Taxation and accountability in sub-Saharan Africa: New Evidence for a Governance Dividend
1. Taxation and Accountability in sub-Saharan Africa
New Evidence for a Governance Dividend
Roel Dom
Institute of Development Studies
February, 2019
Annual Centre Meeting
International Centre for Tax and Development
2. Overview
Puzzle Historical link between taxation and accountability in the
West, but less clear for contemporary developing countries
given different context.
Theory Taxation as a bargaining process between rulers and citizens
in which citizens trade compliance for influence in the
decision-making process.
Question Do increases in the tax ratio translate into increases in
accountability scores for African countries?
Strategy (Dynamic) FE and IV estimation
Result Evidence for a small but positive impact of taxation on
accountability.
1
3. Table of contents
1. Overview
2. Motivation
3. Literature
4. Data & Methodology
5. Results
6. Magnitude of the effect
7. Conclusion
2
4. Puzzle
Historically, the link between tax and accountability is well-established:
• England (1688), Parliament versus the Crown
• France (1789), Louis XV convenes the Estates General
• US (1776), “no taxation without representation”
Today, this governance dividend argument features heavily in the policy
discourse of international development actors.
However ...
... questions about validity for developing countries nowadays
... “evidence was never abundant”
3
5. Theory
Accountability: institutionalised system through which rulers justify
their actions to citizens, and through which citizens can reward or punish
rulers (Schedler, 1999).
Tax bargaining:
• rulers need revenue (Tilly, 1992)
• citizens control tax base and want influence (Bates & Lien, 1989)
• tax compliance exchanged for improved accountability (Levi, 1988)
But current circumstances not the same as historical conditions...
... more revenue sources, e.g. natural resources and aid (Moore, 1998)
... wider set of tax tools with varying saliency (Chetty et al. 2009)
4
6. Evidence
Qualitative work provides narratives for particular cases:
• Colonial export tax on sugar in Mauritius (Brautigam, 2008)
• Local tax revenue and in Somaliland (Eubank, 2012)
• Comparative analysis of Ghana, Kenya and Ethiopia (Prichard, 2015)
The body of quantitative work is small but growing:
• Sub-national variation in tax links with democratic rule (Berger, 2009)
• Cross-country taxes are positively correlated with measures of
democracy (Ross, 2004; Baskaran, 2014 ; Prichard et al. 2018)
Limitations:
• Endogeneity
• Accountability = democracy
5
7. Data
Panel 47 African countries from 1980 until 2015
Revenue ICTD Government Revenue Dataset
Accountability Varieties of Democracy (V-DEM) dataset
Accountability
Horizontal
(within State)
Diagonal
(State - Media/CSO)
Vertical
(State - Citizens)
ElectionsParty
6
8. V-DEM vs. WB measure
30
35
40
45
50
55
Accountabilty
1980 1990 2000 2010 2020
Notes: The solid and dashed lines represent respectively the V-DEM and World Bank
accountability measures. 7
11. EPRI
Pressure index
PIi,t = WE,i
∆Ei,t
Ei,t−1
− WRES,i
∆RESi,t
RESi,t−1
where E is the exchange rate in local currency units and RES the size of
the reserves. WE,i and WRES,i are country-specific weights.
Exchange Rate Pressure Index
ERPIi,t = sign(PI) × log(1 + |PI|)
Assumptions
1. Instruments correlated with tax
2. Shocks should only affect accountability through tax
• Relative changes, not levels
• Economic crisis → democratisation? Include growth rate as control.
• Trade → democratisation? Include trade as control.
10
12. Results - Baseline
FE FE FE-LDV FE-2SLS
Total tax 0.217*** 0.255*** 0.078*** 0.667**
(0.072) (0.064) (0.019) (0.296)
LT effect – – 0.279 –
N 1460 1389 1369 1059
Groups 47 47 47 45
Adj. R2
0.390 0.395 0.721 –
KP F-stat – – – 24.44
Controls –
Notes: Total non-tax is always included. Additional controls are:
GDPPC, ODA as well as Trade and Growth in the 2SLS estimation.
Robust standard errors in parentheses, clustered at the country-level.
***p ≤ 0.01, **p ≤ 0.05, *p ≤ 0.1.
11
13. Results - Sub-components
FE FE FE-LDV
Direct tax 0.211* 0.281** 0.087**
(0.116) (0.131) (0.039)
Indirect tax 0.296***
(0.105)
GST 0.144 0.064
(0.166) (0.063)
Trade tax 0.357** 0.092*
(0.145) (0.048)
N 1182 1046 1035
Groups 45 44 44
Adj. R2
0.384 0.371 0.706
Controls –
Notes: Total non-tax is always included. Additional controls
are: GDPPC and ODA. Robust standard errors in parentheses,
clustered at the country-level. ***p ≤ 0.01, **p ≤ 0.05, *p
≤ 0.1.
12
14. “Placebo” Test
Vertical Diagonal Horizontal Judicial
Total tax 0.067*** -0.001 0.006 -0.006
(0.021) (0.011) (0.012) (0.014)
N 1437 1437 1437 1437
Groups 47 47 47 47
Adj. R2
0.726 0.940 0.833 0.773
Notes: Total non-tax and lagged dependent variable always in-
cluded. Robust standard errors in parentheses, clustered at the
country-level. ***p ≤ 0.01, **p ≤ 0.05, *p ≤ 0.1.
13
16. A large-N perspective
I II III IV
Total tax 0.279*** 0.271*** 0.259*** 0.231***
(0.066) (0.071) (0.069) (0.068)
Urban Pop. 0.109 0.099
(0.310) (0.193)
Growth 0.034* 0.038**
(0.020) (0.018)
Education -0.395* -0.251*
(0.214) (0.149)
Equality 0.095 0.113
(0.178) (0.149)
Critical media 0.510*** 0.536***
(0.095) (0.098)
N 1258 1389 1389 1258
Groups 47 47 47 47
Adj. R2 0.382 0.397 0.501 0.501
Notes: Total non-tax, GDPPC and ODA included as controls. All regressors
are standardised in the sense that they represent deviations from the mean
divided by the standard deviation.
15
17. Conclusion
Puzzle Is taxation still linked with improvements in
accountability in contemporary developing countries?
Result (1) Evidence for a positive effect
(2) Heterogeneity across taxes
Implications (1) Support for state-building narrative in tax reform
(2) But tax reforms mostly focus on indirect taxation
(3) Effect is small
But ...
... what about cross-country heterogeneity?
... what are the necessary conditions?
... is there a responsiveness versus accountability trade-off?
16