1

Relative Resources: Inequality in ethnic wars, revolutions, and genocides

Jan 4, 2005

Marie L. Besançon

Program on Intrastate Conflict

JohnF.KennedySchool of Government

HarvardUniversity

79 JFK Street

Cambridge, MA02138

USA

Word count 10,026

(including all notes, tables, figures, and appendices)

Abstract

Inequality in ethnic wars, revolutions and genocides

Political scientists and economists have exhaustively examined the nexus between economic inequality and political conflict (EI-PC nexus) in aggregated civil wars. The following essay revisits the nexus and its related theories, empirically and parsimoniously testing the effects of inequality on disaggregated intrastate conflicts. The results buttress the notion that traditionally deprived identity groups are more likely to engage in conflict under more economically equal conditions, while class or revolutionary wars fall under the conditions of greater economic inequality and war. Of the three types of conflicts tested—ethnic conflicts, revolutions and genocides—economic inequality seems to have the most ambiguous bearing on genocides. Support follows for recent findings that political and social equalities are of greater importance in mitigating ethnic violence and that greed factors might exacerbate violence in allcivil conflicts, including genocides.

The theoretical argument proposes that the context within which intrastate violence takes place affects the requisite level of relative resources needed for the escalation of violence between groups. The results have policy implications for ethnically divided states that are in the process of equalizing their income differential, but neglect the substantial inclusion of all groups within the political process and the distribution of public goods. The social contracts between the governors and the governed then require careful crafting for a peaceful coexistence of diverse identity groups.

1

Inequality in ethnic wars, revolutions, and genocides[*]

MARIE L. BESANÇON

Program on Intrastate Conflict, JohnF.KennedySchool of Government, HarvardUniversity

Introduction

The threat of spreading civil wars, their implications in regional stability and escalated terrorism have allied economists, political scientists, and practitioners in re-examining an old question—is inequality one of the causal factors in civil wars? Extending that question—what kind of inequality might be pertinent, and in what kind of civil wars?

Political scientists, for decades, have argued that there is a nexus between economic inequality and political violence, yet these decades of studies have empirically challenged this view. Rarely have statistical studies resulted in a robust relationship between the two variables, and the results have often been contradictory and inconclusive.[1] The inequality data itself as well as the dependent variable measuring civil wars have been in part, culpable. Furthermore, comparing different conflict processes in all intrastate wars inherently contains an equivalence problem (Moore, Linström, & O’Regan, 1996). So far, the data problems have been addressed more than the equivalence issues.

Recent studies of civil conflict have broadened to such causal variables as economic growth, geographical terrain, elite manipulation of the masses, exploitable diamond and fuel resources, renewable resources, degrees of heterogeneity among languages, religions, shape, size, and color shades of homo sapiens, while hesitating to focus too much on economic inequality. Indeed one could say that ‘the poor will always with us’, we are not quite sure what economic inequality has to do with conflict, and move on to other factors such as greed, corruption, governance, globalization, and democracy.

Because of spurious findings from these many empirical studies, scholars have suggested disaggregating political violence into particular types of civil conflict or case studies for further investigation as reasonable approaches, in that causes for all types of civil wars are not necessarily parallel.[2] In addition to the case studies that have been produced, Sambanis (2001) and Robinson (2001) have uncovered some evidence that identity wars may have different causes than revolutions—giving credence to the non-equivalence suggestions from earlier literature. It is clear that alternate specifications for the dependent variable are required to interpret the conditions under which inequality may lead to civil political conflict, and why past studies have yielded such variations in results. Furthermore, inequalities between ethnic groups rather than class groups have not been well defined cross nationally.

Theoretically, combining different approaches in the IR and comparative fields, the concept here is to expand the economic inequality-political conflict nexus and propose that within different societies and groups, members have different opportunities and resources for mobilization relative to each other, different histories of deprivation, and different kinds of deprivation, therefore are likely to have different propensities for violence and different perceptions of inequality leading to deadly conflict. Given a certain economic-inequality threshold, beyond which dissatisfied groups resort to violence, additionally, a relative minimum resource threshold is required before certain groups or subgroups will risk fighting. Furthermore, the threshold for violence in ethnic wars may not be an economic inequality grievance.

Consideration is given to the literature on other factors germane to political violence and a few of the most parsimonious are controlled for, while testing the effects of two GiniIndexes and a Human Capital Gini measure (based on education inequality), on intensity of conflict, and on the presence or absence of conflict in ethnic wars, genocides, and revolutions. These models result in distinct anomalies between the types of intrastate violence and different patterns of economic, political, and social divergences contributing to the violence—informing previous difficulties with the economic inequality—political conflict nexus in aggregated intrastate conflict studies.

