3 short critical essays

Sage Publications, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to Journal of Peace Research.

http://www.jstor.org

State Capacity, State Failure, and Human Rights Author(s): NEIL A. ENGLEHART Source: Journal of Peace Research, Vol. 46, No. 2 (march 2009), pp. 163-180 Published by: Sage Publications, Ltd. Stable URL: http://www.jstor.org/stable/25654378 Accessed: 24-08-2014 13:04 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

journal of ? The Author(s), 2009. Reprints and permissions: httpJIwww.sagepub. co. uk/journalsPermissions. nav, vol. 46, no. 2, 2009, pp. 163-180

Sage Publications (Los Angeles, London, New Delhi,

RESEARCH Singapore andWashington DC) http://jpr.sagepub.com DOI10.1177/0022343308100713

State Capacity, State Failure, and Human Rights*

NEIL A. ENGLEHART

Department of Political Science, Bowling Green State University

While it is universally recognized that states are responsible for human rights conditions in their

jurisdictions, it is less often noticed that this responsibility has two dimensions, one normative and one

empirical. Normatively, most people agree that states ought

to prevent human rights abuses. Empiri

cally, however, states may not always be able to do so. In weak and failing states, agency loss and the

inability to police effectively can lead to abuses by private individuals and rogue agents of the

state.

Thus, on balance, weak states typically have worse human rights records than strong ones. This is dem

onstrated by a global time-series cross-section analysis showing that indicators of state weakness ? low

tax revenues, corruption, and lack of law and order ? all have a negative impact on human rights to per

sonal security. The effect differs for different kinds of rights. Extrajudicial killings are highly sensitive to

state capacity, while political imprisonment is more sensitive to democracy. Overall, however, it appears

that the totalitarian model of human rights abuse by excessively strong states applies to a restricted set

of cases. The more common problem is states that cannot effectively protect human rights. We must

take state failure seriously when thinking about the causes of - and remedies for - human rights abuse.

Introduction

There is an implicit tension in the relation

ship between states and rights: states are

simultaneously a threat to human rights and their principal protector (Donnelly, 2003:

35-37). It is now universally accepted that states are responsible for human rights con

ditions within their borders, but this respon

sibility has two dimensions that may conflict with one another. Normatively, it is widely

agreed that states ought to ensure that the

rights of their citizens are respected by both

abstaining from abuse and preventing pri vate parties from committing abuses. Yet,

* I would like to thank Emilie Hafner-Burton, Kisangani

Emizet, Melissa Miller, Steven Poe, James Ron, and three

anonymous reviewers for their helpful comments on this article. The analysis was conducted using Stata 9. The data used in this article and results of alternative speci fications of the model can be found at http://www.prio. no/jpr/datasets. Correspondence should be directed to

[email protected].

empirically, it is possible that states may be

unable to discharge this normative respon

sibility. Many states are weak, plagued with

corruption, and unable to effectively police their territories. Even if well-intentioned,

weak states may not be able to prevent

abuses by powerful private actors. States with corrupt, poorly paid police, judges, and civil servants may be unable to control their own agents.

These possibilities are of real interest, since

many countries with poor human rights records are weak and failing states.1 If the

1 I follow Rotberg's (2004) distinction between weak and

failing states. Weak states are those that have difficulty delivering services to citizens, owing to lack of resources,

corruption, poor infrastructure, and so on. This includes the most basic service: security. Failing states are a subca

tegory of weak states, in which there is armed opposition to

the government. Rotberg also distinguishes failing states, where opposition is active, from failed states, where oppo sition movements have destroyed the central government but have been unable to reconstitute a new order.

163

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

164 journal of PEACE RESEARCH volume 461 number 21 march 2009

weakness of the state itself leads to human

rights abuses, this has important implica tions for where the blame should be lodged and how problems can best be addressed. For

instance, shaming governments with poor human rights records has been one of the

primary ways NGOs have sought leverage on

human rights issues. Shaming can work only if a state has the capacity to remedy human

rights abuses and lacks the will. Willingness to create better conditions is a moot point if states cannot

police their territories or con

trol their agents. I argue that the inability to police terri

tory and to control agents of the state has a

significant negative impact on human rights conditions. Weak states create conditions

ripe for human rights abuse because they cannot restrain powerful citizens and lose control over their own employees. I test the

argument using time-series cross-section

data, accompanied by robustness tests. State

capacity turns out to be a robust predictor of

human rights measures, although its impact varies depending on the particular rights in

question. This has important implications for

determining responsibility for human rights abuses and addressing them effectively.

Agency Loss, Poor Policing, and Human Rights Violations

The presumption that states are responsible for rights is founded on the legal theory of

sovereignty. Theoretically, sovereign states

have legal supremacy over their territories: no greater authority exists. It has long been

recognized that this legal supremacy is Active in many places. States are not always able to control their own territories or otherwise

realize the authority with which they are the

oretically endowed (Jackson, 1990; Krasner, 1999; Badie, 2000). As Nettl (1968) argued almost 40 years ago, empirically the state must be treated as a variable rather than the constant supposed by legal theory.

One useful way to think about this difference between the state in theory and in

practice is through principal-agent theory. Although governments as principals have the

authority to give instructions to their agents, agents may or may not comply. If agents fail to accomplish what their principals intend, this is described as agency loss, a central

problem in principal-agent theory (Ross, 1973; Moe, 1984; Arrow, 1985). Principals

may be unable to control the behavior of

agents because of information asymmetries or because of conflicting incentives. In the former case, the agent often knows more

than the principal and may have incentives to conceal information displeasing to the

principal. In the latter case, the agent may want to do things that the principal does not

want, or may have incentives that do not

motivate the desired behavior. Mitchell has insightfully applied prin

cipal-agent theory to human rights abuse. He assumes that 'human rights violations are a policy'. As with any policy, agents of the state must be given incentives to carry it out (Mitchell, 2004: 17-18). Mitchell

recognizes the problem of agency loss, but his major examples of this phenomenon are instances in which principals surrender control intentionally in order to reward followers or inflict cruelties while denying culpability. For instance, principals may

exploit agency loss by creating 'artificial information asymmetries', choosing not to

know what agents are doing in the expec tation that they will likely commit abuses

(Mitchell, 2004: 48). However, it would be perverse to suggest

that principals lose control over their agents

only by choice. We can easily extend the

principal-agent logic introduced by Mitchell to the context of weak states, where agency loss is rampant. Indeed, Mitchell gives a

few examples of such unintended agency loss, for instance Indian prison guards who

abuse prisoners so that they will be bribed

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart Human rights 165

to stop (Mitchell, 2004: 46). This is not a

policy of the Indian government, but rather

something the guards do on their own initia

tive, exploiting their public power for private gain.

Such circumstances are sadly common:

poorly paid state employees are weakly incentivized by their official salaries to fol low the rules and often face little oversight.

They may choose to use their power to com

mit abuses, either for profit or out of innate

cruelty, and there are few disincentives to

deter them.

