Towards a Multidimensional Measure of Governance

Report
TOWARDS A
MULTIDIMENSIONAL
MEASURE OF GOVERNANCE
SHABANA SINGH
VANDERBILT UNIVERSITY
APRIL 2011
“Government is not reason; it is not eloquence; it is
force! It is a dangerous servant and a terrible master.”
– George Washington, 1st President of USA.
What is governance?
 UNDP: "comprises the mechanism and process for
citizens and groups to articulates their interests, mediate
their differences and exercise their legal rights and
obligations.”
 European Commission: "the way public functions are
carried out, public resources are managed and public
regulatory powers are exercised".
 World Bank: "the state's institutional arrangements; the
processes for formulating policy, decision making, and
implementation; information flows within government;
and the overall relationship between citizens and
government."
INTRODUCTION
The available indices:
 Broad focus indices:
 Political Freedom Index (HDR, 1992)
 World Governance Indices by World Bank
 Mo Ibrahim’s Index of African Governance
 Narrow focus indices:
 Corruption Perception Index by Transparency
International
INTRODUCTION
Governance is a multidimensional phenomena
Political Freedom Index:
• Personal Security
• Rule of Law
• Freedom of expression
• Political Participation
• Equality of Opportunity
Worldwide Governance Indicators
• Voice and Accountability
• Political Stability and Absence of Violence
• Government Effectiveness
• Regulatory Quality
• Rule of Law
• Control of Corruption
Mo Ibrahim’s Index of African Governance
•Safety and Security
•Rule of law, Transparency Corruption.
•Participation and Human Rights
•Human Development
•Sustainable Economic Opportunity
INTRODUCTION
Broad concerns about governance indicators:
What is governance and what indicators should
be incorporated?
 (Arndt and Oman 2006), (Thomas 2010)
Quality of data and cross-country comparisons
 (Kaufmann and Kraay 2007), (Julius Court and Mease 2002)
and (Knack, Kugler, and Manning 2003)
The sensitivity of the measures to the scaling of
data
 This issue is the main focus of this paper
MO IBRAHIM’S IAG
The data:
 57 indicators of Governance for 48 countries in Africa
 Multiples sources for data, like Transparency International,
CIRI Human Rights Data
 18 of the 57 indicators are ordinal variables
Three tier structure to the index.
 57 indicators are divided into 15 sub-categories.
 15 sub-categories are aggregated up to five dimension
indices.
Five dimensions of Governance are aggregated to
get a single measure of Governance
MO IBRAHIM’S IAG
Governance
Safety and
Security
National
Security
Public Security
Rule of law,
Transparency
Corruption.
Ratification of
critical legal
norms
Existence of
Independent
and Efficient
Judicial
Systems
Corruption
Participation
and Human
Rights
Sustainable
Economic
Opportunity
Human
Development
Participation
Wealth
Creation
Poverty and
Inequality
Respect of
Human Rights
The Arteries of
Commerce
Health
Absence of
discrimination
Environmental
Sensitivity
Education
MO IBRAHIM’S IAG
Government involvement in
armed conflict
No. of battle deaths
No. of internally displaced
persons
National Security
Safety and Security
No. of civilian deaths due to
one-sided violence
Ease of Access to Small
Arms and Light Weapons
No. of refugees and asylum
seekers from this nation
Public Security
Level of Violent Crime
(Homicides Rates)
MO IBRAHIM’S IAG
Aggregation Methodology:

Rescaling of raw data (both ordinal and cardinal)
 − ()
 =
  − ()
 Three methods used for choosing the min and max:
 1st method allows for inter-temporal comparisons for each
country
 2nd method allows for cross country comparisons for each year
 3rd method : the “benchmark” case similar to the 2nd but uses
values from 2000

