I evade. Do you? Compliance in groups and the effect of social proximity

How to cite

Homa, Sofie. 2023. "I evade. Do you? Compliance in groups and the effect of social proximity." AEA RCT Registry. AEARCTR-0012570. https://doi.org/10.1257/rct.12570-1.0

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

Sofie Homa*

* Primary investigator

Registry ID
AEARCTR-0012570
DOI
https://doi.org/10.1257/rct.12570-1.0
First registered
Last updated
Status
In development
Trial dates
2023-12-11 – 2023-12-31
Countries
Private
JEL codes
H2, H3, H26, C92
Registry record
http://www.socialscienceregistry.org/trials/12570

Abstract

Collaborative tax evasion is a commonly observable problem, and examples such as the “LuxLeaks” or the “Panama Papers” are only two among many instances of such illegal collusive behavior. To counteract tax evasion and hence induce behavioral change, information provision about other anonymous individuals’ behavior is a recurring type of intervention. However, in accordance with previous research, it is plausible to assume that its use to promote tax compliance and thus fight tax evasion may backfire, when noncompliance is widespread. Further, due to the asymmetric effect of peer information, which describes individuals conforming solely with observed examples of peer’s non-compliance with a certain (informal) rule, it seems sensible to expose individuals to information about peers with whom they are especially socially proximate. So far, there are only few insights on the channels through which the effect of peer information works. Prior research in other domains however suggests, that social proximity between two partners may affect how information about their behavior affects own behavior. Further, empirical evidence suggests social proximity to allow to elicit responses to both compliance and violation of certain (informal) rules. I thus consider it crucial to shed light on the effect of social proximity with peers on collaborative tax evasion efforts.
Using an online sample, I run a randomized control trial in the form of a collaborative tax evasion game. The outcome of interest is the subjects’ tax compliance (income declaration rates of their company’s income). The random treatment assignment to one out of three treatments regarding information on social proximity with peers, the baseline (T1), social proximity (T2), and no social proximity (T3), is based on the subjects’ entry in the study’s experimental sessions.
Given a certain baseline level of compliance, I hypothesize social proximity to decrease the erosion of compliance, as individuals exposed to information on social proximity with their peers, will conform with peer’s compliance and noncompliance to a similar extent. However, I assume this effect to only hold in socially proximate groups and to be absent within non socially proximate groups.

Experimental design

I inquire the causal effect of social proximity with peers on tax compliance. I implement a collaborative online experiment, in which subjects recruited via a crowdsourcing platform, are asked to make corporate income declaration decisions.

The collaborative online experiment takes place in a corporate context and consists of various decision-making rounds. Participants are matched with a “colleague”, i.e., another participant of the same study. The group's task is to file their company's tax return. In each decision-making round participants must individually decide on what fraction of their company's income to declare, but the amount declared in the final tax report is the average of the group's individually declared incomes. The tax report is subject to being audited. The corporate income may vary with each round.

In the beginning of the collaborative online experiment subjects are randomly assigned to one out of three treatments. Participants in the first treatment (the baseline) are exposed to the behavior of their peer (previous income declaration decisions). In the second treatment participants are not only exposed to the behavior of their peer, but also to the information of being socially proximate with their peer. The third treatment resembles the second treatment expect for participants being exposed to the information of being non socially proximate with their peer.

The main outcome of interest is the subjects' tax compliance rates. Compliance behavior is translated from the income declaration decisions made by the subjects in various decisions rounds and allows to determine the inquired causal effect by introducing treatment dummy variables into the regression.

Design details

Randomization method
The random treatment assignment to one out of three treatments regarding information on social proximity with peers, the baseline (T1), social proximity (T2), and no social proximity (T3), is based on the subjects’ entry in one of the experimental sessions of the study. For each treatment, one experimental session is taking place. The entry in one of the experimental sessions is random, as subjects are free to choose when to take part in the online experiment after its implementation. Subjects' characteristics are assumed to be randomly distributed over the different experimental sessions.
Randomization unit
Subjects taking part in the online experiment are randomly assigned to one out of three treatments based on their entry in one of the experimental sessions.
Clusters
I am planning on recruiting around 495 subjects to take part in the online experiment.
Observations
The planned number of observed income declaration decisions is 4950.
Arms
A planned total amount of 495 subjects will be randomly assigned to one out of three treatment. Hence, 165 subjects are assigned to each of the three treatments. This results in 1650 observations per treatment arm.
Primary outcomes
I consider tax compliance rates (individual income declaration rate of company income), obtained from an online experiment, as primary outcomes of the study.

Cited by

No citation data available for this registration yet.

Keywords