An Experimental Study of State-Dependent Preferences: The Case of Inflation

How to cite

TAN, Yang. 2026. "An Experimental Study of State-Dependent Preferences: The Case of Inflation." AEA RCT Registry. AEARCTR-0018495. https://doi.org/10.1257/rct.18495-1.0

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

Yang TAN*

* Primary investigator

Registry ID
AEARCTR-0018495
DOI
https://doi.org/10.1257/rct.18495-1.0
First registered
Last updated
Status
Completed
Trial dates
2026-04-29 – 2026-05-01
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/18495

Abstract

This study examines how people make decisions when outcomes depend on uncertain future events, and whether their preferences change depending on the state of the world. Participants complete a series of simple decision tasks in which they allocate points between two accounts. Which account determines their final payoff depends on a real-world economic indicator—the U.S. inflation rate (CPI). In some cases, payment depends on whether inflation is above or below a historical benchmark; in others, it depends on a feature that is effectively random (such as whether a digit is odd or even). The study investigates whether individuals treat outcomes differently when they are tied to meaningful economic states versus purely random events. This allows us to test for “state-dependent utility” .

Experimental design

The experiment consists of two incentivized decision tasks (22 rounds each) and a final questionnaire. In each round, participants choose a point allocation between two accounts along a linear trade-off. Payments depend on realizations of U.S. inflation (CPI), with treatments varying the source of uncertainty: (i) comparison of future or randomly drawn inflation to a historical median, and (ii) parity (odd/even) of the inflation decimal digit. One round is randomly selected for payment, and outcomes are incentive-compatible.

Design details

Randomization method
Randomization is implemented by the experimental software using Python's random module in oTree. At the beginning of the experiment, each individual participant is independently assigned with equal probability to one of two between-subject conditions: Treatment 1, in which payment-relevant uncertainty is based on a randomly drawn historical U.S. CPI month from March 2025 to February 2026; or Treatment 2, in which payment-relevant uncertainty is based on the December 2026 U.S. CPI. No stratification or blocking is used. In addition to the treatment assignment, the software independently randomizes, at the individual level, the order of the two tasks, the order of the 22 decision rounds within each task, the decision round selected for payment from the 44 total decisions, and the rounds selected for certainty questions.
Randomization unit
The unit of randomization is the individual participant. Each participant is independently randomized into one of the two treatment conditions, and all additional order/payment randomizations are also conducted separately for each participant.
Clusters
300 individuals
Observations
300 individuals
Arms
300 individuals
Primary outcomes
The main outcome variables are participants’ allocation choices between the two accounts in each decision round (i.e., the point combinations chosen along the budget line), which reflect their preferences under different uncertainty structures. Additional outcomes include certainty ratings for selected decisions and individual characteristics from the post-experiment questionnaire (e.g., risk attitudes and demographics)

Cited by

No citation data available for this registration yet.

Keywords