A Chimera of Theories

Though the many theories on inequality and political violence propose arguments that seem contradictory—when they are applied to disaggregated civil wars—they can explain a continuum of reasoning behind economic inequality and conflict in different kinds of intrastate violence. If the conjecture that testing aggregate internal conflicts produces an equivalence problem is true, then logically the effect of income inequality might be different for different types of internal violence. While the relationship may yet exist, disaggregating the dependent variable to types of civil violence should show differences in, magnitude, significance, or direction of the effect of inequality on political violence.

The proposal here is to integrate components of relative deprivation arguments for internal conflict (De Toqueville, 1887; Davies, 1962; Gurr 1970; Nagel, 1974; Inglehart, 1990), and power arguments for systemic wars (Blainey, 1975; Organski & Kugler, 1980; Tammen et al, 2000) with institutional bargaining reasoning (Knight, 1992, 1995), augmented with thoughts on the hierarchy of needs (Popkin 1979; Malsow, 1954), risk and utility (Bueno de Mesquita, 1985; Grossman, 1991), to explain why internal factors might affect the requisite relative economic inequality levels for different kinds of civil wars to either inflate, or for conflict to occur.

Deprivation, Equality, Risk

Not meaning to belabor these much traveled theoretical paths-- deprivation and group perceptions of this phenomenon are at the forefront of one sphere of intrastate conflict theory. At the national and groups levels of analysis, the theoretical approach of Gurr (1970) argues that discontent arising from the perception of relative deprivation is the basic, instigating condition for participants in collective violence. At the extreme end of the spectrum on deprivation, Bueno de Mesquita (1985) shows that those who have nothing will risk conflict because they have nothing to lose. Davies’ conceptualization of relative resources and conflict suggests thatrevolutionswill most likely occur after a ‘prolonged period of objective economic and social development is followed by a short period of sharp reversal’ (1962: 5). Grossman’s (1991) equilibrium model for insurrections also implies that the very deprived have an expected income gain from fighting. The end results of these theories are that economic inequality produces conflict.

Knight (1995) believes that asymmetries in the ownership of resources affect the willingness of actors or groups to accept the bargaining demands of other actors. He argues that those who are the most deprived will be risk averse and will bargain with those having a greater economic base rather than demand greater equality and political representation. Based on the matrix of his institutional bargaining theory—greater resources would mean greater propensity for actors to fight for their political demands. Nagel (1974) proposes that comparison is negatively related to the degree of income disparity. If there is a large gap between groups, they will have less of a tendency to compare than if they are closer in resources. This also implies that groups closer to each other are in a position to compare, and have a greater impetus to rebel. For Popkin (1979) & Maslow (1954), those who are the most deprived are thinking foremost of survival and have nothing left for conflict. The systems theorists expand this ‘glimpse of victory’ to improve understanding of when global wars occur.

At the global systems level of analysis (Organski & Kugler, 1980; Tammen et al, 2000), power parity theory introduces the concept of equality and conflict, proposing that when nations approach equality and are discontent with the existing conditions they are most likely to fight. A nation growing in resources and population can credibly challenge a formerly dominant nation when it achieves adequate resources to do so. Blainey (1975) has a similar concept with his theory of victory in that there must be a belief in the ability to win in order to choose violent rebellion. This proposal that applies globally and regionally (Lemke, 2002) could be applied within nations and between groups. These theories are not incompatible with De Tocqueville (1887) or Denzau & North (1994). On an individual actor level, Denzau & North ask: ‘to what extent does the individual believe that his own choice can effect the real outcomes? (1994: 8).’ Extending the systems theory and the individual beliefs theory—groups will be more likely to challenge each other when there is a greater parity in resources, or some other factor that increases the group’s belief in their chances to win. Ellingsen (2000) shows this parity theory to be true with group size and rebellion between ethnic groups.

The above disparate theories have the common chord of discontent with circumstances, sometimes producing rebellion, while other times concessions. The proposals primarily address economic inequality while ignoring other such inequalities as participation in rule and public goods. Taken individually, each theory has logic and merit, however they each allow for violence to occur at very different levels of relative deprivation or equality.

In situations where the major standing grievances occur between ethnic, religious or cultural groups, with longstanding abuse and repression in many areas, social and political as well as economic, greater economic equality might be a facilitating factor for rebellion. Given a background of repression, self esteem issues, and identity group exclusion from rule and privileges, institutional bargaining theory, power parity theory, and the theory of victory, would infer that ethnic identity groups might perceive inequality in ways other than economic, but would require greater economic equality in order to perceive a higher probability of winning…therefore:[3]

H1- Increasing economic equality coincides with increasing levels of ethnic intrastate conflict.

