Effects of Warning Messages and Tax Inspections on Taxpayer Behaviour

Last registered on September 10, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

Effects of Warning Messages and Tax Inspections on Taxpayer Behaviour

RCT ID

AEARCTR-0008181

Initial registration date

September 08, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

September 10, 2021, 12:06 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Hungary

Region

Primary Investigator

Name

Bálint Ván

Affiliation

Ministry of Finance (Hungary), Tax Policy and Research Department

Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status

On going

Start date

2017-01-01

End date

2021-12-31

Keywords

Behavior, Firms & Productivity

Additional Keywords

Randomised Control Trial, Tax Inpection, Online Cash Register

JEL code(s)

H26

Secondary IDs

Prior work

This trial is based on or builds upon one or more prior RCTs.

Abstract

Tax administrations have limited resources to ensure tax compliance, and thus it is important to use the tools they have in the right combination. Online cash registers were introduced in Hungary in 2013-2014, they are compulsory equipment in certain sectors and they communicate directly with the tax agency. With a series of experiments I investigate the effect of warning letters and local inspections in relation to online cash registers.
At the beginning of April, 2019 a randomized experiment was conducted with Hungarian taxpayers. Selected companies using online cash registers (OCRs) received a warning letter. The letter warned the reader to use the cash register for every sale they make as the agency is monitoring incoming data continuously and suspicious companies are selected for tax inspection. It was a novel approach for the tax administration to send this kind of reminder. Results showed that after receiving the letter, the number of receipts and turnover recorded by the cash registers temporarily increased. The increase was concentrated among taxpayers who did not give receipts during an earlier tax inspection. A series of experiments is planned to find out the best way to increase tax compliance among companies using online cash registers.
During the summer of 2021 a territorially concentrated tax inspection campaign was conducted in the vicinity of lake Balaton, which is a main tourist destination of Hungary during the summer. In the first phase during July several local tax investigations have been conducted. At the beginning of August warning letters have been sent out based on the results of the first phase in a randomised control trial to businesses around Balaton as well as to businesses around other lakes with touristic relevance in Hungary. For the second phase of the inspection campaign conducted during August businesses were selected for inspection in a randomised way and with regard to the results of the first phase of the campaign. The goal of the research is to find out how does a warning letter and a tax inspection compare in their effectiveness in increasing tax compliance and what is the best way to combine them.

External Link(s)

Gitlab project for the code related to the research.

Registration Citation

Citation

Ván, Bálint. 2021. "Effects of Warning Messages and Tax Inspections on Taxpayer Behaviour." AEA RCT Registry. September 10. https://doi.org/10.1257/rct.8181-1.0

Sponsors & Partners

Partner

Name

National Tax and Customs Administration of Hungary

Type

government

URL

https://nav.gov.hu/

Experimental Details

Interventions

Intervention(s)

The intervention has three parts:
1. Warning messages around lake Balaton.
Sending messages online through the official channel. ("Cégkapu") The messages sent had two versions. Both versions asked the taxpayers to record every transaction in their online cash registers. Version "A" highlighted the tax inspection campaign of the summer and the probability of being inspected. Version "B" did not mention the inspection campaign and wrote about the possibility of inspections in a general way only. Version "B" is very similar to the letter sent during the earlier campaign in 2019.
2. Warning messages around other lakes.
The messages sent in the campaign around other lakes are different, because there was not concentrated tax inspection camapign in these areas during this summer. One of the messages is the same as version "B", but is called "C" for easier distinction between the interventions. The other version ("D") is only sent among companies who were found noncompliant during the last two years at least once. It mentions that this comany has been found noncompliant in the past.
3. Tax inspections around lake Balaton.
For the second phase of the tax inspection campaign around lake Balaton the inspections were randomised. For the locations selected the inspectors were told to conduct a tax investigation, but they allowed to make inspections at other companies too. All the tax inspections are local and their main goal is to see if the online cash machines are operated correctly and a receipt is given to the customers or not. This part of the inspection campaign happened in August 2021.

Intervention Start Date

2021-07-01

Intervention End Date

2021-08-31

Primary Outcomes

Primary Outcomes (end points)

Turnover recorded by the online cash registers aggregated by day of purchase.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

Number of transactions recorded by the online cash registers aggregated by day of purchase.

Secondary Outcomes (explanation)

Experimental Design

1. Warning messages around lake Balaton.
Blocks were created based on two variables: number of business locations (one or more) and results of previous tax inspections in the first (July) phase of the tax inspections campaign (inspection(s) with condemning decree, everything else). Companies were distributed equally among treatments "A" and "B" and the control group.
2. Warning messages around other lakes.
Blocks were created based on two variables: number of business locations (one or more) and results of previous tax inspections in the last 2 years (no inspection, inspection(s) without condemning decree, inspection(s) with condemning decree). Companies which had a condemning decree were distributed equally among treatments "C" and "D" and the control group. Other companies were distributed between "C" and the control group.
3. Tax inspections around lake Balaton.
Blocks were created based on the different possible results of tax inspections in the first (July) phase of the tax inspections campaign. Companies for tax inspection were selected in a way that a total of 600 business locations were selected for inspections. Probability of selection was higher in blocks with fewer units.

Experimental Design Details

Randomization Method

Randomization by computer using a reproducible script in R. The used is available in the project's Gitlab repository.

Randomization Unit

For warning messages the randomization unit is the company.
For tax inspections the randomization unit is the shop or restaurant, the place of business. The two type of randomizations are not the same as some companies operate in multiple locations.

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

1. Warning messages around lake Balaton: 4713 companies
2. Warning messages around other lakes: 3552 companies
3. Tax inspections around lake Balaton: 7235 business locations

Sample size: planned number of observations

At this point it is not clear how long the time period will be for the data. I will assume 6 months. 1. Warning messages around lake Balaton: 7235 business location * 183 days = 1 324 005 observations. 2. Warning messages around other lakes. 4997 business location * 183 days = 914 451 observations. 3. Tax inspections around lake Balaton. 7235 business location * 183 days = 1 324 005 observations.

Sample size (or number of clusters) by treatment arms

1. Warning messages around lake Balaton.
1686 companies control, 1686 companies message "C", 181 companies message "D"
2. Warning messages around other lakes.
1571 companies control, 1572 companies message "A", 1570 companies message "D"
3. Tax inspections around lake Balaton.
507 business locations selected for tax inspection, 4206 business location not selected for tax inspection

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Supporting Documents and Materials

IRB