Content uploaded by Todd Landman
Author content
All content in this area was uploaded by Todd Landman on Jul 10, 2014
Content may be subject to copyright.
T
HE
I
NTERNATIONAL
C
OUNCIL ON
H
UMAN
R
IGHTS
P
OLICY
Review Meeting
Corruption and Human Rights
Geneva, 28-29 July 2007
C
ORRUPTION AND
H
UMAN
R
IGHTS
:
E
MPIRICAL
R
ELATIONSHIPS AND
P
OLICY
A
DVICE
Todd Landman &
Carl Jan Willem Schudel
© 2007, International Council on Human Rights Policy
ICHRP commissioned this document as a Working Paper. ICHRP has not edited it and is not responsible
for its accuracy or for the views and opinions expressed. ICHRP welcomes the re-use,
re-publication and re-distribution of this paper, appropriately cited.
A
BSTRACT
This paper explores the empirical relationships between corruption and human rights using
extant quantitative measures for a sample of 186 countries for the period 1980 to 2004. It uses
three measures of corruption and 17 measures of human rights, which are examined using
univariate, bivariate, and multivariate analysis and methods of estimation. The paper argues that
some measures of corruption and human rights are better than others based on an assessment of
their validity, reliability and temporal and spatial coverage. The statistical analysis shows that
more corrupt countries have worse records at protecting human rights, even after controlling for
other explanatory variables, such as the level of democracy, national income, population size,
government consumption, and regional control variables. The implications of these findings for
advocacy strategies are then addressed.
Table of Contents
T
ABLE OF
C
ONTENTS
A
BSTRACT
.............................................................................................................................................................. 1
I
NTRODUCTION
.................................................................................................................................................... 3
C
ORRUPTION AND
H
UMAN
R
IGHTS
................................................................................................................ 4
M
EASURING
C
ORRUPTION AND
H
UMAN
R
IGHTS
......................................................................................... 6
D
ATA
A
NALYSIS AND
F
INDINGS
...................................................................................................................... 9
I
MPLICATIONS FOR
A
NTI
-C
ORRUPTION
C
AMPAIGNS
................................................................................19
R
EFERENCES
.......................................................................................................................................................21
3
I
NTRODUCTION
1.
The explanation of the global variation in the protection of human rights has occupied the
attention of scholars in the social sciences since the late 1970s (Claude 1976; McCamant
1981), while the first cross-national statistical analysis on human rights was not conducted
until the late 1980s (Mitchell and McCormick 1988). Since that time, there has been a
proliferation of studies using increasingly large and complex data sets for which an
expanding list of independent variables has been specified (see Landman 2005a; Moore
2006). These variables most notably include the level, pace, and quality of economic
development (e.g. Henderson 1991; Poe and Tate 1994; Poe, Tate, and Keith 1999); the
level, timing, and quality of democratization (e.g. Davenport 1999; Zanger 2000b;
Davenport and Armstrong 2004; Mesquita, Downs, Smith, and Sherif 2005); involvement in
internal and external conflict (Poe and Tate 1994; Poe, Tate, and Keith 1999); and the size
and growth of the population (Henderson 1993; Poe and Tate 1994; Poe Tate and Keith
1999).
2.
In addition to these more general variables, there have been further and more specific areas
of research conducted that include such variables as foreign direct investment and/or the
presence of multinationals (Meyer 1996; 1998; 1999a; 1999b; Smith, Bolyard, and Ippolito
1999); the level of global interdependence (Landman 2005b); and the proliferation of
international human rights law (Keith 1999; Hathaway 2002; Landman 2005b; Neumayer
2005; Hafner-Burton and Tsuitsui 2005). Across all these studies, human rights are
operationalised to include the protection of various civil and political rights, or more
narrowly, ‘personal integrity rights’, and the data sets tend to vary across time (15 ≤ T ≤ 25)
and space (150 ≤ T ≤ 194), yielding a large total number of observations used for
econometric estimation of empirical relationships (2250 ≤ N*T ≤ 4850) (Landman 2005a).
3.
In drawing on the achievements of the cross-national statistical and comparative literature
on human rights, this paper explores the empirical relationship between corruption and
human rights using extant quantitative measures for a sample of 186 countries for the
period 1980 to 2004. It uses three measures of corruption and seventeen measures of
human rights, which are examined using univariate, bivariate, and multivariate analysis and
methods of estimation. The paper argues that some measures of corruption and human
rights are better than others based on an assessment of their validity, reliability and temporal
and spatial coverage. The statistical analysis shows that more corrupt countries have worse
records at protecting human rights, even after controlling for other explanatory variables,
such as the level of democracy, national income, population size, government consumption,
and regional control variables.
4.
To develop these arguments and sustain the findings, the paper is divided into four sections.
Section One outlines the concept of corruption and develops an argument about why there
ought to be a relationship between corruption and human rights based on ICHRP’s own
definitions of corruption. Section Two describes the main measures of corruption and
human rights, as well as the control variables. Section Three presents increasingly complex
statistical analysis that explores the empirical relationship between corruption and human
rights, including univariate, bivariate, and multivariate techniques. The fourth section
discusses the implications for advocacy and policy making in the area of anti-corruption.
4
C
ORRUPTION AND
H
UMAN
R
IGHTS
5.
The ICHRP uses the following definition of corruption adopted in the UN Convention
against corruption:
the bribery of national and foreign public officials, bribery in the private sector,
embezzlement of property by a public official, trading in influence, abuse of functions,
and illicit enrichment.
6.
It also concedes that this is a working definition that includes the private sector and that the
list of corrupt acts is not an exhaustive one, where interpretation is likely to enlarge the list
to include other acts in the future.
7.
In the spirit of the flexibility and interpretation that ICHRP encourages, we contend that
corruption can indeed take many forms and involves a significant grey area between and
among different sets of practices, institutions, and culture. In the short term, there are rational
reasons and incentive structures that encourage corrupt practices in which particular
individuals are able to make substantial private gains within the public and private sector.
