Voice and Tax Compliance: Experimental Evidence from Kenya

How to cite

Wang, Tianhao; Gilmore, James; Howard, Peyton; Manda, Constantine; Cheruto, Lydia. 2026. "Voice and Tax Compliance: Experimental Evidence from Kenya." AEA RCT Registry. AEARCTR-0018902. https://doi.org/10.1257/rct.18902-1.0

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

Tianhao Wang* , James Gilmore , Peyton Howard , Constantine Manda , Lydia Cheruto

* Primary investigator

Registry ID
AEARCTR-0018902
DOI
https://doi.org/10.1257/rct.18902-1.0
First registered
Last updated
Status
In development
Trial dates
2026-06-14 – 2026-10-01
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/18902

Abstract

This study uses a randomized controlled experiment to test whether government-issued text messages that invite taxpayer input on public spending can improve property tax compliance among registered property owners in Nairobi County, Kenya. Before the tax deadline, the relevant tax and government authority sends two rounds of SMS messages to property owners randomly assigned to one of two treatment groups: a payment reminder that also invites them to indicate which local spending category should receive funding priority, or the same payment reminder without the invitation. A control group receives no message. Tax compliance is measured using administrative payment records (any payment, amount paid, share of bill paid). After the deadline, both treatment groups receive a single SMS eliciting beliefs about whether the county government responds to resident input; comparing beliefs and payments across groups tests whether any compliance response reflects belief updating rather than alternative explanations. A follow-up SMS survey measures beliefs about the use of tax revenue, attitudes toward government, institutional trust, and perceptions of public service quality.

Experimental design

Registered property owners with property tax obligations in Nairobi County, Kenya, are randomized between one control and two treatment groups, based on message content: no message (control), an SMS reminding owners of the upcoming property tax deadline, or the same reminder together with an invitation to indicate which local spending category should receive funding priority. The two treatment groups receive two rounds of SMS messages before the tax payment deadline. After the deadline, both treatment groups receive a single SMS eliciting beliefs about whether the county government responds to resident input; comparing beliefs and payments across groups distinguishes belief updating from alternative explanations for any compliance response. Following the government's delivery of the SMS messages, all participants are invited to complete an SMS survey. The survey includes items on moral attitudes toward non-payment, message recall and credibility, perceived use of tax revenue by the county and national government, trust in public institutions, ease of access to public services, and demographic information. Survey responses are incentivized through a lottery.

Design details

Randomization method
Randomization done by the relevant tax and government authority by computer (in Excel). Individual property owners are assigned to one of three groups (control, reminder, reminder plus participation), with the same number of subjects per arm. Randomization is at the individual level, stratified by electoral ward (85 wards in Nairobi).
Randomization unit
Individual property owner
Clusters
Errors will be clustered at the ward level, with 85 wards in Nairobi.
Observations
We plan to sample 150,000 property owners; the analysis on tax compliance will be done on the subsample that has not made payments by the treatment date.
Arms
For each treatment arm, we will sample 50,000 owners. These owners will be sampled proportionally across the different wards.
Primary outcomes
The key outcome variable is tax compliance, measured by total taxes paid and the percentage of taxes paid. Tax morale, measured by the responses to our survey, is also a key outcome variable.

Cited by

No citation data available for this registration yet.

Keywords