Scores can be below 0 and above 100
MAIN ISSUE
The IAG has 18 of its 57 indicators as ordinal variables.
IAG imputes cardinal values to ordinal data
• Problematic: Choice of scale can affect the rankings of different
nations
Example:
4 nations and 4 dimensions.
We use a simple average across the 4 dimensional values to arrive
at an overall measure.
Observations can take values on a scale of 0-10 in each
dimension. Higher values indicate better performance.
EXAMPLE- ORDINAL DATA
Dimension 1
Dimension 2
Dimension 3
Dimension 4
Average
Score
Country A
10
10
0
0
5
Country B
4.5
7.1
7.1
4.5
5.8
Country C
4.5
4.5
8.9
4.5
5.6
Country D
4.5
8.9
4.5
7.1
6.25
Nation A has the lowest score and Nation D has the
highest.
Undertake the following exercise, with the data above:
Rescale the data to a 0-100 scale.
Square the value in each dimension to get the new rescaled values.
Compute the composite index, as done previously (simple average of
the scores in each dimension).
EXAMPLE- ORDINAL DATA
Dimension 1
Dimension 2
Dimension 3
Dimension 4
Average
Score
Country A
100
100
0
0
50
Country B
20
50
50
20
35
Country C
20
20
80
20
35
Country D
20
80
20
50
42.5
Nation A now has the highest score!
Note:
The ranking of the nations in each specific dimension are preserved.
The aggregation (simple average) results in a new scheme for the composite
index.
REVISITING ALKIREFOSTER METHODOLOGY
 Identifcation stage: Dual cut-off
 Deprivation cutoffs idenfity whether deprived in the that index
 Dimension cutoff: No. of indicators one has to be deprived in
to be considered poor
 Aggregation stage : FGT based measure:
 For ordinal dimensions use the multidimensional adjusted
headcount (M0)
 For cardinal dimensions use the any of the class of measures
REVISIT: EXAMPLE
Original data: With a scale of 0-10
Dimension 1
Dimension 2
Dimension 3
Dimension 4
Av Score
Country A
10
10
0
0
5
Country B
4.5
7.1
7.1
4.5
5.8
Country C
4.5
4.5
8.9
4.5
5.6
Country D
4.5
8.9
4.5
7.1
6.25
Using the AF methodology:
Dimension 1
Dimension 2
Dimension 3 Dimension 4
Av. Score
Country A
0
0
1
1
0.5
Country B
1
1
0
1
0.75
Country C
1
1
0
1
0.75
Country D
1
0
0
0
0.25
Cutoff
5
8
4
5
Note: Lower scores imply better performance
REVISIT: EXAMPLE
Rescaled data: Scale 0-100
Dimension 1
Dimension 2
Dimension 3
Dimension 4
Av Score
Country A
100
100
0
0
50
Country B
20
50
50
20
35
Country C
20
20
80
20
35
Country D
20
80
20
50
42.5
Using AF Methodology on the new data:
Dimension 1
Dimension 2
Dimension 3
Dimension 4 Av Score
Country A
0
0
1
1
0.5
Country B
1
1
0
1
0.75
Country C
1
1
0
1
0.75
Country D
1
0
0
0
0.25
25
64
16
25
Cutoff
Notice: No change in rankings!
A NEW GOVERNANCE
INDEX
 Indicators are aggregated using AF to give dimension
specific governance indices
Average of these dimension specific indices gives overall
level of governance for the nation
Do not aggregate across nations.
 For the IAG the dimensions: Safety and Security, Rule of
law, Transparency Corruption & Participation and Human
Rights
 Treat them as ordinal , use M0
For IAG dimensions: Human Development & Sustainable
Economic Opportunity
 Treat them as cardinal, use M1
A NEW GOVERNANCE
INDEX
 For ordinal dimensions
G
s
i

 I ci  k
s
s

d

 j 1
s
w
s
j
I 

s
y

ij
z
  ;
j 

s
for s  1, 2 ,3
For cardinal dimensions
G
s
i

 I ci  k
s
s

 s
d


j 1

w
s
j
I 

y
s
ij

s  
 z j 


z
s

j
z
y
s
j
Overall governance is
Gi 
1
5
5

s 1
s
Gi
s
ij


;


for s  4 ,5
MAIN RESULTS
Country
Governance index
Rank
IAG Rank
Difference in Ranks
Cape Verde
0.043
1
3
-2
Mauritius
0.104
2
1
1
Lesotho
0.131
3
12
-9
Gabon
0.133
4
8
-4
Ghana
0.148
5
7
-2
Rwanda
0.303
17
18
-1
Swaziland
0.312
18
34
-16
Sierra Leone
0.317
19
37
-18
Benin
0.318
20
13
7
Burkina Faso
Central African
Republic
0.329
21
20
1
0.600
44
43
1
Sudan
0.659
45
45
0
Chad
Congo, Democratic
Rep.
0.693
46
46
0
0.723
47
47
0
Somalia
0.842
48
48
0
RESULTS (STRICT LINES)
Country
Governance index Rank
IAG Rank
Difference in Ranks
Cape Verde
0.128
1
3
-2
Mauritius
0.269
4
1
3
Lesotho
0.292
8
12
-4
Ghana
0.370
10
7
3
Gabon
0.398
11
8
3
Rwanda
0.428
12
18
-6
Benin
0.494
18
13
5
Burkina Faso
0.514
24
20
4
Swaziland
0.520
25
34
-9
Sierra Leone
0.532
28
37
-9
Central African Republic
0.679
43
43
0
Sudan
0.726
45
45
0
Chad
0.733
46
46
0
Congo, Democratic Rep.
0.792
47
47
0
Somalia
0.878
48
48
0
RESULTS (SOFTER LINES)
Country
Governance index Rank
IAG Rank
Difference in Ranks
Cape Verde
0.020
1
3
-2
Ghana
0.059
5
7
-2
Mauritius
0.073
6
1
5
Gabon
0.096
10
8
2
Lesotho
0.107
11
12
-1
Rwanda
0.131
15
18
-3
Benin
0.155
19
13
6
Burkina Faso
0.198
21
20
1
Swaziland
0.205
23
34
-11
Sierra Leone
0.233
28
37
-9
Central African Republic
0.439
44
43
1
Sudan
0.558
45
45
0
Chad
0.579
46
46
0
Congo, Democratic Rep.
0.669
47
47
0
Somalia
0.784
48
48
0
COMPARISON
IAG Methodology
Scaling of all indicators is
necessary
Gives value interpretation
to ordinal variables
Retain information on
depth of deprivation
within a dimension
(indicator).
New Methodology
No scaling required
Can be used with ordinal
variables
Gain information on
depth of deprivation in
governance (in terms of
k) across the indicators
Focus on deprived
nations and dimensions
(with the indicatorspecific cutoffs)
REPORT CARDS
The new index is a counting approach where zero
implies not governance poor and one implies
maximum deprivation.
This allows a very convenient representation a
nation’s performance.
• Governance Report Card
Report card for Rwanda:
Rwanda
Rank: 21
Dimension
Score
Best Score
Safety and Security
0.11
0
Rule of Law
0.67
0
Human Rights
0.11
0
Sustainable Economic Development
0.41
0
Human Development
0.22
0.12
Overall Governance
0.30
0.03

similar documents