State officials may also be bribed by private parties to commit abuses or ignore them. For instance, police might be bribed

by landlords to beat tenants, by business

people to intimidate rivals, by criminals to extort resources, or by politicians

to coerce

voters. Alternatively, private

actors may

bribe or intimidate officials to ignore abuses committed by their own agents. If the state

is unable to police effectively, powerful citizens may commit abuses with impunity, knowing they will not be sanctioned. In

many weak states, locally powerful figures effectively usurp state authority altogether (Migdal, 1988; O'Donnell, 1993). These

petty despots are seldom accountable to

any higher authority, in part because they control the local representatives of those authorities. They are free to use violence and intimidation to maintain their power,

resulting in human rights abuses. The human rights literature increasingly

recognizes private actors as

potential abu

sers, including corporations, international financial institutions, criminals, insurgents, and even families (Paust, 1992; Thomas &

Beasley, 1993; Weissbrodt, 2002; Brysk, 2005). Similarly, the growing literature on

the 'dark side' of civil society suggests that

private actors can seriously threaten human

rights if not policed by the state (Walzer, 1991; Kumar, 1993; Keane, 1996; Berman, 1997, 2003; Colletta & Cullen, 2000).

State Failure

Agency loss and poor policing occur to some degree in all countries but are most

widespread in weak and failing states. In these 'lame leviathans' (Callaghy, 1987), resources available to pay salaries and moni tor employees are limited, encouraging cor

ruption. Many suffer from the legacies of weak colonial regimes: corruption, ethnic

tensions, and uneven development (Badie, 2000). They also face significant infrastruc ture problems: lack of good roads, rail lines,

telephone connections, and so on. 'Rough terrain' and lack of roads have been recog nized as potential factors in civil war, but

they create barriers to governance even in

the absence of violent opposition (Herbst, 2000). Indeed, the very weakness of the state may encourage the formation of non

state armed groups that sharply increase the likelihood of civil war (Fairbanks, 2002; Fearon & Laitin, 2003).

Colombia provides a useful illustration of these dynamics. It has an exceptionally poor human rights record, falling into the worst

category of the Political Terror Scale (PTS) for ten years running and earning the con

demnation of all major global human rights organizations.

Although insurgent groups and drug gangs contribute to poor human rights con

ditions in Colombia, right-wing paramilita ries are the country's

most notorious

abusers. Outlawed by the central govern ment in 1989, they continue to operate, sometimes with active support from military units. Thus, Human Rights Watch (2001: 6) recommended that the Colombian gov ernment should:

Effectively combat paramilitaries and per

manently dismantle paramilitary organiza tions by capturing leaders and prosecuting and punishing those responsible for forming, organizing, leading, belonging to, assisting, and financing paramilitary groups, including security forces.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

166 journal of PEACE RESEARCH volume 461 number 21 march 2009

These recommendations appear reasonable and straightforward: the Colombian gov ernment controls the military, the military controls the paramilitaries, and so the gov ernment should be able to end paramilitary violence. As the ultimate principal embody

ing Colombian sovereign authority, the

government ought to be able to control its

agents.

There is, however, considerable agency loss in these relationships, making it unclear

who controls whom. In Human Rights Watch's own report, it is evident that para militaries manipulate military officers and

government officials through bribery. In a

country with a 2001 per capita income of

US$7,040 a year, military officers received bribes ranging from US$1,500 to 3,000 a

month (Human Rights Watch, 2001: 23). Civilian government officials and demo

cratically elected politicians were bribed too.

Under these conditions, officials are more

strongly incentivized by the paramilitaries than the government:

The established powers are governed by alter

native powers that, in the territories under

their control, are the supreme authorities, control decision making, and inspire the

respect and obedience of the local population. The Colombian state has neither the power nor the authority to settle social conflicts

or

enforce the law. It is obliged to

negotiate its

own sovereignty with the other, alternative

powers. (Melquizo & Crenshaw, 2001: 112)

Colombian judges, too, have noted that in

practice sovereignty in Colombia is frag mented into 'small independent states that

defy the law', each under the control of a

different armed group (Kline, 1999: 66). The standard legal model of sovereign state

responsibility for human rights is clearly

inadequate to understand this situation.

In this context, the Human Rights Watch

recommendations to 'effectively combat the

paramilitaries', punish their leaders and sup

porters, and 'ensure that suspects, including

government members, are prosecuted in civilian courts' ring hollow (Human Rights

Watch, 2001: 6). Even while detailing the

ways in which Colombia fails, Human Rights Watch clings to a presumption that the state

is responsible both normatively and empiri cally for controlling human rights abuse and

simply lacks the will to do so.

The specific dynamics of weak states

vary. In some cases, such as Afghanistan and Somalia, the state may collapse utterly, so that no public authority exists to provide security and enforce rights. In others, such as Colombia and India, a partial failure of

authority may expose some citizens to oppres sion by others, or by rogue agents of the state.

The common thread is that in weak states it is impossible to make state employees fully accountable and adequately enforce laws to

create better human rights conditions.

A Global View

The dynamics found in Colombia can be found in any number of vignettes, but do these examples really reflect global patterns?

Does state capacity on balance lead to better human rights conditions or worse?

The hypothesis tested here is that state

capacity has a positive impact on human

rights conditions. Stronger states perform better on measures of human rights protec tion because they are more effective at polic

ing abusive private actors and controlling the

actions of their own agents. This produces more restraint in the use of violence by both

public and private actors, which on balance will lead to fewer abuses.

The plausibility of this hypothesis is

suggested by the fact that in 2002, the most recent year for which PTS data are

available, the worst countries in the world were Colombia, Burundi, the Democratic

Republic of Congo, and Sudan. None can

be characterized as strong states. In order to test the hypothesis that weak states

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart Human rights 167

lead to poor human rights conditions, 1 conduct an ordered logit analysis of time

series cross-section data covering over

140 countries.

Measures of Human Rights Abuse

The dependent variable - human rights

abuse -

is measured using two data sources:

the PTS and Cingranelli & Richards's

CIRI project. The PTS data have been

widely used and therefore have the advan

tage of providing a baseline for comparison with earlier studies. Gibney's most recently

updated PTS scores based on the State

Department annual human rights reports are used here, because they

are more com

plete than those based on Amnesty Inter

national annual reports (see Gibney &

Dalton, 1996). The scale is inverted, so

that higher scores reflect better human

rights conditions. Mitchell & McCormick (1997) argue

that human rights violations should be bro

ken down into subtypes for analysis. If so,

the PTS may actually muddy the analysis by

combining abuses with different causes in

the same measure.

The CIRI data separately code disappea rances, killings, political prisoners, torture, freedom of movement, religious freedom,

rights to association, political participation and free speech, workers' rights, and

women's economic, social, and political

rights. This permits a test of Mitchell &

McCormick's argument with time-series

cross-section data. Owing to limits of

space, I focus on just two kinds of rights to

personal security: extrajudicial killing and

political imprisonment.2

2 An important limitation on the CIRI data is their severely truncated scale. Countries are coded into only three catego ries, with 0 indicating no abuses, 1 indicating some abuse, and 2 indicating serious abuse. In the case of extrajudicial killing, for instance, a 2 indicates 50 or more killings. Once countries have over 50 killings, no distinctions are made

among them.

Measures of State Capacity State capacity is a complex concept. I adopt a relatively restricted use of the term: it refers

to the willingness and capability of the state

apparatus to carry out government policy.

One could think of it as the inverse of agency loss: it reflects the degree to which the prin

cipal (i.e. the government) controls its agents and the degree to which the government

can

police private citizens.

State capacity here does not refer to the

stability, longevity, or popularity of particu lar governments. It is important to distin

guish between the state apparatus and the

government. In studies of the politics of

developed countries, this distinction is basic: no one would confuse the Labour govern ment with the professional civil service and

military forces of the United Kingdom. The same distinction is less often made in studies

of developing countries, but I would argue that it is just as important.

