This document provides a summary of preliminary research on the effects of district proliferation in Ghana. It finds that while new districts are justified as bringing development closer to people, politics often drives the creation of new districts. An analysis of Ghana's 2012 district splits found they were more likely in areas that voted for the ruling party and in poorer, smaller places. This has led to malapportionment and may contradict the goal of equal representation. However, new districts remain dependent on intergovernmental transfers. Further analysis of expenditures, development outcomes, and longer term impacts is still needed.
Danielle Resnick - Are there dividends from district proliferation? The case of Ghana
1. Are there dividends from district
proliferation?
The case of Ghana
(Preliminary research)
Danielle Resnick, IFPRI Washington DC
11.10.2015
2. District Proliferation in Africa
At least 25 African countries identify decentralization as a priority
in their national development strategies and focus of 10% of
democracy and governance foreign aid
Administration divisions are viewed as most basic means of
“implementing” decentralization policies
• Zambia (2011 until present): more than 30 new districts and 1 new
province
• DRC (June 2015): increased from 11 to 26 provinces
• South Sudan (Oct. 2015): increased from 10 to 28 states
Justified as bringing “development closer to the people”
“One purpose of the new states is to decentralize power, placing resources
closer to the rural population while at the same time reducing the size of the
national government.” (Pres. Salva Kiir, Oct. 2015)
3. Developmental or Political?
Potential developmental goals
• Each district has more resources through inter-governmental
transfers and more expenditures on goods and services
(Dollery et al. 2006; Tiebout 1956)
• Downward accountability because administrative units are
closer to population
• Upward accountability because citizens can reward/sanction
through elections (Brinkerhoff 2010; Rodinelli et al. 1983)
Potential political goals
• Vote buying between parties and citizens (Khemani 2010)
• Elite patronage (Mohammad 2015) and building majority
support within parliaments (Snyder and Samuels 2001)
• Undermine influence of popularly, elected leaders or chiefs
(Lambright 2013; Resnick 2013)
4. Developmental or Political? Ghana Case
Drivers and consequences of district proliferation are way
of looking at trade-offs with decentralization
Questions Elements of Analysis
Why are some
districts split and
others not?
Driven by population, poverty, or politics
What are the
consequences of
splitting districts?
Politics
Voting, Malapportionment
Public finance
Revenues and expenditure allocations
Public goods and services (2000-2010)
Development outcomes (2000-2010)
5. Why Ghana?
One of most committed democratizers in Africa with
extensive decentralization
• Began decentralization in 1988, reinforced by 1992 Constitution
• Three level structure below government:
1) Regional Coordinating Committees,
2) Metropolitan, Municipal, and District Assemblies (MMDAs),
3) Sub-district structures
• Fiscal transfers through DACF, Common Fund, and DDF
• Devolved health, agricultural, and education responsibilities
Established two-party system
• NDC (1992-2000, 2008-today) and NPP (2000-2008)
DCEs are appointed by the President
6. From 110 to 216 Districts, 2000-2012
30
27
20
26
16
26
13
11
25
22
0
5
10
15
20
25
30
35
Ashanti Brong
Ahafo
Central Eastern Greater
Accra
Northern Upper East Upper West Volta Western
2000 2004 2008 2012
7. Ghana’s 2012 district creation
The President can, through executive instrument, declare an
area to be a district and assign a name to a district
• President directs Electoral Commission (EC) to make
recommendations, considering population, geographic
contiguity and economic viability of the area
• New districts inaugurated in June 2012
Concerned Citizens of Ghana took EC to court
• Claimed Government attempting to gain unfair majority in
Parliament
• New districts contravened Act 1993, Act 462 whereby there
should be equal representation in local government
8. Data Sources
Own built database of district splits since 2000
Elections
• Electoral Commission of Ghana and civil society sources
• Constituency level data aggregated to district level
• Available for parliamentary and presidential elections
Population, Poverty, and Development Data
• 2000 and 2010 Population and Housing Census
• 2012/2013 Ghana Living Standards Survey
Revenues and Expenditures
• Composite budgets (2011-2015) from MoFEP
• Audit reports of the MMDAs (various years)
• DACF allocations from Ghana’s Common Fund
9. Population as a criteria?
Median population of all
districts before 2012 was
111,504
• Median for split vs.
non-split was 131,424
and 98,046,respectively
• Statistically significant
10000
100000
1000000
10000000
pop20080 pop20081
Logofpopulation
10. Poverty as a criteria?
0
10
20
30
40
50
60
70
80
90
100
pov2008 pov20080 pov20081
PovertyRate
Median poverty of all
districts before 2012 was
26%
• Median for split vs.