These practices violate the norms and principles of openness, transparency, and
accountability. Such gains would not have been possible under conditions in which there is
public scrutiny of decision-making, oversight authority and mechanisms for horizontal
accountability, and a larger culture of integrity in public and private life.
8.
In the longer term, as corrupt practices are iterated over time they become institutionalised
and develop their own cultural logics that create a quasi-acceptance of such practices by
society at large. Typically, patron-client and neo-patrimonial forms of interest
intermediation create their own ‘acceptable’ systems of rules and norms in which it is
expected that one must pay tribute to the patron in return for tangible benefits. Pork barrel
politics treads a fine line between legitimate and corrupt forms of exchange. Where the
stakes are higher and resources limited or poorly distributed, the maintenance of gains
through corrupt practices not only creates a demand for the use of coercion and violence
among state and non-state actors, but also a supply of such violence in the form of the
violation of human rights. Moreover, corruption permeates state institutions in ways that
undermine the protection and promotion of human rights, which continue with impunity.
State agents within the police and judiciary can engage in corrupt practices where rapid
confessions gained through torture and other forms of ill-treatment are rewarded through
material and non-material means. We thus expect, ceteris paribus, that patterns of corruption
and the abuse of human rights ought to be related empirically.
9.
Since we adopt a quantitative approach in this paper as per its terms of reference, we can
test the general hypothesis that there is a relationship between corruption and human rights
using quantitative measures of both concepts that have been collected on a sample of
countries over time. Simple bivariate analysis can reveal the magnitude and significance of
this positive relationship using standard measures of correlation (e.g. Pearson’s r and tau b).
Indeed, Lindberg (2006: 153) reports preliminary findings that corruption and the
protection of civil liberties are related across 48 African countries. But such analysis says
little about the direction of the relation or the presence of other factors that may account for
human rights violations (see Diamond, Linz, and Lipset 1989; Lindberg .2006: 152-154).
5
10.
First, it is our view that corruption has a certain quality that makes it more institutionalised
and ‘sedimented’ within the organs of the state than everyday forms of human rights abuse
and that since corruption undermines mechanisms of accountability and oversight it should
be seen as prior to human rights abuse. Second, human rights violations occur for many
different reasons beyond the presence of corruption. As outlined in the introduction to his
paper, the extant social scientific literature is replete with possible explanations for
variations in human rights protection ranging from domestic variables such as democracy
and economic development to international variables such as foreign direct investment and
international human rights law. Corruption is thus one of many possible factors that
account for variation in human rights abuse.
11.
In the analysis presented here there is an underlying assumption that corruption ought to be
specified as an explanatory variable for human rights protection alongside other key features
of countries (see Figure 1). For the purposes of this paper, the other factors include the
level of democracy, the level of economic development, international trade, population size,
government consumption, and a series of regional ‘dummy’ variables that control for
possible differences in human rights between the countries in Africa, Latin America, Middle
East, Western Europe and North America, South Asia, and East Asia and the Pacific.
12.
As we shall see, the model also includes variation over time, the possible ‘feedback’ effects
between one year’s human rights record and another, as well as the presence of ‘error’ or
the variation in human rights violations that remains unexplained. But before examining the
descriptive patterns and empirical relationship between corruption and human rights, it is
first necessary to describe the main measures that we use.
Figure 1. Corruption, human rights, and other explanatory factors
Corruption
Other factors
Human rights
protection
6
M
EASURING
C
ORRUPTION AND
H
UMAN
R
IGHTS
13.
The analysis uses a global data set on 186 countries between 1980 and 2004. The process of
case selection turned mainly to questions of data availability over time and was in no way a
function of values on the dependent variable. Microstates with less than half a million
inhabitants were eliminated but the remaining cases provide meaningful geographical spread
across different regions of the world.
Corruption
14.
We use three measures of corruption: (1) the Corruption Perception Index (CPI) developed
by Transparency International, (2) the corruption index from the International Country Risk
Guide (ICRG) developed by the Political Risk Studies Group and (3) the corruption
indicator from the governance indicators developed by Kaufmann, Kraay and Mastruzzi of
the World Bank (see Landman and Häusermann 2003).
15.
The CPI and the ICRG are indices based on so-called ‘expert surveys’, which are typically
carried out on an annual basis. The experts include politicians, businessmen, scholars,
among others are thought to have in-depth knowledge of the country. The interviews are
then coded, and aggregated into an index. The World Bank index is a weighted average of
many different corruption indices, among which the CPI and the ICRG. The great difficulty
with expert surveys on corruption lies in their subjective nature and their small sample size.
The indices are aggregations of mere perceptions of corruption and are typically derived
from a sample of fewer than 100 people per country, where the ICRG measure is largely
constructed from perceptions of business elites. The use of subjective judgement is partly
explained by its ‘latent’ quality and its contested nature, while small sample sizes limit the
reliability of the indices.
16.
Of the three indices used here, the ICRG provides the most temporal and spatial coverage.
The data range from 1984 to 2002, which is a much longer time span than either the CPI or
the WB data can provide. A major weakness of the ICRG measure is its limited range (0-6)
and limited set of intervals (.5), which means that a country’s level of corruption has to
change significantly in order for this value to increase or decrease. Consequently, ‘within
country’ variation for this variable is smaller than ‘between-country’ variation (this will
become important for the multivariate analysis).
17.
The CPI has a larger range (from 0 to 10) and with more intervals (0.1), but its temporal and
spatial coverage is significantly smaller than that provided by the ICRG. Transparency
International started collecting its data in 1995 and has achieved global coverage only a few
years ago. This limited coverage has two effects. First, the number of observations is
relatively low, as the dataset used in this study covers the years from 1980 to 2003. Second,
the data are biased since they mostly comprise industrialised countries. Such a bias makes
sense at one level, as there have been more ‘expert reports’ available in developed countries
earlier. However, it means that an uneven majority of the CPI data concerns a small group
of states that performs relatively well on corruption, which will produce biases in the
observations and the results. Inclusion of regional dummy variables in the multivariate
analysis ameliorates this problem in some degree.