In some cases (e.g. Burma or the Congo), a strong government can persist for a long time with - or because of - a very weak state. A government may keep the state

weak intentionally, for instance to main

tain informal clientage networks (Chabal &

Daloz, 1999). This leaves the government free to indulge in patronage, corruption, and

repression without fear of being checked by other public authorities, such as an indepen dent judiciary, honest police, or effective

budgetary institutions. In extreme cases, the

government may attack the state apparatus

to prevent it becoming a site of resistance

(Englehart, 2007). Under such circum

stances, the power of petty local despots is

likely to grow, increasing the level of human

rights abuse. There is no generally accepted way of

measuring state capacity. The concept itself

is multidimensional, and no single sum

mary measure is satisfactory. Three different measures are used here; each taps a different

aspect of state capacity.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

168 journal of PEACE RESEARCH volume 461 number 21 march 2009

The first measure is the Political Risk Service's (PRS) Law and Order measure. This variable uses surveys of expert opin ion to measure the strength and impar tiality of the legal system, and the degree to which the law is generally respected and order is maintained (PRS, n.d.). Coun tries with poor law enforcement score low on this variable, while those in which the law is generally enforced score high. Since the particular form of agency loss discussed above leads to a breakdown of the state's

capacity to provide justice and create order, this variable is valuable because it provides a measure of the degree to which the state

is capable of enforcing the law within its borders. In countries with poor Law and Order scores, we would expect to find that

powerful private interests are freer to coerce

citizens, and that police and military forces

would be less effective, less subject to con

trol by the government, and more prone to

'capture' by private actors.

Measures of corruption may provide another way of measuring state capacity: states with high capacity should have less cor

ruption than those with less capacity. Here I use the Transparency International Corrup tion Perceptions Index (CPI) as a measure of

corruption. Like the PRS data, the CPI are

based on surveys of expert opinion (Trans

parency International, 2007). A third measure of state capacity employed

here is tax as a proportion of gross domestic

product (Fauvelle-Aymar, 1999; Lieberman,

2002). Extracting taxes has traditionally been regarded as one of the most important functions of the state. Indeed, in historical

sociology, it is generally assumed to drive

the process of state formation, which is

sometimes depicted as almost entirely about

collecting revenue to pay for a powerful military (Hintze, 1975; Tilly, 1975, 1992, 2003; Brewer, 1989). As Tilly (1992: 20) puts it, 'preparation for

war ... involves

rulers ineluctably in extraction. It builds up

an infrastructure of taxation, supply, and administration that requires maintenance of itself and often grows faster than the armies and navies that it serves'. The process of col

lecting revenue requires the development of an effective bureaucratic apparatus.

The state's ability to extract taxes is thus at the heart of its control over the use of vio

lence, or indeed its ability to engage in any

policy. In order to collect taxes effectively, states need to be able to collect information about their citizens, induce them to pay, sanction defaulters, and keep tax officials

reasonably honest. The ability to raise rev enues without taxation, for instance through the sale of natural resource endowments or

by collecting foreign aid, inhibits the deve

lopment of this strong state apparatus and

produces weaker, more corrupt, and inef fective states. States that are better able to

collect taxes therefore should tend to suffer less agency loss and be better at policing their territories. These in turn should lead to bet ter control over the use of violence by both

private and public actors and, therefore, bet ter human rights conditions.3

The tax data constitute an objective indi

cator, as opposed to the PRS and CPI vari

ables, which are both based on surveys of

expert opinion. They thus provide an impor tant check on the validity of those indicators.

They also have the advantage of covering more country-years than either the PRS or

the CPI. The tax data used here are drawn

3 Using tax as a proportion of GDP as a measure of state

capacity does assume that states ordinarily collect as much revenue as they can, given the institutional constraints they face. This may seem a strong assumption, but the data

suggest it is probably accurate. For instance, after Ronald

Reagan's tax reform in 1986, tax collections in the United States actually went up, both in absolute terms and as a

proportion of GDP. Bueno de Mesquita et al. (2003) argue that parliamentary governments have a higher propensity to tax than presidential systems, but overlook the possibi lity that this may be because, on balance, they have higher capacity. Parliamentary systems do have higher tax collec tions as a proportion of GDP than presidential systems, but they also score better on all other measures of state

capacity, including corruption and law and order.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart HUMAN RIGHTS 169

primarily from the World Bank's World

Development Indicators, supplemented with

data from the African Development Indica

tors, IMF Country Reports, and the Asian

Development Bank.

These three indicators of state capacity yield slightly different samples. Difference of means tests demonstrate that the countries

included when the CPI and PRS data are

employed are significantly wealthier, more

democratic, more involved in the interna

tional system, and more populous than the tax/GDP sample. They also tend to extract a larger share of tax as a proportion of GDP. In effect, the countries for which CPI and PRS data are available are a subsample of the tax/GDP group, biased toward the higher capacity

states.

Control Variables Control variables have been selected to account for important findings in the quan titative human rights literature. They include

measures of conflict (both civil and interna

tional), democracy, economic development, population, and engagement with the inter national human rights legal regime and the international community generally. Some

variables, such as population and the lagged dependent variable, are included to control for characteristics of the dependent variables. Others are included to control for standard

explanations for human rights conditions, most importantly democracy and per capita GDP. Because they have been important predictors of human rights in other studies, it is useful to compare their impact to that of state

capacity. Civil and international wars have been

found to be significant negative predictors of human rights abuse in earlier studies (Poe & Tate, 1994; Poe, Tate & Keith, 1999). This may reflect abuses by combatants

against civilians, or the general breakdown of the social control of violence that often takes place in wartime. Abuses may include

opportunism by military units exploiting citizens without any direct military logic (Tilly, 2003: 131), or by civilians taking advantage of the chaos to settle grudges.

The Lacina & Gleditsch (2005) data on

battle-deaths in international and civil wars are used here. Battle-deaths are employed rather than a dummy for the presence of

war because the intensity of conflict is

probably relevant. Following the PRIO data-collection methodology, these data have the lowest threshold for including a war among the various datasets available

(25 deaths per year or 100 for the total con

flict). This dataset is thus more sensitive to

small conflicts.

Democracy has been found to be a sig nificant positive predictor of human rights conditions in numerous studies. The com

bined Polity IV score is used here, ranging from -10 (perfectly authoritarian) to +10

(perfectly democratic). Some studies adopt a linear interpretation

of the relationship between democracy and human rights: more is better (Poe & Tate, 1994; Poe, Tate & Keith, 1999). Others argue that there are significant non-linearities. Fein

(1995) argues that there is 'more murder in the middle': transitional and semi-democratic

regimes have more human rights abuses than fully democratic or fully authoritarian

political systems. By contrast, Davenport &

Armstrong (2004) and Bueno de Mesquita et al. (2005) suggest a tipping point, with most of the effect of democracy coming only with the establishment of full multi

party competition. Even this understates the non-linearities involved.4 Figure 1 makes the

point graphically, showing the mean values

4 A couple of recent articles suggest that a single democracy variable may be too crude. Davenport (2007) finds impor tant variations within diffetent categories of authoritarian

regimes. Richards & Gelleny (2007) find significant differ ences between the effects of presidential and legislative elec tions in democtacies. These findings together suggest that the categories of authoritarianism and democracy are multi dimensional and cannot be captured in a single variable.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

170 journal of PEACE RESEARCH volume 461 number 21 march 2009

Figure 1. Security Rights by Polity Score

?tw^i TiJliy TV, TVilirirnl Wki rtirT ii lui Inn 1ufiH iMil>l?#^Tjl ifii<i< li^H i<*i tn i nnlfwr^liw rn rti#^^mr

of the three measures of personal security

rights for each point on the Polity scale.

