What makes redistribution popular?

How to cite

Sun, Gregory. 2026. "What makes redistribution popular?." AEA RCT Registry. AEARCTR-0019220. https://doi.org/10.1257/rct.19220-1.0

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

Gregory Sun*

* Primary investigator

Registry ID
AEARCTR-0019220
DOI
https://doi.org/10.1257/rct.19220-1.0
First registered
Last updated
Status
Completed
Trial dates
2026-07-22 – 2026-07-23
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/19220

Abstract

This experiment validates the predictions of a structural model of voter preferences over redistribution policy. In a previous study, I had participants make choices over who people would rather transfer money to, where I varied the characteristics of transfer recipients. I used the data from this prior study to estimate a model of how citizens form preferences over redistribution policy. In this study, I wish to test the predictions of the model in terms of which policies are popular versus unpopular.

Intervention

Participants will be asked to choose yes or no about if they like the policy proposed in the experiment. The randomization will include three policies:

Policy 1: The policy described will be computed by solving the standard Mirrleesian social planner's problem. The policy will be motivated as helping medium and low-income Americans.
Policy 2: The policy described will be computed by solving the a modified Mirrleesian social planner's problem, with a direct penalty on labor supply effects of the policy. The motivation of the policy will be identical as Group 1.
Policy 3: The policy described will be identical to the one in Group 2, but it will be framed as also helping to mitigate concerns about job loss due to AI automation.

The randomization will simply involve randomizing the order of when these three policies are shwon.

Experimental design

The only outcome variable in the experiment is support for the policy described. As detailed above, the intervention is simply to change order in which the following 3 policy descriptions are shown.

Policy 1: The policy described will be computed by solving the standard Mirrleesian social planner's problem. The policy will be motivated as helping medium and low-income Americans.
Policy 2: The policy described will be computed by solving the a modified Mirrleesian social planner's problem, with a direct penalty on labor supply effects of the policy. The motivation of the policy will be identical as Group 1.
Policy 3: The policy described will be identical to the one in Group 2, but it will be framed as also helping to mitigate concerns about job loss due to AI automation.

Design details

Randomization method
Randomization will be implemented by a computer. More specifically, question order will follow a simple cyclic latin squares design, with row of the latin square chosen via the randomization engine of Qualtrics.
Randomization unit
Randomization is at the survey-taker level.
Clusters
750
Observations
750
Arms
250/group
Primary outcomes
Indicator for if the participant supports the policy or not.

Cited by

No citation data available for this registration yet.

Keywords