7
18.
Finally, the WB index shares some of the advantages and problems with the CPI. The
variable has a small range (from –2.5 to +2.5), but this is not so much a problem as it is
continuous between these values, with interval changes of one hundredth. This means that
even the smallest changes of the level of corruption are measured by this variable. The time
span however is problematic, since it is relatively short and skips a year each year between
1996 and 2002. This makes it hard to measure variation over time. Also, the number of
countries is small in comparison to the other two variables discussed here.
19.
All three variables share a methodological shortcoming by being inherently subjective. Both
the ICRG and the CPI are developed based on surveys. Therefore, they are prone to bias.
The WB index can overcome some of this criticism, as it is a weighted variable. This does
not mean that the WB index is superior to the other two. On the contrary, despite its
continuous nature the WB index is the least usable variable here. The fact that it misses time
points makes it unsuitable for analysis of within-variation. Also, it has the smallest number
of available countries of the variables presented here.
20.
The CPI scores better on this point and is also reasonably continuous, but suffers from
selection bias and has a small number of time points available for most countries in the
sample. Therefore, it is our view that of the three measures, and taking into account their
various strengths and weaknesses, the ICRG is the most useful variable for our purposes as
it provides the best temporal and spatial coverage.
Human rights
21.
Human rights are operationalised using several ‘standards-based’ (Landman 2004) human
rights scales: (1) the Amnesty International version of the Political Terror Scale, (2) the US
State Department version of the Political Terror Scale, and (3) a series of measures from the
Cingranelli and Richards human rights data set (www.humanrightsdata.com) (see UNDP
2006; Landman 2004, 2005a, 2005b, 2006).
1
22.
The two versions of the Political Terror Scale use a coding protocol to convert source
material about particular human rights practices into ordinal scales that range from 1 (low
violations) to 5 (high violations). The rights covered by these scales include ‘personal
integrity rights’ violations, such as political imprisonment, exile, arbitrary detention, and
forced disappearance. The human rights data from the Cingranelli and Richards use
narrower coding schemes (0-1, 0-2, and 0-3) and similar source material to provide separate
measures for disappearances, extra-judicial killings, torture, and political imprisonment; the
right to association, movement, speech, political participation, and religious freedom;
empowerment rights; worker rights; and women’s economic, political, and social rights. In
addition, Cingranelli and Richards provide a combined ‘personal integrity rights’ scale that
ranges from 0 (high violations) to 8 (low violations). For the purposes of comparability, the
subsequent analysis transformed all the human rights variables so that a low score denotes a
low protection of human rights (i.e. frequent violations) and a high score denotes a better
protection of human rights (i.e. less frequent violations).
1
We opted not to use the civil and political rights measures available from Freedom House since there are
significant problems with their source materials, transparency of coding procedures, absence of inter-coding
reliability tests, and additional problems that make them unreliable (see Munck and Verkuilen 2002).
8
23.
The main advantages of these scales include their wide temporal and spatial coverage, their
use of a standardised coding protocol that provides comparability, and their use of inter-
coder reliability tests and adjudication of scoring by the project coordinators and coding
teams. Their main disadvantages include over-reliance on single sources of information and
their inherent reductionism (known as variance truncation). Amnesty International and US
State Department annual reports provide particular accounts of country human rights
events, conditions, and practices, where subsequent analysis has identified significant biases
for or against particular sets of countries (see Landman and Häusermann 2003; Landman
2004, 2005b). Like the measures of corruption, these measures of human rights are limited
in their range, leading to a reduction in variation. They do not provide much differentiation
among the world’s best and worst protectors of human rights, and ultimately lead to a
three-level grouping of countries across the world (see Landman 2005b: 98-108). The large
number of observations across time and space, however, does provide significant variation
and degrees of freedom to carry out tests on the relationship between corruption and
human rights.
Additional variables
24.
The level of democracy is measured using the combined democracy scale from Polity IV,
which codes countries from -10 (full autocracy) to full democracy (+10) using a minimal
and procedural definition of democracy (see e.g. Jaggers and Gurr 1995; Foweraker and
Krznaric 2001; Munck and Verkuilen 2002; Landman and Häusermann 2003). The
economic variables all come from the World Bank’s World Development Indicators
(ww.worldbank.org). The level of economic development is measured using the logged
value of real per capita income (GDP, constant 2000 US $). Population size is logged. Trade
is measured as the total imports and exports as a percentage of GDP. Government
consumption is measured using total government expenditure as a percentage of GDP.
9
D
ATA
A
NALYSIS AND
F
INDINGS
25.
The first step in analysing our data is to examine the descriptive statistics for the corruption
and human rights variables. Table 1 lists their mean values, standard deviation, range
(minimum and maximum values) and the number of observations (N). The large number of
observations comes from the fact that we collected data on 186 countries for 25 years. The
descriptive statistics confirm the observation that we have sufficient variation and degrees
of freedom to carry out the multivariate analysis below.
26.
Figure 2 compares the mean corruption score across the different regions of Africa, Latin
America, Middle East and North Africa, Europe and North America, South Asia, and East
Asia and the Pacific. The scales for the ICRG and CPI are roughly comparable, while as
Table 1 shows, the range of values for the World Bank measure is completely different due
to the way in which it is calculated. Nevertheless, the figure shows that levels of corruption
are worse in South Asia, followed by Africa, Latin America, the Middle East and North
Africa, East Asia and the Pacific, and then Europe and North America. Figure 3 shows the
mean score for three of the main human measures (the two versions of the political terror
scale and the Cingranelli and Richards measure of physical integrity rights) across regions.
Again South Asia scores the worst, followed by very little difference between the remaining
regions other than Europe and North America
Table 1. Descriptive Statistics
Variable description
Acronym
Mean
Std. Dev.
Range
N
Corrup
tion
Corruption Perception Index
CPI
4.660
2.421
[0.4
–
10]
869
Country Risk
ICRG
3.204
1.414
[0
–
6]
2,483
World Bank
WB
-
0.064
1.012
[
-
2.13
–
2.52]
1,064
Human Rights
Physical Integrity Rights
PHYSINT
4.864
2.367
[0
–
8]
3,493
Disapp
earances
DISAP
1.653
0.645
[0
–
2]
3,510
Killings
KILL
1.317
0.778
[0
–
2]
3,507
Political imprisonment
POLPRIS
1.088
0.855
[0
–
2]
3,515
Torture
TORT
0.798
0.751
[0
–
2]
3,513
Empowerment
EMPINX
5.884
3.282
[0
–
10]
3,515
Association
ASSN
1.071
0.855
[0
–
2]
3,605
Movement
MOVE
0.706
0.456
[0
–
1]
3,527
Speech
SPEECH
1.039
0.740
[0
–
2]
3,526
Political Participation
POLPAR
1.104
0.854
[0
–
2]
3,525
Religious Freedom
RELFRE
0.617
0.470
[0
–
1]
3,526
Worker rights
WORKER
0.993
0.797
[0
–
2]
3,523
Women’s Econ Rights
WECON
1.316
0.634
[0
–
3]
3,455
Women’s political rights
WOPOL
1.702
0.662
[0
–
3]
3,510
Women’s social rights
WOSOC
1.228
0.838
[0
–
3]
3,408
Political Terror Scale
(Amnesty)
PTSAI
2.672
1.133
[1
–
5]
3,279
Political Terror Scale
(US
State)
PTSSD
2.493
1.169
[1
–
5]
3,554
10
02468
Africa Latin America MENA W Europe N America South Asia East Asia & Paciific
mean of CPI mean of ICRG
mean of WB
Figure 2. Mean corruption score across regions
0 2 4 6 8
Africa Latin America MENA W Europe N America South Asia East Asia & Paciific
mean of ptsai mean of ptssd
mean of physint
Figure 3. Mean human rights scores across regions
11
27.
Bivariate analysis of the corruption and human rights variables (reported in Table 2) shows
a remarkable consistency in the first order empirical relationships. All the correlation co-
efficients are significant at the 99.99% probability level of confidence (
p
< .001). Reading
down the co-efficients reported in the first three columns of Table 2 shows that the three
corruption measures are highly inter-correlated, and all the corruption measures are all
positively correlated with the various human rights measures. The co-efficients vary from a
low association of .26 to a high association of .71. This variance in the co-efficients is
expected since corruption will be differently related to different rights violations and
corruption is one of many factors that accounts for variation in human rights protection.
On balance, however, the table shows that as expected, countries with more corruption
have a worse record at protecting human rights.
28.
As outlined above, corruption is one of many possible factors that may account for the
variation in human rights protection. Thus, it is imperative to move the analysis beyond
simple bi-variate correlations to a fuller specified multivariate model that includes other
explanatory variables alongside corruption. We thus have specified the additional variables
drawing on the extant social scientific literature in this area. The analysis tests for the
independent effects of the corruption on human rights while also testing for the
independent effects of the other explanatory variables. In this way, the statistical estimations
control for the presence of the other variables while allowing us to focus on the main
relationship between corruption and human rights. Based on our arguments in Section One
of the paper, we specify human rights as the dependent variable, while corruption and the
other variables are considered the independent variables.
29.
Our data set follows by now what has become a standard construction of a matrix of cross-
section and time-series units, where variation in the variables and the number of
observations are maximised across time and space. Such data sets do, however present a
number of problems. In addition to the standard problems (for which we introduce
appropriate controls) such as auto-correlation and heteroscedasticity common to these data
sets (see Beck and Katz 1995), our data set has the additional problems associated with time
invariant or nearly time invariant variables (Plümper and Troeger 2007). Standard fixed
effects regression techniques for pooled cross section time series models that include such
invariant or nearly invariant variables have been shown to produce inefficient estimators,
which may lead to making false inferences about the empirical relationships that have been
analysed. Plümper and Troeger (2007) have devised a three-stage regression technique that
‘decomposes’ the explained and unexplained elements of the fixed effects and produces
final estimates that take into account the particular qualities of invariant or nearly invariant
variables.
30.
The basic rule of thumb in using this method of estimation is to compare the ‘between-unit
variation’ to the ‘within unit variation’ of our variables. If the between unit variation is 2.5
times greater than the within unit variation, then we specify the variables as invariant or
nearly invariant. In other words, those variables that exhibit greater variation across
countries than over time are considered time invariant or nearly time invariant. For our data
set, economic development, trade, government consumption, and population size are the
time invariant or nearly time invariant variables. We thus adopt the fixed effect vector
decomposition method of estimation and specify these variables in the procedure as
invariant.