Only the trend for political imprisonment comes close to a linear relationship, with worse conditions in non-democracies giving way to improved conditions as the level of

democracy increases. By contrast, the Politi cal Terror Scale has a shallow u-shape, with the worst human rights conditions coming in the middle of the Polity scale. The trend for extrajudicial killings has a similar shape, but even more exaggerated: it also reaches its lowest point at a Polity score of zero, and the scores for -10 (perfecdy authoritarian) and +10 (perfectly democratic) are virtu

ally identical. The non-linearities are not

simply curvilinear, and they differ among variables, meaning that no simple expedient (such as squaring the Polity score) would be

adequate. The problem is minimized in the

analysis below because the logistic regression

employed does not assume a linear relation

ship between variables.

Level of economic development has often been found to be a strong positive predic tor of human rights conditions. Economic

development may reduce competition over resources and raise the opportunity costs of conflicts. More economic resources may also make it easier for the government to deter or buy off potential opposition. Economic

development is measured by logged GDP

per capita in constant 2000 US dollars. The natural log of population has been

included because the original State Depart ment and Amnesty International data are

events-based: they report individual examples of abuse. This makes the data sensitive to

population size, because all other things being equal, a country with a large population has more interactions between individuals and thus more 'opportunities' for abuse to occur.5

5 Hegre & Sambanis (2006) make a similar argument

about the relationship between population size and the

propensity for civil war onset.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart Human Rights 171

Human rights are defined in part by an

international regime, and thus interaction

with the international community could be a significant predictor of human rights con

ditions. To account for this, two additional variables are included in the model: explicit commitments to various human rights instruments and more general engagement

with the international community.

Engagement with the international com

munity may make states more likely to par

ticipate in the international human rights regime and may, therefore, have a positive effect on rights (Risse, Ropp & Sikkink, 1999; Landman, 2005). Engagement is

measured by involvement with intergov ernmental organizations (IGOs). The data are drawn from Pevehouse, Nordstrom &

Warnke's (2003) IGOs dataset. Pevehouse, Nordstrom & Warnke recognize that coun

tries have different levels of engagement with IGOs, distinguishing between observ

ers, associate members, and full members.

I score observers 1, associates 2 and full members 3, and sum these for each coun

try-year. Since the number of IGOs in the world has grown rapidly (Huntington, 1973; Iriye, 2002), the measure is not con

sistent over time. A score that represents

high engagement in 1960 would look small

by 2000. To adjust for this growth, I calcu late the mean IGO membership score for all countries for each year and then divide each

country's score by that mean. The result is the proportion of IGO engagement to the

global mean.

A dummy variable is included for rati fication of the first optional protocol of the International Covenant on Civil and Political Rights (ICCPR). One might expect this to predict better human rights practices, since adoption of the proto col is voluntary. However, earlier work

(Keith, 1999; Neumeyer, 2005; Hafner Burton & Tsutsui, 2007) suggests that the

treaty has little impact on human rights.

Landman (2005) finds only a weak positive relationship between treaty membership and human rights protection.6

Finally, a lagged dependent variable is

included to control for serial autocorrela tion. Previous violations of human rights are likely to lead to future abuses, as abusers often discover they can act with impunity, and victims seek revenge. Conversely, in

places with low levels of human rights abuse, these dynamics do not get started, and so

conditions remain relatively good. Including a lagged dependent variable controls for such effects and also for countries' idiosyncratic features that persist

over time.

Results

Ordered logistic regression is employed to test the hypothesis, because both dependent variables (the PTS and CIRI scores) are cat

egorical. All countries for which complete data are available are included. In addition, Huber-White ('robust') standard errors are

employed, because the data are not drawn from a random sample. The standard errors are adjusted for intragroup correlation within panels. Year dummies are included in the estimates but are not shown, owing to

limits of space. Table I shows the results of ordered logit

analyses of the three measures of personal security rights using all three measures of state capacity. The state capacity variables are significant, positive predictors of rights to personal security in 8 of the 9 models esti

mated in Table I, suggesting strong support for the hypothesis. Only in the CPI estimate for political imprisonment does state capacity fail to achieve significance. State capacity is

6 Landman's finding is based on a variable he constructed

to weight ratification by the number and significance of

treaty reservations. He generously shared these data with me, but they had little impact on the results. Because the differences are so slight, the more straightforward ratifica tion dummy is used here.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

to & tn > n w W cn tn > n ? i

Table I. Predicting Three Different Measures of Personal Security Rights

State capacity PTS Extrajudicial Political imprisonment

measured as killing

Law & Order 0.55** - - 0.49** - - 0.23**

(0.07) - - (0.07) - - (0.06) CPI - 0.23** - - 0.22** - - 0.06

(0.07) - - (0.08) - - (0.07)

Tax/GDP - - 1.80* - - 2.39** - - 1.59*

(0.77) - - (0.67) - - (0.78)

Civil war -O.OOt -0.00 -.00f -0.00** -0.00* -0.00** -0.00 -0.00 -O.OOt

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

International war -0.00 -O.OOt -0.00 -0.00 -0.00 -O.OOt 0.00 -0.00 -0.00

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

Democracy 0.05** 0.02 0.05** 0.00 -0.01 -0.01 0.10** 0.07** 0.08**

(0.01) (0.02) (0.01) (0.01) (0.02) (0.01) (0.01) (0.02) (0.01)

GDP/cap (In) 0.06 0.15 0.19** 0.11 0.20 0.26** -0.02 0.22t O.llt

(0.07) (0.11) (0.05) (0.08) (0.14) (0.04) (0.07) (0.13) (0.06)

Population (In) -0.37**

-0.27**

-0.27** -0.32** -0.17t -0.23** -0.38** -0.16 -0.32**

(0.06) (0.09) (0.05) (0.06) (0.09) (0.05) (0.07) (0.10) (0.07)

IGO 0.76** 1.15* 0.51* 0.36 0.29 0.13 0.62* 0.44 0.35

engagement (0.28) (0.48) (0.26) (0.33) (0.58) (0.22) (0.31) (0.47) (0.28)

ICCPR 1st optional -0.10 0.15 -0.21 0.01 0.27 0.02 -0.02 0.06 0.05

protocol (0.15) (0.271) (0.14) (0.16) (0.31) (0.12) (0.16) (0.24) (0.15) Lagged DV 2.34** 2.76** 2.44** 1.70** 2.01** 1.95** 1.93** 2.20** 2.00**

(0.12) (0.23) (0.10) (0.12)

(0.17)

(0.08) (0.11) (0.16) (0.11)

PseudoR2 0.50 0.55 0.46 0.35 0.42 0.34 0.41 0.45 0.39 Log pseudolikelihood -1,348.85 -5,30.51 -1,962.73 -1,195.80 -4,53.56 -1,412.62 -1,165.24 -457.04 -1,412.40

N 1,801/125 852/97 2,526/146 1,810/127 791/100 2,144/145 1,812/127 789/100 2,141/145

(observations/countries)_

Entries ate coefficients and (in patentheses) standatd errors from ordered logit models with robust standard errors adjusted for intragroup cotrelation.

t/>>.10; *p> .05;**p> M.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart HUMAN RIGHTS 173

the most robust of the variables of substantive

interest, even more so than stalwarts of ear

lier studies such as democracy and per capita GDP. However, the relative importance of state capacity varies depending on the depen dent variable. It appears to be more impor tant for security rights generally, as coded by the PTS, and for extrajudicial killing than for

political imprisonment. This suggests sup

port for Mitchell & McCormick's thesis that

different kinds of abuses will have different causes.

