Denis Mukama, Head of Research, Rwanda Revenue Authority (RRA); Giulia Mascagni, Fellow, Institute of Development Studies (IDS) and Research Director, ICTD; and Fabrizio Santoro, ICTD Research Officer
4. Overview of the project
• Genesis
- Improving voluntary compliance with limited resources – call for internal assessment
- TADAT assessment and emphasis on evaluation of compliance initiatives
- Institutional focus on internal data consumption (Data analytics)
• Main challenges
- Quick solutions to tax admin challenges Vs longer term research
- No data (linkage of training to declaration, reasons for non attendance)
- Capacity to undertake program evaluation
• Involvement of ICTD
- Previous work on compliance projects – revenue benefit
- Experience in experimental research (balance btn long term and quick wins, capacity
building opportunities)
7. Two Phases: Phase 1
• Objectives:
• Information about attendance and self selection
• Effectiveness on knowledge and perceptions
• Get TP’s feedback on the training
• Data: attendance data and survey
• Small sample
• New registration process
• Attendance data to allow matching with declarations
• Pilot a new registration process – TIN, business name, participants name (to allow
matching with declarations – phase2)
8. Two Phases: Phase 2
• Objectives:
• Analyse effectiveness on actual compliance
• Pilot a new way to boost attendance (experimental)
• Pilot a new hand-holding program
• Data: attendance and administrative data
• Data from tax declarations
• All new taxpayers
• Attendance data on all trainings (new registration procedure)
9. RRA invites all new
taxpayers to the training
Some taxpayers attend the
training
Effects of the training on:
• Knowledge
• Perceptions
• Compliance behaviour
Some taxpayers do not
attend the training
No expected effects of the
training (except possible
spillovers)
RRA’s list of
new taxpayers
Attendance data,
Survey data
Survey data
Data from returns
Data sources: Analysis:
1. Attendance rates
and self-selection
2. Effectiveness of
RRA education
program
Overview of evaluation
14. Attendance rates and self-selection
• How many taxpayers attend?
• Population of new taxpayers: 30% attendance rate
• Kigali: 23%
• Out of Kigali: 40%
• Survey participants: 53% attendance rate
• In Kigali: 51%
• Out of Kigali: 56%
• Self reported and actual attendance are largely consistent
• Who goes to the training? Self-selection
• Differences between attendees and non-attendees are not significant, except:
• Age: attendees are 1.5 years older
• Attendees have slightly higher knowledge (0.3 points)
• No difference in education
15. Effect on taxpayer knowledge
• 86% of attendees see an improvement in the knowledge index
• 52% increase in knowledge from training
• Controls: gender, age, usage of email, bank account, previously trained,
secondary education, above secondary education, Kigali, books, baseline value
Teach to comply? An experimental evaluation of
a taxpayer education program in Rwanda
Giulia Mascagni, Fabrizio santoro
October 15, 2017
Table 1: Training e↵ect on knowledge
(1) (2) (3)
Know di↵ Know gain Know index
Attendance 1.303⇤⇤⇤
0.518⇤⇤⇤
2.569⇤⇤⇤
(0.094) (0.069) (0.155)
controls yes yes yes
N 821 816 821
Table 2: Training e↵ect on perceptions
(1) (2) (3) (4) (5)
file getintouch fair duty evade
selfrep attend -0.077⇤⇤
-0.057⇤⇤
0.011 -0.004 0.007
16. Effect on taxpayer perceptions
• 8% and 6% increase in perceptions about complexity of the tax system
• “it is difficult to file” + “it is difficult to get in touch with RRA”
• Others not significant: tax system is fair, tax is a social duty, evaders will be caught, …
• Controls: gender, age, usage of email, bank account, previously trained, secondary
education, above secondary education, Kigali, books, baseline value
Table 1: Training e↵ect on knowledge
(1) (2) (3)
Know di↵ Know gain Know index
Attendance 1.303⇤⇤⇤
0.518⇤⇤⇤
2.569⇤⇤⇤
(0.094) (0.069) (0.155)
controls yes yes yes
N 821 816 821
Table 2: Training e↵ect on perceptions
(1) (2) (3) (4) (5)
file getintouch fair duty evade
Attendance -0.077⇤⇤
-0.057⇤⇤
0.011 -0.004 0.007
(0.037) (0.029) (0.023) (0.004) (0.006)
controls yes yes yes yes yes
N 793 815 803 819 800
22. Overview of evaluation
RRA invites all new
taxpayers to the training
Some taxpayers attend the
training
Effects of the training on:
• Knowledge
• Perceptions
• Compliance behaviour
Some taxpayers do not
attend the training
No expected effects of the
training (except possible
spillovers)
RRA’s list of
new taxpayers
Attendance data,
Survey data
Survey data
Data from returns
Data sources: Analysis:
1. Attendance rates
and self-selection
2. Effectiveness of
RRA education
program
Lottery
One-to-one Training