12
Table 2. Correlation Matrix for all corruption and human rights measures
CPI
ICRG
WB
Physint
Disap
Kill
Polpris
Tort
Empinx
Assn
Move
Speech
Polpar
CPI
1.00
ICRG
0.81
1.00
WB
0.97
0.76
1.00
Physint
0.66
0.48
0.60
1.00
Disap
0.36
0.27
0.35
0.74
1.00
Kill
0.56
0.37
0.51
0.83
0.59
1.00
Polpris
0.46
0.38
0.42
0.78
0.41
0.4
6
1.00
Tort
0.68
0.68
0.58
0.78
0.39
0.56
0.48
1.00
Empinx
0.55
0.42
0.49
0.51
0.26
0.29
0.61
0.39
1.00
Assn
0.39
0.35
0.38
0.44
0.20
0.23
0.57
0.31
0.77
1.00
Move
0.39
0.26
0.35
0.40
0.24
0.25
0.44
0.29
0.76
0.50
1.00
Speech
0.54
0.40
0.49
0.47
0.26
0.29
0.55
0.35
0.80
0.67
0.49
1.00
Polpar
0.43
0.38
0.42
0.42
0.21
0.22
0.54
0.29
0.83
0.74
0.52
0.68
1.00
Relfre
0.35
0.26
0.29
0.30
0.13
0.13
0.41
0.25
0.74
0.49
0.45
0.45
0.47
Worker
0.51
0.39
0.44
0.41
0.19
0.28
0.45
0.34
0.75
0
.59
0.43
0.56
0.56
Wecon
0.63
0.46
0.57
0.41
0.23
0.32
0.36
0.35
0.43
0.37
0.28
0.36
0.41
Wopol
0.42
0.27
0.30
0.18
0.10
0.07
0.27
0.10
0.37
0.38
0.18
0.30
0.41
Wosoc
0.66
0.53
0.54
0.42
0.20
0.32
0.41
0.37
0.50
0.45
0.31
0.44
0.47
Ptsai
0.58
0.51
0.53
0.78
0.57
0.66
0.58
0.58
0.40
0.33
0.31
0.38
0.32
Ptssd
0.71
0.55
0.67
0.82
0.60
0.72
0.60
0.65
0.49
0.41
0.38
0.46
0.40
Relfre
Worker
Wecon
Wopol
Wosoc
Ptssai
Ptssd
Relfre
1.00
Worker
0.42
1.00
Wecon
0.25
0.40
1.00
Wopol
0.24
0.31
0.40
1.00
Wosoc
0.30
0.43
0.73
0.45
1.00
Ptssai
0.21
0.36
0.39
0.18
0.43
1.00
Ptssd
0.27
0.42
0.43
0.18
0.45
0.82
1.00
All correlations are significant at the p<0.001 level.
13
31.
In addition to these considerations relating to our method of estimation, we reduced the
number of human rights variables under consideration to include the two versions of
the political terror scale, the physical integrity rights measures from Cingranelli and
Richards, and a combined human rights score. For the combined score, the bi-variate
correlations in Table 2 show the existence of clusters of large and significant correlation
coefficients between the human rights scales, suggesting that they are measuring aspects
of the same underlying dimension. Given this high degree of agreement among the
different scales, we extracted a single component in an effort to reduce the group of
interrelated human rights variables into one common factor-score using regression.
2
The
resulting factor loadings (not reported here) showed a strong relationship between each
variable and the common underlying dimension they all measure. The resulting human
rights factor score ranges from low values denoting poor record of human rights
protection (i.e. high violations) to high values denoting a better record of human rights
protection (i.e. low violations).
32.
The results of the multivariate regression using the vector decomposition method of
estimation are reported in Tables 3a, 3b, 3c, and 3d. Table 3a reports the results for the
Cingranelli and Richards physical integrity rights measure of human rights and the three
different measure of corruption. The ICRG and World Bank measures of corruption
show a positive and significant relationship between better scores for corruption and
better levels of human rights protection. The CPI is not significant. These findings are
upheld in the presence of the additional independent variables and regional dummy
variables. Better levels of democracy and economic development are positively related
to better records of human rights protection, while trade and government consumption
are not significantly related to human rights. The regional variables show that while the
relationship between corruption and human rights is positive and significant, countries
in Africa, Latin America, South Asia, and East Asia and the Pacific start out with
significantly worse human rights records than those countries in Europe and North
America. We have thus controlled for the regional differences and have demonstrated a
positive and significant relationship between corruption and human rights.
33.
Table 3b reports the results for the political terror scale that codes the US State
Department reports, while Table 3c reports the results for the version that codes the
Amnesty International reports. The results across the two tables are consistent. Only the
ICRG measure of corruption is significant, while the co-efficients for the other variables
show approximately the same magnitude and direction and levels of significance as
those reported in Table 3a. The absence of significance for the World Bank and CPI
measures of corruption is partly explained by the fact that they have not been produced
on annual basis, and in the case of the CPI, were initially produced for developed
countries only. Finally, Table 3d reports the results for the combined human rights
factor score. As in Table 3a, the ICRG and World Bank measures are significant, while
the CPI is not. The co-efficients for the other variables have approximately the same
magnitude, direction and significance.
2
Given a different time coverage across the scales, we adopted the ‘substitute missing values with the
mean’ option to deal with missing cases, and ensure the widest coverage of the factor-score. This
procedure is justified by the fact that missing cases are randomly distributed both across indicators and
across countries (note also that for each country year between 1980 and 2003, at least 2 indicators were
available).