Below, I first examine the PTS results

and then compare the results from the CIRI

extrajudicial killing and political imprison ment variables. Interpretation of the logit coefficients is facilitated by estimating the

effect of variables by using simulations and

altering the values of particular variables of

interest, to assess their impact.7 If we set all variables to the mean, the

base probability of falling into the 'best' cat

egory of the PTS (no systematic abuse) in

the estimate employing tax as a proportion of GDP is .07. Raising the tax variable to the

maximum increases this probability to .18.

By contrast, raising democracy to the maxi mum increases that value only to .10. In the

estimation using the PRS Law and Order

variable, the base probability of falling into

the best category is .04. Raising the Law and

Order variable to the maximum increases

that probability to .09; raising democracy to the maximum increases it only to .05. In

the CPI estimate, where democracy does not even achieve significance, the impact of state

capacity is similarly large: the base probability of falling into the best category of the PTS is

.12, and increasing the CPI to its maximum raises this to .30. Thus, improving from the

lowest level of state capacity to the highest

more than doubles the chance of eliminating abuses of physical security rights.

The biggest surprise is that per capita GDP is significant for predicting the PTS

only in the models with the largest N, those

employing tax as a proportion of GDP.

The same is true for the models predict

ing extrajudicial killing. This is troubling, because GDP per capita has traditionally been regarded as an important positive

predictor of respect for human rights. One

possibility is that multicollinearity between state capacity and GDP per capita under

mines its effect. However, multicollinearity issues in Table I do not appear especially severe.8

There are, however, good reasons to con

sider whether earlier GDP per capita find

ings actually proxied state capacity. Indeed, Fearon & Laitin (2003) treat GDP per capita as an indicator of state capacity. There is an

extensive literature suggesting the impor tance of state capacity for economic develop

ment.9 This suggests that state capacity will

have a positive impact on both level of eco

nomic development and human rights condi tions. One would thus expect that including both would result in some multicollinearity, which would be endogenous to the model.

7 This analysis was conducted using Clarify (King, Tomz

& Wittenberg, 2000), available at http://gking.harvard. edu/clarify/docs/clarify.html.

8 Although there is no formal cutoff, a variance inflation

factor (VIF) of 10 is often seen as a cause for concern

about multicollinearity. A more conservative cutoff of 4 is sometimes used. In the PRS and tax models predicting the

PTS, none of the VIFs come close to this level. The high est VIF in these models is 2.38. VIFs for the CIRI models are very similar. The CPI model is more problematic, with the VIF for GDP per capita edging just above the more conservative cutoff at 4.44. This reflects the high corre

lation between GDP per capita and the CPI (.83). Since the CPI reports data on a disproportionately wealthy set of

countries, the range of variation is truncated, exacerbating multicollinearity issues. 9 The literature is too voluminous to cite entirely, but

highlights include Wade (1990), Evans (1995), Knack & Keefer (1997), World Bank (1997), Weiss (1998), and Rodrik, Subramanium & Trebbi (2002). At the same time, there is no systematic literatute pointing the causal arrow in the opposite direction, i.e. that economic

development leads to the rise of a larger and more effec tive state.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

174 journal of Peace research volume 461 number 21 march 2009

Notably, dropping the state capacity measure

substantially increases the coefficient on

GDP per capita in all models, and lowers

^-values. However, dropping GDP per capita from the model has a very modest effect on

state capacity measures. The greater stability of the state capacity variables suggests that it is GDP per capita rather than state capacity

which is the problem. The positive, significant coefficient on

GDP per capita in the tax model suggests a

separate, smaller effect of level of develop ment on human rights, however. This may be related to a reduction in the tendency to armed conflict, as suggested in earlier

studies.

As noted above, the PTS has been faulted for lumping different kinds of human rights abuses together. It is possible that different

types of abuses are mediated by different factors. The CIRI data allow us to test this

proposition. Table I includes models using CIRI data to measure two kinds of abuse:

extrajudicial killing and political imprison ment. These different types of abuse have

different causes and, as we shall see, different

relationships to state

capacity.

In the results for extrajudicial killing in

Table I, state capacity again has a significant, positive, and robust effect. The magnitude of

the impact of state capacity on extrajudicial

killing is again high. The base probability of

falling into the 'best' category (no reported

killings) is .50; raising tax as a proportion of GDP to the maximum raises this prob

ability to .70. In the Law and Order model, the base probability is .46; raising Law and

Order to the maximum raises this probabil

ity to .66.

Democracy loses significance and even

takes on a negative coefficient in these esti

mations. This is consistent with the pattern seen in Figure 1, where four of the five points

where the mean value for extrajudicial killing falls below 1 are non-negative polity scores

(0, 1, 3, and 8). Murder is not quite in the

middle: imperfect democracy appears to be more dangerous in terms of extrajudicial killing than imperfect authoritarian rule.10

Again, this resonates with Colombia, where a significant democratic breakthrough in the

early 1990s had no impact on the level of

extrajudicial executions. The results are quite different when we

turn to political imprisonment, however. State capacity is not as robust a predictor as it was for the PTS or extrajudicial kill

ing, failing to achieve significance in the CPI estimate. Democracy is the most robust

predictor among the variables of substantive interest. It takes on a positive coefficient, and its substantive impact is greater than that of state capacity. In the estimation using tax as a proportion of GDP as the measure of state capacity, the base probability of falling into the best category (no political impris onment) is .30. When democracy is raised to the maximum, the probability of falling into the best category increases to .46. By contrast, raising tax as a proportion of GDP to the maximum increases the base probabi

lity to .44, a performance near to - but not

quite as good as -

democracy. The Law and

Order estimation yields similar results: the

base probability is .29. Raising democracy to

the maximum raises this to .45, while rais

ing Law and Order to the maximum raises it to only .40. This result resonates with the

Colombia case, where the democratic break

through of the early 1990s led to a sudden

and dramatic decline in political imprison ment, albeit with significant backsliding in

the late 1990s.

Thus, there are important substantive

differences between these two types of human

rights abuse. A high-capacity state appears more important for preventing extrajudi

10 Similarly, Eck & Hultman (2007) find that rebels

tend to be more violent in democratic contexts and that

they inflict more 'one-sided' violence on civilians than

governments.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart HUMAN RIGHTS 175

cial killings, while democracy appears more

important for preventing political imprison ment. Overall, however, state capacity is

a

significant, positive, and robust predictor of

respect for human rights.

Alternative Specifications and Robustness Checks

Causal Ordering The results presented above

are contem

poraneous, but one might ask if the causal

ordering is correct: is state capacity really prior to human rights conditions? Or could

repression instead somehow lead to a decline in the capacity to tax, enforce the law, and deter corruption? Each specification of the model was tested with one- and five-year lags on the independent variables. The results are not shown, owing

to space constraints,

but largely replicate the same patterns of sig nificance, suggesting that state capacity does indeed precede human rights conditions in time.

