Revenue Windfalls & Accountability

How to cite

James, Alexander; Murphy, James; Servátka, Maroš. 2024. "Revenue Windfalls & Accountability." AEA RCT Registry. AEARCTR-0014212. https://doi.org/10.1257/rct.14212-1.0

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

Alexander James* , James Murphy , Maroš Servátka

* Primary investigator

Registry ID
AEARCTR-0014212
DOI
https://doi.org/10.1257/rct.14212-1.0
First registered
Last updated
Status
In development
Trial dates
2024-09-30 – 2024-10-05
Countries
Private
JEL codes
D72, H7, H27, H11
Registry record
http://www.socialscienceregistry.org/trials/14212

Abstract

Existing literature suggests that taxation is necessary for robust democratic representation. The argument is centered around the idea that money is not fungible. Rather, how money is spent depends on how money is obtained. In the context of public finance, this implies that people's preferences over government expenditures depend on how the public revenue was collected. Revenue generated from a broad-based tax is potentially viewed differently than revenue generated from, say, an intergovernmental grant or resource windfall. We explore this idea using an experiment in which people are assigned the role of either "Governor" or "Citizen". Depending on the treatment, Citizens receive either i) earned income or ii) a windfall. An exogenously determined tax is implemented and the resulting tax revenue is given to the Governor. The tax revenue can be embezzled by a Governor, or invested in a risky asset that pays the Citizen $1.20 with a 5/6 probability and $0 with a 1/6 probability. Modeled as a hidden-action game, should a Citizen receive $0, they do not know if the revenue was embezzled or if the investment paid $0 by chance. We also augment this basic framework with a treatment in which Citizens can punish Governors by reducing their earnings by $0.20. Outcomes of interest include i) the rate at which Governor's embezzle money in windfall and earned-income treatments (in both the "punishment" and "no punishment" treatments), ii) the rate at which Citizens punish Governor's in windfall and earned-income treatments.

Intervention

Subjects in this study will be randomly assigned the role of either "Citizen" or "Governor". Governors will tax either i) earned or ii) windfall income from the Citizen. The Governor can choose to keep the money for themselves or invest in a risky asset. In one branch of the experiment, the Citizen will have the opportunity to punish the Governor by reducing their pay once the Citizen realizes their return. The 2x2 design consists of interventions where i) the Citizen earning their money rather than simply being endowed with it and ii) the Citizen being able to punish the Governor rather than having no matter of recourse.

Experimental design

Subjects will be recruited on Prolific. Subjects in this study will be randomly assigned the role of either "Citizen" or "Governor". Governors will tax either i) earned or ii) windfall income from the Citizen. The tax is automatic and is not a choice. The Governor can choose to keep the money for themselves or invest in a risky asset. In one branch of the experiment, the Citizen will have the opportunity to punish the Governor by reducing their pay once the Citizen realizes their return. The interventions include i) the Citizen earning their money rather than simply being endowed with it and ii) the Citizen being able to punish the Governor rather than having no matter of recourse. The experiment is conducted using the strategy method. Subjects make decisions for all possible scenarios without knowing which one will be chosen by their counterparts. For example, Citizens are asked to choose a punishment should they be matched with a Governor who returns nothing to them and also choose a punishment should they be matched with a Governor who returns a positive amount to them.

Design details

Randomization method
Subjects are randomly assigned by Qualtrics into the various treatments.
Randomization unit
Individual
Clusters
N/A
Observations
We will have roughly 250 Citizens and 250 Governors in each treatment. With four treatments, our sample size will therefore be roughly 2000.
Arms
We will have roughly 250 Citizens and 250 Governors in each treatment (so our sample size will be roughly 500 per treatment).
Primary outcomes
We are interested in the rates at which i) Governors steal money across the two treatments (earned and windfall) and ii) the rate at which Citizens punish Governors across the two treatments (earned and windfall) and whether having the opportunity to punish affects the Governors' stealing.

Cited by

No citation data available for this registration yet.

Keywords