non-split was 25 and 28
percent, respectively
• Not statistically
significant
11. Politics as a criteria?
Results from 2008
elections
Split (%) Not split (%) Total (%)
NPP won 31 50 45.3
NDC won 69 50 54.7
Total percentage 100 100 100
Total districts 45 125 170
Chi Square 4.968**
**P<0.05
12. Drivers of District Splitting in 2012
Independent
Variable
Full sample Excludes
metro areas
Log of population 3.129*** 3.768***
(0.679) (0.790)
Poverty rate 0.026** 0.028**
(0.011) (0.012)
Vote margin favoring
NDC in 2008
elections
0.017** 0.015**
(0.006) (0.007)
N 170 164
***p<0.01; **p<0.05
13. Malapportionment
• When the votes of some citizens outweigh those of
others
MAL = (1/2) 𝒔𝒊 − 𝒗𝒊
where si refers to the percentage of total parliamentary seats
allocated to district i and vi is the percentage of the total population
residing in district I
• Almost 16 percent of seats were allocated to districts that
would not have otherwise received
• Comparisons: South Africa (3%), Senegal (4%), Botswana
(9%), Malawi (14%), Tanzania (28%)
14. Is there a rural bias in district allocation?
Region Population
(2010)
Share of
pop, rural
(%)
Actual number
of
Constituencies
Average
population
per
constituency
Ideal number
of
constituencies
Change
between
Ideal
and
Actual
Ashanti 4,780,380 39.4 47 101,710.21 53 6
Brong-Ahafo 2,310,983 55.4 29 79,689.07 26 -3
Central 2,201,863 53.0 23 95,733.17 25 2
Eastern 2,633,154 56.4 33 79,792.55 29 -4
Greater Accra 4,010,054 9.3 34 117,942.76 45 11
Northern 2,479,461 69.9 31 79,982.61 28 -3
Upper East 1,046,545 79.0 15 69,769.67 12 -3
Upper West 702,110 83.8 11 63,828.18 8 -3
Volta 2,118,252 66.5 26 81,471.23 24 -2
Western 2,376,021 57.5 26 91,385.42 26 0
Total 24,658,823 49.1 275 89,668.45 275 0
Malapportionment of Constituency Size due to 2012 Splits
Population quota = total population by number of constituencies
15. Electoral Benefits of Splitting in 2012
**p<0.05
Results from 2012
elections
Split (%) Not split
(%)
Total
(%)
Parent New
NPP won 37.8 21.7 42.4 100
NDC won 62.1 78.3 57.6 100
Total percentage 100 100 100 100
Total districts 45 46 125 216
Chi Square 6.169**
17. New districts are poorer and smaller
29.5
35.3
24.2
0
20000
40000
60000
80000
100000
120000
0
5
10
15
20
25
30
35
40
Parent District New District Non-Split District
Medianpopulation
Medianpovertyrate
• 67 percent of the new districts have less than 75,000 people
18. Preliminary Conclusions
Important political motivations for dividing districts
• Targeting seems aimed at securing a parliamentary
majority and elite patronage but it’s also “rewarded” by
citizens in subsequent elections
Newer districts are poorer, which maintains dependence on
DACF and also violates the “economic viability” stipulation
that the government is obligated to follow
Possible contradiction in bringing resources closer to people
even as others lose representation through
malapportionment, supporting suggestions of “rural bias”
19. Next Steps
Examine whether the results hold up to parliamentary
elections and if splitting helps ruling party “consolidate”
their hold
Examine the full distribution of expenditures, including
the share allocated to administration, within the
composite budgets
Analyze over time to trace whether split districts have
had substantially different development outcomes by
using the 2000 and 2010 Censuses
20. Intergovernmental Transfer Changes
Years No.
Districts
Total
population
(millions)
Average
Population
per district
Total DACF,
real GHC
Millions
DACF per
district, real
GHC
DACF per
person, real
GHC
2000 110 18.9 171,818 75.7 688,227 4.0
2001 110 19.3 175,455 70.9 644,340 3.7
2002 110 19.8 180,000 81.1 737,592 4.1
2003 110 20.3 184,545 154.1 1,401,078 7.6
2004 138 21.2 153,623 178.4 1,292,865 8.4
2005 138 21.4 154,855 126.9 919,341 5.9
2006 138 21.9 158,551 139.2 1,008,406 6.4
2007 138 22.4 162,246 149.1 1,080,264 6.7
2008 170 22.9 134,706 155.3 913,331 6.8
2009 170 23.4 137,765 115.7 680,542 4.9
2010 170 24.2 142,529 179.3 1,054,702 7.4
2011 170 24.6 144,765 183.2 1,077,596 7.4
2012 216 25.9 119,769 231.4 1,071,381 8.9
2013 216 26.5 122,593 225.4 1,043,517 8.5
Sources: Calculated from World Development Indicators, 2000 and 2010 Census, and the District Assemblies Common Fund (DACF)
Editor's Notes
Would note that this will most likely be more than one paper
Need to look at all of these together; most work only focuses on one element of it
Accompanied by increase in 200 to 275 constituencies; note that this corresponds to election years
NEED TO UPDATE BOX AND WHISKERS
Need to give the odds ratios;
Need to mention what happened in 2008 with NPP; same dynamic