It is beyond the scope of this article to

work out the complex temporal dynam ics. Some specifications show the strongest effects for state capacity after only a one-year

lag, while others show stronger effects after five years. However, the lagged versions of the model do not suggest that human rights conditions or practices are logically or tem

porally prior to state capacity.11

Alternative Measures: War

The Lacina & Gleditsch data on battle-deaths for civil and international war are employed above because their low threshold makes them sensitive to small conflicts. The most

commonly used war data are derived from the Correlates of War (COW) project. As a

11 Only in the models using Law and Order to pf edict

political imprisonment and extrajudicial killing is the

strongest coefficient for state capacity found in the con

temporaneous version. In all other models, the strongest effect comes after either a one- or five-year lag.

robustness check, Gleditsch's (2004) revised COW data on battle-deaths were substituted for the Lacina & Gleditsch data. This pro duced little change on the state capacity variables, although it did produce lower coefficients on both of the war variables.12

Marshall's data on major armed conflicts

(Marshall & Gurr, 2005) include an estimate of the total impact of conflicts rather than

just battle-deaths. Marshall seeks to incorpo rate factors such as civilian casualties, refu

gee flows, infrastructure damage, war-related

disease, loss of confidence in government, and so on.

Substituting Marshall's data for the Lacina & Gleditsch battle-deaths data has limited effects on the impact of state capacity but substantial effects on the war variables. In

most estimations, Marshall's data produce much stronger effects for war than the battle death data. There are thus major differences between these measures, something that

clearly merits further investigation. How ever, the differences have little impact on the effect of state capacity.

Alternative Measures: Democracy Measures of democracy

are controversial

(Munck & Verkuilen, 2002). The Polity scores are employed above, because they are so widely used. However, alternatives exist

with quite different characteristics. As a robustness check, all models were run using the Vanhanen index (2003). Vanhanen's index is constructed by multiplying voter turnout by the percentage of the vote lost by the winning party. It is therefore an objec tive measure, as opposed

to most others,

including Polity, that are constructed using

12 In the models for extrajudicial killing, both civil and international war lose significance altogether. This appears to be a function of the higher cutoff reducing the total number of conflicts recorded in the data and thus increas

ing the number of cells in which no conflict is recorded as

occurring. In effect, this reduces the amount of variance on the war variables.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

176 journal of Peace research volume 461 number 21 march 2009

complex scoring protocols. However, it

has certain problems. The turnout element creates bias (Bollen, 1993). For instance, the advanced industrial countries tend to

have lower voter turnout and thus lower Vanhanen scores. In addition, using the

percentage of the vote lost by the winning party tends to punish majoritarian systems

(Lijphart, 1999: 277-280).

Keeping these caveats in mind, using Vanhanen's index largely reproduces the results in Table I, although it tends to

strengthen the effect of democracy and weaken state capacity somewhat. The major

change is that tax as a proportion of GDP

loses significance as a predictor of PTS scores.

In all other models, however, the state capa

city variables remain robust.

Alternative Measures: State Capacity As noted above, no single measure of state

capacity is ideal for present purposes. Above, we have seen results using three of the most

promising indicators. Here, I discuss some

alternatives.

Fearon & Laitin (2003) interpret GDP

per capita as a measure of state capacity.

Observing that GDP per capita is a signifi cant (negative) predictor of civil war, they argue that this 'is due largely to its acting as

a proxy for state military and police strength relative to potential insurgents' (Fearon &

Laitin, 2003: 80). GDP per capita, however, seems inferior to tax extraction as a measure

of state capacity, because it is more distant

from the concept to be measured: GDP per

capita indicates the resources available for

extraction, while tax as a proportion of GDP

indicates the actual level of extraction.

However, if we were to adopt GDP per

capita as a measure of state capacity, the results are robust to that measure. As noted above,

eliminating other measures of state capacity from the model substantially improves the

performance of the GDP measure, so that it

becomes significant in all estimations.

One of the better sets of state capacity indicators are those developed by Kaufman and his collaborators (Kaufman, Kraay &

Zoido-Lobaton, 1999; Kaufman, Kraay &

Mastruzzi, 2003). Based on an unobserved

components analysis of a variety of indica

tors, these data are arguably superior because the combination of sources makes the analy sis more reliable. Unfortunately, they cannot

be used in a time-series analysis because they are available for only four years (1996, 1998, 2000, and 2002). If we employ cross-section models of those years, we see much the same

pattern as observed in Table I. Of the set of variables developed by

Kaufman and his collaborators, two are of interest here: rule of law and government effectiveness. The first is similar to the PRS Law and Order variable, the second evalu ates the effectiveness of the civil service in

carrying out government policy. Results can

not be included here, owing to space con

straints, but show the same pattern seen in

Table I: they are significant, positive, and

strong predictors of the Political Terror Scale

and Extrajudicial Killing, but not Political

Imprisonment.13 A number of other variables are arguably

endogenous to the concept of state capacity. These include variables such as the size of a

country, rough terrain and forest cover, and

infrastructure, such as roads and telephone lines. In each of these cases, the available data are generally supportive of the hypothesis. However, space limitations preclude a full

analysis here.

The quantitative study of state failure is

still in its infancy, and few variables have been

collected to specifically target its concerns

(Englehart, 2007). Information on non-state

13 There is some instability in these results, because the cross sections have rather low Ns. The Ns differ slightly in

different years, but are usually in the mid-140s. The main issues are that government effectiveness is not significant for extrajudicial killing in 2000 only, while it is significant for political imprisonment in 1998 only.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart human Rights 177

armed groups and the availability of weapons would be useful measures of the state's theo

retical monopoly on the use of force. The

Human Security Centre's (2005) dataset on

non-state armed conflicts is a laudable but still incomplete step in this direction. Better

data on corruption, particularly data disag gregated by different types of corruption, would be useful as well (Johnston, 2005).

Finally, data on human rights violations

disaggregated to the sub-state level would be useful. State capacity clearly varies spatially: typically states have the most control over the

capital city and its environs, but often lose control over peripheral areas. The absence of state authority does not produce a void:

petty despots fill the vacuum. In these areas, the state is weakest, and we would expect to

find worse human rights conditions. In the absence of disaggregated data, however, this can only remain a hypothesis. In Colombia, for instance, there is considerable regional variation in both state capacity and human

rights conditions, but it is impossible from the data currently available to determine if

they co-vary systematically. Buhaug & Rod (2006) show that disag

gregating the analysis from nation-states to smaller units provides new insights into the propensity for civil war. Human rights researchers would be well advised to follow their lead in developing disaggregated data.

Human rights data so far have remained

firmly attached to the state as the unit of

analysis, and this may obscure important dynamics, since it is clear that levels and

types of abuse vary within states.

Conclusions

While it is often assumed that states are the

primary abusers of rights, it is also usually recognized that they are the primary guaran tors of rights as well. WTiich of these tenden cies is more important? This is a significant question for activists as well as scholars. If

states are the primary violators of rights, then we might be best advised to look to interna

tional regimes to improve respect for human

rights; we might even want to weaken the

authority of states and undermine the prin ciple of sovereignty. If the role of the state as a guarantor of rights predominates, how

ever, then we would be better off trying to

enhance domestic enforcement capabilities, promoting the rule of law and transparency while strengthening state capacity.