14
Table 3a Physical Integrity Rights (Cingranelli and Richards)
ICRG
CPI
WB/KKM
ICRG
0.065*
(0.038)
CPI
0.080
(0.095)
WB/KKM
0.472**
(0.206)
Polity 4 (net)
0.035***
(0.008)
0.057***
(0.013)
0.044**
(0.015)
Ln GDP per ca
pita
0.405***
(0.090)
0.458**
(0.149)
0.164
(0.156)
Ln Population
-
0.746***
(0.081)
-
0.560***
(0.128)
-
0.695***
(0.137)
Ln Trade
-
0.048
(0.071)
-
0.440***
(0.102)
0.174
(0.124)
Ln Government
Consumption
0.013
(0.077)
-
0.148
(0.126)
0.113
(0.132)
Africa -0.599***
(0.121)
-1.120***
(0.145)
-0.233
(0.151)
Latin America
-
1.513***
(0.111)
-
1.729***
(0.130)
-
0.863***
(0.155)
Middle East
-
1.451***
(0.120)
-
2.398***
(0.160)
-
1.450***
(0.166)
Western Europe
North America
0.674***
(0.132)
0.411**
(0.149)
0.211
(0.202)
South Asia
-
0.445**
(0.189)
-
2.152***
(0.242)
-
1.282***
(0.284)
East Asia Pacific
-
0.049
(0.132)
-
0.503**
(0.146)
-
0.525**
(0.204)
Constant
14.223***
(0.815)
16.002***
(1.189)
12.105***
(1.309)
R²
0.747
0.887
0.824
Observations 1,570 411 568
15
Table 3b. Political Terror Scale (US State Department)
ICRG
CPI
WB/KKM
ICRG
0.083***
(0.017)
CPI
0.023
(0.047)
WB/KKM
0.142
(0.102)
Polity 4 (net)
0.012**
(0.004)
-
0.013**
(0.006)
0.005
(0.007)
Ln GDP per capita
0.197***
(0.041)
0.298***
(0.075)
0.027
0.079
Ln Population -0.369***
(0.037)
-0.385***
(0.064)
-0.494***
0.068
Ln Trade 0.117***
(0.032)
-0.099*
(0.051)
0.010
(0.061)
Ln Government
Consumption
0.066*
(0.035)
0.004
(0.062)
0.163**
(0.066)
Africa
-
0.117**
(0.058)
-
0.334***
(0.080)
-
0.266**
(0.079)
Latin America
-
0.663***
(0.051)
-
0.551***
(0.064)
-
0.367***
(0.077)
Middle East
-
0.664***
(0.055)
-
0.890***
(0.078)
-
0.530***
(0.082)
Western Europe
North America
0.173**
(0.061)
0.224**
(0.076)
0.306**
(0.103)
South Asia
0.040
(0.087)
-
0.146
(0.119)
-
0.073
(0.142)
East Asia Pacific
0.125**
(0.061)
0.099
(0.072)
0.153
(0.102)
Constant
6.101**
(0.380)
8.174***
(0.607)
7.908***
(0.640)
R²
0.760
0.887
0.828
Observations
1,543
385
556
Note: *p<0.1, **p<0.05, ***p<0.001. Standard deviations are given between parentheses.
All models have Prais-Winsten autoregressive controls.
16
Table 3c. Political Terror Scale (Amnesty International)
ICRG
CPI
WB/KKM
ICRG
0.053**
(0.021)
CPI
-
0.038
(0.067)
WB/KKM
0.063
(0.120)
Polity 4 (net)
0.016***
(0.004)
0.014*
(0.008)
0.024**
(0.008)
Ln GDP per capita
0.256***
(0.049)
0.701***
(0.098)
0.349***
(0.091)
Ln Population
-
0.306***
(0.044)
0.003
(0.086)
-
0.137**
(0.079)
Ln Trade
0.077**
(0.038)
-
0.244***
(0.068)
-
0.010
(0.070)
Ln Government
Consumption
0.002
(0.042)
-
0.393***
(0.083)
-
0.185**
(0.076)
Africa
0.116
(0.071)
-
0.295**
(0.110)
-
0.349***
(0.094)
Latin America
-
0.676***
0.062
-
0.871***
(0.087)
-
0.674***
(0.090)
Middle East
-
0.721***
(0.066)
-
1.009***
(0.100)
-
0.754***
(0.094)
Wes
tern Europe
North America
0.306***
(0.077)
0.416***
(0.106)
0.322**
(0.124)
South Asia
0.187*
(0.101)
-
0.272*
(0.153)
-
0.093
(0.157)
East Asia Pacific
0.184**
(0.073)
0.095
(0.095)
-
0.090
(0.116)
Constant 6.085***
(0.446)
8.141***
(0.819)
7.302***
(0.746)
R²
0.721
0.830
0.792
Observations
1,339
307
490
Note: *p<0.1, **p<0.05, ***p<0.001. Standard deviations are given between parentheses.
All models have Prais-Winsten autoregressive controls.
17
Table 3d. Human Rights Factor
ICRG
CPI
WB/KKM
ICRG
0.04
4**
(0.013)
CPI
-
0.023
(0.037)
WB/KKM
0.136*
(0.075)
Polity 4 (net)
0.016***
(0.003)
0.006
(0.005)
0.014**
(0.005)
Ln GDP per capita
0.189***
(0.032)
0.418***
(0.056)
0.142**
(0.057)
Ln Population
-
0.362***
(0.028)
-
0.234***
(0.049)
-
0.298***
(0.049)
Ln Trade
0.046*
(0.025)
-
0.204***
(0.039)
0.046
(0.045)
Ln Government
Consumption
0.039
(0.027)
-
0.160**
(0.047)
0.001
(0.047)
Africa
-
0.016
(0.046)
-
0.372***
(0.061)
-
0.240***
(0.058)
Latin America
-
0.623***
(0.040)
-
0.716***
(0.049)
-
0.434***
(0.056)
Middle East
-
0.564***
(0.043)
-
0.879***
(0.056)
-
0.602***
(0.059)
Western Europe
North America
0.271***
(0.050)
0.355***
(0.060)
0.291***
(0.077)
South Asia
0.115*
(0.066)
-
0.308***
(0.086)
-
0.209**
(0.098)
East Asia Pacific
0.169***
(0.047)
0
.114**
(0.054)
-
0.019
(0.072)
Constant 3.530***
(0.290)
5.496***
(0.461)
3.801***
(0.471)
R²
0.803
0.922
0.891
Observations
1,289
299
474
Note: *p<0.1, **p<0.05, ***p<0.001. Standard deviations are given between parentheses.
All models have Prais-Winsten autoregressive controls.
18
34.
The final step in our analysis is to examine the cross-national patterns in the relationship
between corruption and human rights. For this analysis, we use the ICRG measure of
corruption, which was the most consistent and significant, and the human rights factor
score, which has a normal distribution across the world. Figure 4 is a scatter plot
between corruption on the horizontal axis and the human rights factor score on the
vertical axis. We use the year 2003. The relationships revealed through the bivariate and
multivariate analysis suggest that there should be a positive relationship between high
corruption scores (i.e. low corruption) and high human rights scores (i.e. good records
of protection). Countries that fall on the line confirm the general hypothesis, while the
‘outliers’ provide a good insight into those countries in which corruption is high and
rights protection is good, or those countries in which corruption is low and rights
protection is not particularly good. While battling corruption in general is good for
improving the human rights situation in a country, it is these contradictory and ‘deviant’
cases that require additional attention.
