investigate whether subjects hold biased beliefs about the project when sunk costs are present. Furthermore, I will explore the behavioral mechanisms that potentially connect subjects' beliefs and the sunk-cost effect.

Intervention

The investment tasks will be varied across treatments in a within-subject design. This variation includes a baseline investment task in which an ex-ante optimal choice might be ex-post suboptimal, a treatment where the ex-ante optimal choice is always ex-post optimal, and a treatment where computers make some decisions for the subjects. In addition, the information setting will be varied across treatments in a between-subject design.

Experimental design

This is a laboratory experiment, conducted in a standard university experimental economics lab, recruiting students broadly across the university. Subjects will be randomly assigned to a treatment, which is fixed throughout their experimental session.

The experiment comprises three blocks of 10-round repeated individual investment tasks, a lottery selection task, and a demographic survey. The three blocks of individual investment tasks consist of one block for each of the within-subject treatments. The order of the three blocks is randomized at the individual level.

Design details

Randomization method
Subjects will be recruited by email using ORSEE. They choose between a list of available sessions, and the session is randomized to a treatment before it is initialized. Within the session, individuals are randomized into different orders of tasks.
Randomization unit
Individual
Clusters
200 subjects, each subject is a cluster.
Observations
200 subjects.
Arms
I plan to collect data from about 150 subjects (6 sessions) in the standard information setting and about 50 subjects (2 sessions) in the rich information setting.
Primary outcomes
Investment rates First-order beliefs
Secondary outcomes
Learning behavior across rounds Risk attitudes

Cited by

No citation data available for this registration yet.

Keywords