Applying Behavioral Insights to Improve Tax Compliance in Papua New Guinea

How to cite

Hoy, Christopher; Sinning, Mathias. 2019. "Applying Behavioral Insights to Improve Tax Compliance in Papua New Guinea." AEA RCT Registry. AEARCTR-0004056. https://doi.org/10.1257/rct.4056-7.0

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

Christopher Hoy* , Mathias Sinning

* Primary investigator

Registry ID
AEARCTR-0004056
DOI
https://doi.org/10.1257/rct.4056-7.0
First registered
Last updated
Status
Completed
Trial dates
2019-03-27 – 2020-03-05
Countries
Papua New Guinea ()
JEL codes
C93, H24, H25, H26
Registry record
http://www.socialscienceregistry.org/trials/4056

Abstract

The aim of this trial is to test the effectiveness of messages to improve tax compliance in Papua New Guinea (PNG) in a setting where the threat of punishment from non-compliance lacks credibility. We will test two types of interventions in collaboration with the Internal Revenue Commission (IRC). Firstly, we will test the effects of SMS messages on taxpayer behavior. Two alternative messages will be sent to taxpayers to remind them to lodge their tax return (Treatment Group 1A) and to inform them that their taxes are being used to fund education, healthcare and infrastructure (Treatment Group 2A). A control group will not receive any communication from the IRC over the duration of the trial (Control Group A). Taxpayers are being assigned randomly to the two treatment groups and to the control group. Secondly, we will test the effect of a flyer that is designed to encourage taxpayers to pay their taxes by informing taxpayers that their taxes are being used to fund education and healthcare. Taxpayers that receive a letter from the IRC for notification purposes will be randomly assigned to a group that will receive a letter and a flyer (Treatment Group 1B) and a group that will only receive a letter but no flyer (Control Group B).

Intervention

Two types of interventions will be studied:

1. SMS messages: a message that reminds taxpayer to lodge their tax return (Treatment 1A) and a message that informs taxpayers that their taxes will be used to fund education, healthcare and infrastructure (Treatment 2A). The messages are attached in Appendix A at the end of the pre-analysis plan.

2. Flyer: a flyer that informs taxpayers that their taxes will be used to fund education and healthcare (Treatment 1B). A draft flyer is attached in Appendix A of the pre-analysis plan.

Treatments will be compared to randomly assigned control groups consisting of individuals who are not being contacted by the IRC over the duration of the trial. The treatment effect of each intervention will be measured by comparing average outcomes between treatment and control groups. No interaction between treatments will be tested.

Experimental design

We will test two types of interventions in collaboration with the Internal Revenue Commission (IRC). Firstly, we will test the effects of SMS messages on taxpayer behavior. Two alternative messages will be sent to taxpayers to remind them to lodge their tax return (Treatment Group 1A) and to inform them that their taxes are being used to fund education, healthcare and infrastructure (Treatment Group 2A). A control group will not receive any communication from the IRC over the duration of the trial (Control Group A). Taxpayers are being assigned randomly to the two treatment groups and to the control group. Secondly, we will test the effect of a flyer that is designed to encourage taxpayers to pay their taxes by informing taxpayers that their taxes are being used to fund education and healthcare. Taxpayers that receive a letter from the IRC for notification purposes will be randomly assigned to a group that will receive a letter and a flyer (Treatment Group 1B) and a group that will only receive a letter but no flyer (Control Group B).

Design details

Randomization method
For the SMS trial, de-identified baseline data were provided by the IRC for the population of taxpayers registered to pay Salary and Wages Tax (SWT) and Goods and Services Tax (GST). The baseline data were used for stratified randomization. A list of variables that was used to stratify is provided in Appendix B of the pre-analysis plan. Randomization was done in Stata using the command -randtreat- (version 1.4, 5 April 2017) using a random seed. For the flyer trial, the population of taxpayers who missed the deadline for lodging taxes in the previous month is randomly divided based upon whether the last digit of their Tax Identification Number (TIN) is an odd or even number. Within this population, taxpayers with a TIN that is an odd number are assigned to receive a flyer in addition to the letter.
Randomization unit
Individual
Clusters
N/A
Observations
For the SMS trial, there are up to 23,489 taxpayers. For the flyer trial, it is expected there will be around 3,000-4,000 taxpayers.
Arms
For the SMS trial, there are around 8,000 taxpayers in each group. For the flyer trial, there will be around 1,500-2,000 taxpayers in each group.
Primary outcomes
The main outcomes of interest are: Timing of SWT lodgement; Number of times SWT was lodged; Amount of SWT paid; Total payment of SWT; Number of employees reported by taxpayers; Timing of GST lodgement; Number of GST was lodged; Amount of GST paid; Gross income reported by taxpayer Timing of Salaries and Wages Tax lodgement Number of times Salaries and Wages Tax was lodged Amount of Salaries and Wages Tax paid Total payment of Salaries and Wages Number of employees reported by tax payers Timing of Goods and Services Tax lodgement Number of Goods and Services Tax was lodged Amount of Goods and Services Tax paid Gross Income reported by tax payer

Cited by

No citation data available for this registration yet.

Keywords