Canada
Cuba
Haiti
Dominican Republic
Jamaica
Trinidad and Tobago
Mexico
Guatemala
Honduras
El SalvadorNicaragua
Colombia
Venezuela, RB
Guyana
EcuadorPeru
Brazil
Bolivia
Paraguay
Chile
Argentina
Uruguay
United Kingdom
Ireland
Belgium
France
Switzerland
Spain
Portugal Germany
Poland
Austria
Hungary
Slovak Republic
Italy
Albania
Croatia
Slovenia
Greece
Bulgaria
Romania
Russian Federation
EstoniaLatvia Lithuania
Ukraine
Belarus
Armenia
Azerbaijan
FinlandSweden
Guinea-Bissau
Gambia, The
Senegal
Niger
Cote d'Ivoire
Guinea
Burkina Faso
Liberia
Sierra Leone
Ghana
Togo
Cameroon
Nigeria
Congo, Rep.
Congo, Dem. Rep.
Uganda
Kenya
Tanzania
Somalia
Ethiopia
Angola
Mozambique
Zambia
Zimbabwe
Malawi
Namibia
Madagascar
Morocco
Algeria
Tunisia
Libya
Sudan
Iran, Islamic Rep.
Turkey
Egypt, Arab Rep.
Syrian Arab Republic
Lebanon
Jordan
Israel
Saudi Arabia
Yemen, Rep.
Kuwait
Bahrain
Qatar
United Arab Emirates
Kazakhstan
China
Mongolia
Korea, Dem. Rep.
Korea, Rep.
Japan
India
Pakistan
Bangladesh
Myanmar
Sri LankaThailandVietnam
Malaysia
Singapore
Philippines
Indonesia
Australia
Papua New Guinea
-2 -1 0 1 2
Human Rights Factor Score
0 2 4 6
Corruption (ICRG)
Figure 4. Corruption and human rights, 2003
35.
Figure 4 shows that there are many countries of interest for this particular year. On the
one hand, there are those countries that have reasonably good scores on corruption and
yet have weaker records for human rights, such as Brazil, Kenya, Cote D’Ivoire, Israel,
and Colombia (although Colombia has experienced severe internal conflict). On the
other hand, there are those countries that have relatively bad scores on corruption and
yet have stronger records for human rights, such as Niger, Paraguay, Armenia,
Kazakhstan, and Mozambique. The number and position of countries that fall off the
line explain the relatively weak relationship overall between corruption and human rights
and confirm that factors other than corruption account for the variation in human rights
that we observe. Moreover, while the scatter plot is illuminating, it is but a snapshot of
the world and even if the relationship is shown for other years (see Figure 5), the
19
correlation reveals little as to why at different moments of comparison countries have a
mixed record on corruption and human rights.
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
Corruption
HR
0 2 4 6
-2
0
2
1984 1985 1986 1987 1988
1989 1990 1991 1992 1993
1994 1995 1996 1997 1998
1999 2000 2001 2002 2003
Graphs by Year
Figure 5. Corruption and human rights, 1984-2003
I
MPLICATIONS FOR
A
NTI
-C
ORRUPTION
C
AMPAIGNS
36.
This paper has presented measures of corruption and human rights, and then specified
and tested a model of the relationship between the two using bivariate and multivariate
analysis. The bivariate analysis showed a consistent and positive relationship between
the three measures of corruption on the one hand and between the measures of
corruption and human rights on the other. The positive and significant coefficients lend
initial support to the hypothesis that higher levels of corruption are related to worse
records of human rights protection. Scatter plot analysis also revealed that certain outlier
cases warrant additional attention, since there are several countries that have good
records on corruption and bad records on human rights, and vice versa. But in any one
year, the list of such outlier countries may well differ and further analysis may want to
include internal conflict as an additional explanatory variable (see, e.g. Poe and Tate
1994).
37.
The multivariate analysis, however, weakened the support for this hypothesis in some
degree. First, not all the measures of corruption were significantly related to human
rights protection. Transparency International’s CPI is not significant across the different
models. In fairness, Transparency International has argued that the nature of the coding
precludes time-series analysis even though other studies such as this have used it in this
way. Second, variables other than corruption were also significantly related to human
rights protection, suggesting that any policy prescriptions should include these factors
alongside corruption as targets for a reform agenda.
20
38.
In particular, the results of our analysis suggest that promoting democracy and
economic development alongside a reduction in corruption seems a sensible package of
activities to improve the overall human rights situation. The extant literature on human
rights that utilises this method of comparison has confirmed the importance of
economic development and democracy for human rights as has been demonstrated in
the analysis presented here (see Landman 2005a for a review). Moreover, the work on
democratization both at a global level (e.g. Zanger 2000b) and a regional level in Latin
America (e.g. Foweraker and Landman 1997) and Africa (e.g. Lindberg 2006) that initial
democratization has tangible benefits for the protection of human rights. But beyond
efforts at initial democratization, strengthening the mechanisms for horizontal
accountability such as the independence of judiciaries, oversight authority and capacity
of legislative assemblies and greater transparency of decision making will contribute to
an improvement in human rights protection since perpetrators can no longer hide
behind dysfunctional institutions. Long term cultural change, however, is much harder
to instil as sedimented practices over time have become institutionalised and reified to
such a degree that corruption becomes an acceptable form of ‘doing politics’ and human
rights violations continue with impunity.
21
R
EFERENCES
Beck, N. and Katz, J. N. (1995) 'What to Do (And Not to Do) with Time-Series Cross-
Section Data', American Political Science Review, 89 (3): 634-47.