The relative deprivation theories have traditionally maintained that violence occurs at greater economic inequality thresholds in situations where the major standing grievances are heavily economically based with large disparities in income, thus inequality is a causal factor. Given income inequality (particularly in the absence of ethnic or identity groups), relative deprivation, risk, and utility theory imply that discontent arising from increasing economic inequality is endemic to revolutions, and since groups might perceive that there is nothing to lose and everything to gain, greater economic inequality should lead to conflict.

H2- Increasing economic inequality coincides with increasing levels of violence in revolutions.

This study does not attempt to theoretically examine the complexity of reasons for a state to sponsor violence against a particular group (as defined by the State Failures Task Force), or one identity group to carry out systematic killing of another group. No hypothesis for economic inequality’s effect on genocide is inferred here by the various inequality theories on conflict. However, the effects of the explanatory inequality variables on genocide are reported, given the clear divergence from the other two types of intrastate conflict. The results imply that genocides are caused by other factors addressed in a growing body of literature on that subject (Krain, 1997; Harff, 2003; Midlarsky, 2005) and that the disaggregated data provide an avenue for further testing. The models give a clear differentiation between cause and effects on all different types of civil wars. The methods and results are as follows.

Methods[4]

The State Failures Task Force is the single publicly available source for disaggregated civil war data compiled cross nationally, thus it forms the basis for the dependent variable —number of deaths in ethnic wars, revolutions, and genocides (and the absence or presence of conflict in the disaggregated wars). The data, which only include the cases with positive results for conflict measured, were extended to include nonviolent country years as described below.

Dependent variables:

The dependent variable for political conflict is specified as deaths caused by insurgent activity (see definition in following section) in ethnic wars, in revolutions, or in genocides, and is measured on a Guttman scale. The MAGFATAL variable in the State Failures Data Set (Marshall, Gurr & Harff, 2002) is used for ethnic war deaths (Ethnic), and revolution deaths (Revolutions), and the DEATHMAG variable is employed for genocide deaths (Genocide). These data taken from the State Failures Data Set, span from 1960 to 2001. The magnitude of deaths from ethnic conflicts and revolutions is subdivided between zero and four (rounded up to whole numbers), and the magnitude of deaths for genocide is subdivided between zero and five (rounded up to whole numbers) as follows in Table 1:

______

Table I in here

______

The zero values for nonviolent country years were added with the aid of Taylor & Jodice (1983), newspapers, and other sources. This admittedly is not complete, but provided adequate data for eliminating the dependent variable being only non-zero country-years.[5]The data were also converted to binary variables (war being greater than 100 deaths for ethnic wars and revolutions and greater than 300 deaths for genocides) for secondary assays with a logit model testing for war, or no war.

The definitions of the type of civil wars disaggregated by the State Failures Project are as follows (Gurr & Harff, 1997).

Ethnic Wars: ‘Episodes of violent conflict between governments and national, ethnic, religious, or other communal minorities (ethnic challengers) in which the challengers seek major changes in their status…Rioting and warfare between rival communal groups is not coded as ethnic warfare unless it involves conflict over political power or government policy.’

Revolutions: ‘Episodes of violent conflict between governments and politically organized groups (political challengers) that seek to overthrow the central government, to replace its leaders, or to seize power in one region.’

Genocide: ‘The promotion, execution, and/or implied consent of sustained policies by governing elites or their agents—or in the case of civil war, either of the contending authorities—that result in the deaths of a substantial portion of a communal group or politicized non-communal group.’[6]

Independent variables:

The two major inequality variables are from data sets that use predicted variables based on hard data, the Gini Index and the Human Capital Gini. The third inequality variable, the World Bank Gini was tested with the logit and probit models as a control matrix for comparison. They are as follows:

Gini Index: The Gini Index for economic inequality measures derive from Feng’s (Feng, Kugler, & Zak, 2005) time series. Feng uses the Deininger and Squire (1996) data as a base and predicts the missing variables and country years with an OLS model using GDP per capita, literacy rate, RPE (relative political extraction) and regional control variables. The data span from 1960 to 1992. These data have a wider coverage than the World Bank Data, show less outlier variances and cover countries not included in the World Bank Gini.[7]

Human Capital Gini Index: Castelló & Doménech (2002) formulate a human capital inequality measure based on education attainment levels from Barro and Lee (2001) for 108 countries (every five years from 1960-1999). They claim more robust results than income inequality measures in estimating growth and investment equations. Their data are not highly correlated to the above Gini Index (.49) that derives from Deininger and Squire. The Human Capital Gini used here is the measure for ages 15 and above and the missing data are imputed with ‘ipolate’ in Stata. The index is scaled to a non-fraction (percent) to be comparable to the Gini Index and the World Bank Gini Index.