Local authorities and tax collection: Experimental evidence from Tanzania

Last registered on September 11, 2021

Pre-Trial

Trial Information

General Information

Title
Local authorities and tax collection: Experimental evidence from Tanzania
RCT ID
AEARCTR-0004875
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, 12:00 PM EDT

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

Last updated
September 11, 2021, 1:05 PM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Primary Investigator

Affiliation
Mzumbe University

Other Primary Investigator(s)

PI Affiliation
MASSA Institute of Social Sciences Research
PI Affiliation
Simiyu Administrative Region
PI Affiliation
Simiyu Administrative Region
PI Affiliation
Simiyu Administrative Region
PI Affiliation
Tanzania Revenue Authority
PI Affiliation
Simiyu Administrative Region
PI Affiliation
Simiyu Administrative Region
PI Affiliation
MASSA Institute of Social Sciences Research
PI Affiliation
MASSA Institute of Social Sciences Research

Additional Trial Information

Status
Completed
Start date
2020-01-02
End date
2021-06-30
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Property tax is one of the potential revenue sources in developing countries, but has shown little contribution to the total tax revenue. The existing evidence indicates that property tax contributes on average less than 1% of GDP in developing countries and below 0.5% of GDP for African countries. The Tanzanian context of property tax is of no difference from other developing countries, where property tax is one of the potential tax base that has limited contribution to the total tax revenue. While we acknowledge that there are many studies using SMS reminders to influence tax compliance; we leverage this strategy to study the additional effect of involvement of local authorities in tax compliance. We employed RCT method to examine the effect of messaging strategies and involvement of local leaders to increase property tax payment compliance. Participants were randomly allocated to three treatment arms and a control group using big-stick randomization method which led to a final sample of 668 Mjumbes (local leaders) and 21,328 taxpayers. The first treatment arm involved SMS reminder with a threat of fine, and second arm combined treatment for the first arm with an additional threat of involving local leaders in the tax collection process. The third arm involved the treatment in the first arm with an actual request for help from local leaders to increase tax compliance. Results indicate that in all treatment arms, the regression coefficients show positive impact and are significant at 1%. Furthermore, when the coefficient of the first treatment (T1) in column 1 of table 5 are compared to the same column for third treatment (T3), the coefficient of the third treatment arm (T3) revealed a 2.2 percentage point increase in payment status for all tax payers. These results suggest that actual Mjumbe involvement triggered an additional effect in compliance relative to the standard reminder SMS alone.
External Link(s)

Registration Citation

Citation
Mwaijande, Francis et al. 2021. "Local authorities and tax collection: Experimental evidence from Tanzania." AEA RCT Registry. September 11. https://doi.org/10.1257/rct.4875-1.2000000000000002
Experimental Details

Interventions

Intervention(s)
A Randomized Control Trial (RCT) will be set to experiment the Effect of Mobile Phone Reminder Text Messages on Property Tax Compliance in Tanzania. A list of registered tax payers from tax authority will be obtained and then validated for quality check of phone numbers and other demographic information. Participants will be owners of properties (residential or business). From the current Tanzanian tax laws, property tax is imposed to buildings with residential or business purposes, but exclusion is given to public buildings, e.g., schools, health centers, and religious buildings such as mosques and churches. Random assignment draw will be made to assign the participants into two groups; treatment groups (T) and control (C). Randomization will be at the street level but outcome will be measured at the individual level of property owner for both groups. Peer effect design will be tested in the identified clusters. The treatment group will consist of one treatment arm (deterrence message) with three (3) varying levels described in section B. Participants will be selected from urban and semi-urban in which there are eligible property tax payers according to the law.
Intervention Start Date
2020-01-02
Intervention End Date
2021-06-30

Primary Outcomes

Primary Outcomes (end points)
Evidence of use of phone sms for increasing property tax payment compliance
Increased property tax revenues to government
Primary Outcomes (explanation)
Improved revenue collection from property tax

Secondary Outcomes

Secondary Outcomes (end points)
Government ability to provide public services in education, water, environment and waste management
Secondary Outcomes (explanation)
Property tax compliance rate

Experimental Design

Experimental Design
Deterrence message contents with varied exposure messages will be sent to clusters so as to measure the peer effects (PEs).
To test the peer effect, groups of 20 registered households will be created within the same street randomly assigned in Treatment arms indicated as:

o Treatment 1a: message content a, exposure level 1 (2 households in 20 hh group)
o Treatment 1a: message content a, exposure level 2 (10 households in 20 hh group)
o Treatment 1a: message content a, exposure level 3 (18 households in 20 hh group)
A Control: create groups of 50 registered property owners within the street, but No message /placebo-reminder message will be sent.
This means, a deterrent reminder message will be sent to the treatment cluster and send none to the untreated cluster groups. Testing for the peer effect, sms messages will be sent to property owners in randomly selected treatment clusters, and send no any messages to all property owners in the control clusters.
Experimental Design Details
Randomization Method
Computer randomization for assigning subjects into Treatment and Control groups
Randomization Unit
Clusters in Streets
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
Property owners will be randomly selected treatment and control groups registered by streets. We will therefore randomize clusters and individuals defined at street/sub-street level and test the peer effect by analyzing the compliance between treated and untreated clusters
Sample size: planned number of observations
groups of 20 registered households within the same street randomly assigned in Treatment arms as shown below: o Treatment 1a: message content a, exposure level 1 (2 households in 20 hh group) o Treatment 1a: message content a, exposure level 2 (10 households in 20 hh group) o Treatment 1a: message content a, exposure level 3 (18 households in 20 hh group
Sample size (or number of clusters) by treatment arms
Observed groups of 20 registered households within the same street randomly assigned in Treatment arms as shown below:

o Treatment 1a: message content a, exposure level 1 (2 households in 20 hh group)
o Treatment 1a: message content a, exposure level 2 (10 households in 20 hh group)
o Treatment 1a: message content a, exposure level 3 (18 households in 20 hh group)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
There are limited studies to use as reference point in the calculation of sample size. However, the information available at the Tax Authority indicates that compliance rate stands at 14%. We will therefore use this rate as a baseline outcome rate and set a 6% increase above the baseline rate arbitrary and therefore set the Minimum Detectable Effect (MDE) at 20%. Other information on power, significance level, and intra-cluster correlation are given below 80% power to detect the cause effect if does exist 5% significant level (alpha =0.05) Probability of success in control group = 0.14% Probability of success in treatment = 0.2 Cluster size = 20 Intra-cluster correlation = 0.4 Our outcome variable (compliance) is measured as binary variable. Based on the given information, we adopt the STATA command
Supporting Documents and Materials

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
Yes
Intervention Completion Date
June 30, 2021, 12:00 +00:00
Data Collection Complete
Yes
Data Collection Completion Date
June 30, 2021, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
21,328 property tax payers and 668 Mjumbes
Was attrition correlated with treatment status?
No
Final Sample Size: Total Number of Observations
21,328 property tax payers and 668 Mjumbes
Final Sample Size (or Number of Clusters) by Treatment Arms
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials