Accountability and taxation: Experimental evidence

How to cite

Sjursen, Ingrid Hoem. 2017. "Accountability and taxation: Experimental evidence." AEA RCT Registry. AEARCTR-0002233. https://doi.org/10.1257/rct.2233-3.1

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

Ingrid Hoem Sjursen*

* Primary investigator

Registry ID
AEARCTR-0002233
DOI
https://doi.org/10.1257/rct.2233-3.1
First registered
Last updated
Status
Completed
Trial dates
2017-05-24 – 2017-08-14
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/2233

Abstract

Corruption and mismanagement of government revenue is a widespread and serious obstacle to social and economic development in may poor countries. It has frequently been argued that taxing citizens is a promising strategy to promote good governance, because paying tax makes citizens demand more accountability in government expenditures. There is however very little causal evidence of the effect and what may be the underlying mechanisms. Using an online experiment, this project investigates whether and why demand for accountability is higher when revenue is generated through tax compared to when it is windfall. I propose that citizens care more about tax than windfall revenue for two reasons. First, tax revenue is generated by citizens’ hard work, increasing the weight they attach to revenue being spent fairly. Second, money paid in tax has been citizens’ pockets before being collected, increasing their expectations of what to get in return from their payment.

Experimental design

Two-player investment game with possibilities for punishment. The participants are randomly assigned to the role as leader or citizen. The leader decides how much of a group endowment to invest in a common pool with a multiplier of 1.5. He or she gets the share not invested in the common pool for him or her self. The experimental treatments vary how the group endowment is generated.

Design details

Randomization method
Randomization is done by the survey software (Qualtrics).
Randomization unit
The treatments and roles are randomized at the individual level.
Clusters
2000 individuals.
Observations
2000 individuals .
Arms
200 citizen and 200 leaders in each of the five treatments.
Primary outcomes
Primary (citizens): Demand for accountability, measured by willingness to punish. Secondary (leaders): Supply of accountability, measured by share invested in the common pool.

Cited by

No citation data available for this registration yet.

Keywords