The analysis above suggests that states'

protective role outweighs tendencies to

violate human rights. On balance, human

rights abuse tends to happen in relatively low-capacity states. In general, the better the state apparatus, the safer its citizens will be from the depredations of non-state actors

and rogue government officials. The human

rights situation in any country must, of

course, be examined on a case-by-case basis.

There will always be some cases of states

using their capacities to abuse their citizens.

However, on average, it is more hazardous

to one's rights to live in a weak state than a

strong one.

In the post-Cold War world, we need to

squarely face the issue of state capacity, rec

ognizing that weak states cannot secure the

rights of their citizens. This is true irrespective of the good or ill will of their rulers. Diagnos ing the pathologies of such states, however,

will require careful attention to the particu lar abuses that take place and the particular reasons for them. We cannot assume that

states are responsible for human rights abuses - we must make them so by enabling them to

police their citizens and control their agents.

References

Arrow, Kenneth, 1985. 'The Economics of

Agency', in John W. Pratt & Richard Zeckhauser, eds, Principals and Agents: The Structure of Business. Boston, MA: Harvard

University Business School (37-51).

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

178 journal of PEACE RESEARCH volume 461 number 21 march 2009

Badie, Bertrand, 2000. The Imported State: The Westernization of the Political Order. Stanford, CA: Stanford University Press.

Berman, Sheri, 1997. 'Civil Society and the Col

lapse of the Weimar Republic', World Politics

49(3): 401-429. Berman, Sheri, 2003. 'Islamism, Revolution and

Civil Society', Perspectives on Politics 1(2): 237-272.

Bollen, Kenneth, 1993. 'Liberal Democracy:

Validity and Method Factors in Cross

National Measures', American JournalofPoliti cal Science 37'(4): 1207-1230.

Brewer, John, 1989. The Sinews of Power: War,

Money, and the English State, 1688-1783. New York: Knopf.

Brysk, Alison, 2005. Human Rights and Private

Wrongs: Constructing Global Civil Society. New York: Routledge.

Bueno de Mesquita, Bruce; Alastair Smith,

Randolph M. Siverson & James D. Morrow,

2003. The Logic of Political Survival.

Cambridge, MA: MIT Press. Bueno de Mesquita, Bruce; George W. Downs,

Alastair Smith & Feryal Marie Cherif, 2005.

'Thinking Inside the Box: A Closer Look at

Democracy and Human Rights', International

Studies Quarterly 49(1): 439-457.

Buhaug, Harvard & Jan Ketil Rod, 2006. 'Local Determinants of African Civil Wars,

1970-2001', Political Geography-25(3): 315-335.

Callaghy, Thomas M., 1987. 'The State as Lame

Leviathan: The Patrimonial Administrative

State in Africa', in Zaki Ergas, ed., The African State in Transition. New York: St. Martin's

(87-116). Chabal, Patrick & Jean-Pascal Daloz, 1999.

Africa Works: Disorder as Political Instrument.

Bloomington, IN: Indiana University Press.

Colletta, Nat J. & Michelle L. Cullen, 2000. Violent Conflict and the Transformation of Social

Capital. Washington, DC: World Bank.

Davenport, Christian, 2007. 'State Repression and the Tyrannical Peace', Journal of Peace

Research 44(4): 485-504.

Davenport, Christian & David A. Armstrong II,

2004. 'Democracy and the Violation of

Human Rights: A Statistical Analysis from 1976 to 1996', American Journal of Political Science 48(3): 538-554.

Donnelly, Jack, 2003. Universal Human Rights: In Theory and Practice, 2nd edn. Ithaca, NY: Cornell University Press.

Eck, Kristine & Lisa Hultman, 2007. 'One-Sided Violence Against Civilians in War: Insights from New Fatality Data', Journal of Peace

Research 44(2): 223-246.

Englehart, Neil A., 2007. 'Governments Against States: The Logic of Self-Destructive Despo

tism', International Political Science Review

28(2): 133-153. Evans, Peter, 1995. Embedded Autonomy: States

and Industrial Transformation. Princeton, NJ: Princeton University Press.

Fairbanks, Charles H. Jr., 2002. 'Weak States

and Private Armies', in Mark Beissinger &

Crawfurd Young, eds, Beyond State Failure? Post-Colonial Africa and Post-Soviet Eurasia

in Comparative Perspective. Washington, DC:

Woodrow Wilson Center (129-160). Fauvelle-Aymar, Christine, 1999. 'Political and

Tax Capacity in Developing Countries',

Kyklos 52(3): 391-414. Fearon, James D. & David D. Laitin, 2003. 'Eth

nicity, Insurgency and Civil War', American

Political Science Review 97(1): 75-90. Fein, Helen, 1995. 'More Murder in the Middle:

Life Integrity Violations and Democracy in the World', Human Rights Quarterly 17(1): 176-191.

Gibney, Mark & Matthew Dalton, 1996. 'The Political Terror Scale', Policy Studies and

Developing Nations 4(1): 73-84 (http: //www.

nca.edu/political science/DOCS/Gibney/ Gibney and Dalton.pdf).

Gleditsch, Kristian Skrede, 2004. A Revised List of Wars Between and Within Independent States, 1816-2002', International Interactions

30(3): 231-262. Hafner-Burton & Kiyoteru Tsutsui, 2007. 'Jus

tice Lost! The Failure of International Human

Rights Law to Matter Where Needed Most',

Journal of Peace Research 44(4): 407-425.

Hegre, Havard & Nicholas Sambanis, 2006.

'Sensitivity Analysis of Empirical Results on Civil War Onset', Journal of Conflict Resolu tion 50(4): 508-535.

Herbst, Jeffrey, 2000. States and Power in Africa: Comparative Lessons in Authority and Control.

Princeton, NJ: Princeton University Press.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

Neil A. Englehart HUMAN RIGHTS 179

Hintze, Otto, 1975. The Historical Essays of Otto Hintze, Felix Gilbert, ed. New York: Oxford

University Press.

Human Rights Watch, 2001. The 'Sixth Division ;

Military-Paramilitary Ties and US Policy in Colombia. New York: Human Rights Watch.

Human Security Centre, 2005. Human Security

Report 2005. Vancouver: University of British

Columbia (http://www.humansecurityreport.

info/). Huntington, Samuel, 1973. 'Transnational

Organizations in World Polities', World Poli

tics25(3): 333-368.

Iriye, Akira, 2002. Global Community: The Role

of International Organizations in the Making of the Contemporary World. Berkeley, CA: Uni

versity of California Press.

Jackson, Robert, 1990. Quasi-States: Sovereignty, International Relations, and the Third World.

New York: Cambridge University Press.

Johnston, Michael, 2005. Syndromes of Corrup tion: Wealth, Power, and Democracy. New

York: Cambridge University Press.

Kaufmann, Daniel; Aart Kraay & Pablo Zoido

Lobaton, 1999. 'Governance Matters', World

Bank Policy Research Paper 2196. Kaufmann, Daniel; Aart Kraay & Massimo

Mastruzzi, 2003. 'Governance Matters III:

Governance Indicators for 1996?2002',

World Bank Policy Research Paper 3106. Keane, John, 1996. Reflections

on Violence. New

York: Verso.

