Tax amnesties in low-income countries: Evidence from administrative and experimental data in Uganda

How to cite

Markhof, Yannick; Vincent, Rose Camille; Dietrich, Stephan. 2025. "Tax amnesties in low-income countries: Evidence from administrative and experimental data in Uganda." AEA RCT Registry. AEARCTR-0016605. https://doi.org/10.1257/rct.16605-1.0

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

Yannick Markhof* , Rose Camille Vincent , Stephan Dietrich

* Primary investigator

Registry ID
AEARCTR-0016605
DOI
https://doi.org/10.1257/rct.16605-1.0
First registered
Last updated
Status
On going
Trial dates
2023-11-30 – 2026-08-31
Countries
Private
JEL codes
H26, H71
Registry record
http://www.socialscienceregistry.org/trials/16605

Abstract

We study the effectiveness of tax amnesties in low-income countries in the context of a tax amnesty law passed by the parliament of Uganda in 2023. Using comprehensive administrative records on liabilities before and payments into the amnesty, two randomized controlled trials in which we send uptake nudges to eligible debtors, as well as original survey data we collect from 1,922 firms, we study uptake of the amnesty, its fiscal impact and distributional consequences, as well as debtors‘ underlying perceptions and attitudes.

Intervention

We conduct two interventions. In November and December 2023, we send uptake nudges to 6,479 firms in the form of two text messages highlighting the amnesty and that enforcement of unpaid debt will resume upon the amnesty‘s expiration on December 31, 2023. Exploiting the unanticipated extension of the amnesty, we conduct a second uptake nudge intervention in December 2024 with 90,000 eligible debtors highlighting either (i) the existence of the amnesty (information, active control), (ii) information + partial payment option, (iii) information + threat of compounding debt, (iv) information + debt‘s threat to business continuity, (v) information + return of enforcement action after the amnesty deadline. Using detailed administrative data on debtors and payments into the amnesty, we study the effect of our text-message nudges on amnesty uptake and payments.

Experimental design

In the first experiment, we randomly select 6,479 medium-sized, registered enterprises in Kampala to receive our messages or comprise the control group. As we did not have information on debtors at the time, this sample consists partially of firms without debt which we will filter out. Treatment firms receive two text messages, one on Nov 30, 2023 and one on Dec 14. The second experiment randomly selects 90,000 eligible debtors in Kampala. Each debtor receives a text message on Dec 20, 2024.

Design details

Randomization method
computer (Stata)
Randomization unit
individual firm
Clusters
Experiment 1: 6,479 medium-sized, registered enterprises in Kampala Experiment 2: 90,000 eligible debtors in Kampala
Observations
Experiment 1: 6,479 medium-sized, registered enterprises in Kampala Experiment 2: 90,000 eligible debtors in Kampala
Arms
Experiment 1: 3,240 control, 3,239 treatment Experiment 2: 18,000 in each of the five treatment arms
Primary outcomes
amnesty uptake; amount paid into the amnesty

Cited by

No citation data available for this registration yet.

Keywords