Repeated Interventions for Long-Term Tax Compliance: Evidence from A Natural Field Experiment in Indonesia

How to cite

Satyadini, Agung. 2023. "Repeated Interventions for Long-Term Tax Compliance: Evidence from A Natural Field Experiment in Indonesia." AEA RCT Registry. AEARCTR-0011485. https://doi.org/10.1257/rct.11485-1.0

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

Agung Satyadini*

* Primary investigator

Registry ID
AEARCTR-0011485
DOI
https://doi.org/10.1257/rct.11485-1.0
First registered
Last updated
Status
In development
Trial dates
2023-06-05 – 2023-06-24
Countries
Private
JEL codes
C93, H25, H26
Registry record
http://www.socialscienceregistry.org/trials/11485

Abstract

The objective of this trial is to investigate the endurance effect of interventions on small businesses' tax compliance in Indonesia. Specifically, this study attempts to address whether there are sustainable effects rather than immediate effects following an intervention. We explore whether taxpayers will learn from previous treatments, or if regular communication is necessary to maintain long-term effects and sustainability. In doing so, we replicate and resend the identical treatment letters (i.e., deterrence, literacy, and public goods provision letters) one year after our initial interventions. The target group for this study comprises around 5,600 small taxpayers. The treatment groups will be contrasted with a control group comprising individuals who did not receive any letters. The target populations were randomly divided into four equal groups, with each group consisting of around 1,400 observations.

Intervention

The experimental design is a replication of the previous interventions that were concluded in December 2021. The design is focused on improving small businesses' tax compliance, particularly related to their tax monthly payment and tax return submission. Three treatments will be tested: (1) Treatment 1: "deterrence letter", which highlights the submission deadline, administrative late-filing penalty IDR100,000, utilisation of computerised audit, the possibility of utilisation computerised audit and further enforcement actions towards non-compliers; (2) Treatment 2: "literacy letter", in which we simplify the letters, adopts less-formal style letter, emphasizes the simple guidance on how to pay and report taxes, and attaches the QR code to access official mobile apps and DGT webpage; (3) Treatment 3: "public goods provision letter", which highlights the taxpayer’s contribution for national budget, particularly for COVID-19 expenditure and future generation education. The control group consists of individuals who are not received letters from the DGT. Copies of sample letters and emails are attached in "Docs & Materials" of this registration.

Experimental design

To accomplish a plausible stratified randomisation, taxpayers with comparable baseline characteristics were grouped into strata. Each case was randomly assigned to one of the treatment groups or the control group within each stratum. This technique provided an even distribution of taxpayers across groups based on their baseline characteristics. The random assignment resulted in the following allocation of taxpayers: (1) Treatment Group 1 (deterrence letters): 1,424 observations; (2) Treatment Group 2 (literacy letters): 1,379 observations; (3) Treatment Group 3 (public goods provision letters): 1,417 observations; and (4) Control Group: 1,379 observations.

Design details

Randomization method
The full de-identified population was provided by the DGT. Each case was randomly assigned to one of the treatment groups or the control group within each stratum. This technique provided an even distribution of taxpayers across groups based on their baseline characteristics. Randomization was performed using a random variable generator in STATA and a random selection of the seed. For stratified randomisation, the user-written command ’randtreat’ version 1.4 was utilised.
Randomization unit
individual taxpayers
Clusters
5600 taxpayers
Observations
5600 taxpayers
Arms
5600 taxpayers
Primary outcomes
The outcome measures consist of (1) a dummy indicating tax returns filing, (2) a dummy indicating non-zero payment, (3) a dummy indicating timely filing, (4) a dummy indi- cating both non-zero payment and timely filing, (5) a dummy indicating improvement in tax filing, which derived from the indicator of Compliance Risk Management (CRM) dashboard – a combination assessment score between taxpayer’s fiscal importance and risks 10, (6) a dummy indicating electronic payment, whether taxpayers utilised e-billing; (7) number of inbound communications to tax office, a log contact from taxpayer in the following month; and (8) amount of tax payment.

Cited by

No citation data available for this registration yet.

Keywords