Keith, Linda Camp, 1999. 'The United Nations International Covenant on Civil and Political

Rights: Does it Make a Difference in Human

Rights Behavior?', Journal of Peace Research

36(1): 95-118.

King, Gary; Michael Tomz & Jason Wittenberg, 2000. 'Making the Most of Statistical Analy sis: Improving Interpretation and Presenta

tion', American Journal of Political Science

44(2): 347-361. Kline, Harvey F., 1999. State Building and Con

flict Resolution in Colombia, 1986-1994. Tuscaloosa, AL: University of Alabama

Press.

Knack, Stephen & Philip Keefer, 1997. 'Does Social Capital Have an Economic Payoff? A

Cross-Country Investigation', Quarterly Jour nal ofEconomics 112(4): 1251-1288.

Krasner, Stephen, 1999. Sovereignty: Organized

Hypocrisy. Princeton, NJ: Princeton Univer

sity Press.

Kumar, Krishnan, 1993. 'Civil Society: An

Inquiry into the Usefulness of an Histori

cal Term', British Journal of Sociology 44(3): 375-395.

Lacina, Bethany & Nils Petter Gleditsch, 2005.

'Monitoring Trends in Global Combat:

A New Dataset of Battle Deaths', European

Journal of Population 2\(2): 145-166. Landman, Todd, 2005. Protecting Human Rights:

A Comparative Study. Washington, DC:

Georgetown University Press.

Lieberman, Evan S., 2002. 'Taxation Data as

Indicators of State-Society Relations: Possibi

lities and Pitfalls in Cross-National Research',

Studies in Comparative International Develop

ment36(4): 89-115.

Lijphart, Arend, 1999. Patterns of Democracy. New York: Yale University Press.

Marshall, Monty & Ted Robert Gurr, 2005. Peace and Conflict 2005. College Park, MD: Center for International Development and

Conflict Management.

Melquizo, Ramiro Ceballos & Francine

Crenshaw, 2001. 'An Analysis of Armed

Conflict in Medellin: An Analysis of the

Major Actors', Latin American Perspectives

28(1): 110-131.

Migdal, Joel S., 1988. Strong Societies and Weak States: State-Society Relations and State Capa bilities in the Third World. Princeton, NJ: Princeton University Press.

Mitchell, Neil J., 2004. Agents of Atrocity: Leaders, Followers, and the Violation of Human Rights in Civil War. New York: Palgrave Macmillan.

Mitchell, Neil J. & James M. McCormick, 1997. 'Human Rights Violations, Umbrella Con

cepts, and Empirical Analysis', World Politics

49(4): 510-525. Moe, Terry, 1984. 'The New Economics of Orga

nization', American Journal of Political Science

28(4): 739-777. Munck, Gerardo L. & Jay Verkuilen, 2002.

'Conceptualizing and Measuring Democracy:

Evaluating Alternative Indices', Comparative PoliticalStudies 35(1): 5-34.

Nettl, J. P., 1968. 'The State as a Conceptual Variable', World Politics 20(4): 559-592.

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

180 journal of PEACE RESEARCH volume 461 number 2 / march 2009

Neumeyer, Eric, 2005. 'Asylum Recognition Rates in Western Europe: Their Determi

nants, Variation, and Lack of Convergence',

Journal of Conflict Resolution 49(1): 43-66. O'Donnell, Guillermo, 1993. 'Democratization

and Some Conceptual Problems: A Latin

American View with Glances at Some Post

Communist Problems', World Development

21(8): 1355-1369. Paust, Jordan J., 1992. 'The Other Side of Right:

Private Duties Under Human Rights Law',

Harvard Human Rights Journal 5: 51-63. Pevehouse, Jon; Timothy Nordstrom & Kevin

Warnke, 2003. 'Intergovernmental Organiza

tions, 1815-2000: A New Correlates of War

Data Set' (http://www.correlatesofwar.org). Poe, Steven & Neal Tate, 1994. 'Repression of

Human Rights to Personal Integrity in the 1980s: A Global Analysis', American Political Science Review 88(4): 853-872.

Poe, Steven C; Neal Tate & Linda Camp Keith, 1999. 'Repression of Human Rights to Personal Integrity Revisited: A Global,

Cross-National Study Covering the Years

1976-1993', International Studies Quarterly

43(2): 291-313. PRS Group, n.d. ICRGMethodology (http://www.

prsgroup.com/PDFS/icrgmethodology.pdf).

Richards, David L. & Ronald D. Gelleny, 2007. 'Good Things to Those Who Wait? National Elections and Government Respect for Human Rights', Journal of Peace Research

44(4): 505-523. Risse, Thomas; Stephen C. Ropp & Kathryn

Sikkink, 1999. The Power of Human Rights: International Norms and Domestic Change. New York: Cambridge University Press.

Rodrik, Dani; Arvind Subramanium & Francesco

Trebbi, 2002. Institutions Rule: The Primacy of Institutions over Geography and Integration in Economic Development. NBER Working Paper 9305.

Ross, Stephen, 1973. 'The Economic Theory of

Agency: The Principal's Problem', American

Economic Review 63(2): 134?139.

Rotberg, Robert, 2004. 'The Failure and Collapse of Nation-States: Breakdown, Prevention, and

Repair', in Robert I. Rotberg, ed., When States

Fail: Causes and Consequences. Princeton, NJ: Princeton University Press (1-50).

Thomas, Dorothy Q. & Michele E. Beasley, 1993. 'Domestic Violence as a Human

Rights Issue', Human Rights Quarterly 15(1): 36-62.

Tilly, Charles, ed., 1975. The Formation of National States in Western Europe. Princeton,

NJ: Princeton University Press.

Tilly, Charles, 1992. Coercion, Capital, and Euro

pean States. Cambridge, MA: Blackwell.

Tilly, Charles, 2003. The Logic of Collective Violence. New York: Cambridge University Press.

Transparency International, 2007. The Methodo

logy of the Corruption Perceptions Index (http:// www.transparency.org/policy_research/

surveys_indices/cpi).

Vanhanen, Tatu, 2003. Democratization: A Com

parative Analysis of 170 Countries. New York:

Routledge. Wade, Robert, 1990. Governing the Market: Eco

nomic Theory and the Role of Government in East Asian Industrialization. Princeton, NJ: Princeton University Press.

Walzer, Michael, 1991. 'The Idea of Civil Soci

ety', Dissent 39 (Spring): 293-304. Weiss, Linda, 1998. The Myth of the Powerless

State. Ithaca, NY: Cornell University Press.

Weissbrodt, David, 2002. 'Human Rights and

Responsibilities of Individuals and Non-State

Entities', in Godmundur Alfredsson & Maria

Stavropolou, eds, Justice Pending: Indigenous

Peoples and Other Good Causes. New York:

Martinus Nijhoff (239-262). World Bank, 1997. World Development Report

1997: The State in a Changing World. New York: Oxford University Press.

NEIL A. ENGLEHART, b. 1964, PhD in Political Science (University of California, San Diego, 1996); Associate Professor of Political Science, Bowling Green State Uni

versity (2004- ); main interests: state failure

and human rights. Most recent book (edited):

Constructing Human Rights in an Age of Glo balization (M. E. Sharpe, 2003).

This content downloaded from 131.94.16.10 on Sun, 24 Aug 2014 13:04:45 UTC All use subject to JSTOR Terms and Conditions

1101462866.pdf