Claude, R. P. (1976) 'The Classical Model of Human Rights Development' in R. P. Claude
(ed) Comparative Human Rights, Baltimore and London: Johns Hopkins University Press.
Davenport. C. (1999) 'Human Rights and the Democratic Proposition', Journal of Conflict
Resolution, 43 (1): 92-116.
Davenport, C. and Armstrong, D. A. (2004) ‘Democracy and the Violation of Human Rights:
A Statistical Exploration from 1976 to 1996’, American Journal of Political Science, 48 (3): 538-
554.
De Mesquita, B. B., Cherif, F. M., Downs, G. W., and Smith, A. (2005) ‘Thinking Inside the
Box: A Closer Look at Democracy and Human Rights’, International Studies Quarterly, 49 (3):
439-480.
Diamond, L, Linz, J. J., Lipset, S. M. (1989) Comparing Experiences with Democracy: Democracy in
Developing Countries, Boulder: Lynne Reiner.
Foweraker J. and Krznaric, R. (2001) 'Measuring Liberal Democratic Performance: A
Conceptual and Empirical Critique' Political Studies, 45 (3): 759-787.
Foweraker, J. and Landman, T. (1997) Citizenship Rights and Social Movements: A Comparative and
Statistical Analysis, Oxford: Oxford University Press.
Hafner Burton E. M. and Tsutsui K. (2005), ‘Human Rights in a Globalizing World: The
Paradox of Empty Promises,’ American Journal of Sociology, 110 (2005): 1373-1411
Hathaway, O. (2002) 'Do Treaties Make a Difference? Human Rights Treaties and the
Problem of Compliance', Yale Law Journal, 111: 1932-2042.
Henderson, C. (1991) ‘Conditions Affecting the Use of Political Repression’, Journal of Conflict
Resolution, 35 (1): 120-142.
Henderson, C. (1993) ‘Population Pressures and Political Repression’, Social Science Quarterly,
74: 322-33.
Jaggers, K. and Gurr, T. R. (1995) 'Tracking Democracy's Third Wave with the Polity III
Data', Journal of Peace Research, 32 (4): 469-482.
Keith, L. C. (1999) 'The United Nations International Covenant on Civil and Political Rights:
Does it Make a Difference in Human Rights Behaviour?' Journal of Peace Research, 36 (1): 95-
118.
Landman, T. (2004) ‘Measuring Human Rights: Principle, Practice, and Policy’, Human Rights
Quarterly, 26 (November): 906-931.
Landman, T. (2005a) ‘Review Article: The Political Science of Human Rights,’ British Journal
of Political Science, 35 (July): 549-572.
22
Landman, T. (2005b) Protecting Human Rights: A Comparative Study, Washington DC:
Georgetown University Press.
Landman, T. (2006) Studying Human Rights, Oxford and New York: Routledge.
Landman, T. and Häusermann, J. (2003) Map-Making and Analysis of the Main International
Initiative Measuring Democracy, Good Governance, and Human Rights, report prepared for
Eurostat.
Lindberg, S. (2006) Democracy and Elections in Africa, Baltimore: Johns Hopkins University
Press.
McCamant, J. F. (1981) ‘Social Science and Human Rights’, International Organization, 35 (3):
531-552.
Meyer, W. H. (1996) 'Human Rights and MNCs: Theory vs. Quantitative Evidence', Human
Rights Quarterly, 18 (2): 368-397.
Meyer, W. (1998), Human Rights and International Political Economy in Third World Nations:
Multinational Corporations, Foreign Aid, and Repression, Westport, CT: Praeger.
Meyer, W. H. (1999a) 'Confirming, Infirming, and Falsifying Theories of Human Rights:
Reflections on Smith, Bolyard, and Ippolito Through the Lens of Lakatos', Human Rights
Quarterly, 21 (1): 220-228.
Meyer, W. H. (1999b) 'Human Rights and International Political Economy in Third World
Nations: Multinational Corporations, Foreign Aid, and Repression,' Human Rights Quarterly,
21 (3): 824-830.
Mitchell N. J. and McCormick, J. M. (1988) 'Economic and Political Explanations of Human
Rights Violations, World Politics, 40: 476-498.
Moore, W. (2006) ‘Synthesis v. Purity and Large-N Studies: How Might we Assess the Gap
between Promise and Performance?’ Human Rights and Human Welfare, 6: 89-97.
Munck, G. and Verkuilen, J. (2002), ‘Conceptualizing and Measuring Democracy: Evaluating
Alternative Ideas’, Comparative Political Studies, 35 (1): 5-34.
Neumayer, E. (2005) ‘Do International Human Rights Treaties Improve Human Rights?’
Journal of Conflict Resolution, 49: 925-953.
Plümper, T. and Troeger, V. (2007) Efficient Estimation of Time Invariant and Rarely
Changing Variables in Panel Data Analysis with Unit Effects, Political Analysis (forthcoming).
Poe, S. and Tate, C.N (1994) 'Repression of Human Rights to Personal Integrity in the 1980s:
A Global Analysis, American Political Science Review, 88: 853-872.
Poe, S. C., Tate, C. N., and Keith, L. C. (1999) 'Repression of the Human Right to Personal
Integrity Revisited: A Global Cross-National Study Covering the Years 1976-1993',
International Studies Quarterly, 43: 291-313.
Smith, J, Bolyard, M. and Ippolito, A. (1999) ‘Human Rights and the Global Economy: A
Response to Meyer,’ Human Rights Quarterly, 21: 207-219.
23
United Nations Development Programme (2006) Indicators for Human Rights Based Approaches
to Development in UNDP Programming: A Users’ Guide, New York and Oslo: UNDP.
Zanger, S. C. (2000a) ‘Good Governance and European Aid: The Impact of Political
Conditionality’, European Union Politics, 1 (3): 293-317.
Zanger, S. C. (2000b) 'A Global Analysis of the Effect of Regime Changes on Life Integrity
Violations, 1977-1993', Journal of Peace Research, 33.