Resource Access in Multiethnic States: The Role of Economic Discrimination in Conflict Onset

Abstract: 

Previous research into ethnic conflict has focused on the role of political representation in advancing interethnic cooperation and incorporating minority concerns. The nature of the state influences inclusion, but ethnic conflict may relate to the state’s ability to distribute goods. I propose that when an ethnic group can influence state stability through protest, rebellion, or changes in voting behavior, access to state resources should be assured. When groups are denied economic opportunity conflict may result. I tested the role of resource access on conflict onset from 1950 to 2003 for countries identified in the Minority at Risk data set. An aggregate economic discrimination score tested against nine measures of conflict onset revealed a mixed relationship. Greater levels of economic discrimination increased the likelihood of ethnic violence onset, but decreased the likelihood of conflict onset more generally. Significantly, conflict over government incompatibilities declined in anocracies as discrimination increased. The major conclusion of the research is that the promise of democratic change may potentially deter ethnic conflict.

Table of Contents: 

    Introduction

    Ethnic conflict has long been a focus of the literature in comparative politics (Horowitz 1985; Fearon and Laitin 1996; Hechter and Okamoto 2001; Bermeo 2002). The literature considers the role of the state in accommodating ethnic groups, whether through social, political, or economic means. In particular, studies address how political systems represent ethnic interests. In turn, representation should allow groups to address social or economic grievances. The literature assumes that conflict results from political conditions rather than ethnicity directly. While political representation aids ethnic inclusion, what role does economic opportunity alone have on limiting conflict onset? I propose that access to state resources decreases the likelihood of ethnic conflict. I also address the gap in the literature between political and economic grievances among ethnic groups (Besancon 2005; Ostby 2008; Cederman, Gleditsch, and Weidmann 2011).

    I begin with an overview of the relevant literature with a specific focus on political systems and modes of representation. I then present a theory of ethnic access to resources. I draw from coalition theory to propose that when a group can influence regime stability, either through protest, rebellion, or changes in voting behavior, the state should accommodate the group. I suggest that when the state denies a group resources or economic opportunity, conflict will result. Likewise, when a group benefits from state policy, that group should avoid conflict. I then test the theory across nine conflict categories to reveal varied relations between economic discrimination and ethnic conflict onset. I conclude with a summary of my findings and suggestions for future research into the role of economic discrimination.

    Literature Review

    How can states avoid conflict? At the strategic level, conflict can emerge from common security concerns (Posen 1993). The ethnic security dilemma proposes that under a condition of anarchy, nations will enhance group security. The case of the former Yugoslavia illustrates how regional states consolidate control after the breakup of multinational federations. Security concerns prompt groups to build up defensive forces. Ethnic fears emerge when rivals mistake the defensive actions taken by groups for offensive gestures. The perception of threatening conditions causes the benefits of striking first to outweigh common security concerns, leading to conflict. In contrast, when the state guarantees security, informal institutions govern interethnic tensions (Fearon and Laitin 1996). The costs of conflict encourage ethnic groups to punish members who disrupt relations and threaten security. If a group fails to punish violators, interethnic institutions break down. To reinforce interethnic peace and in hopes of deterring future breakdowns, groups punish each other. The resulting cycle of violence can eventually escalate into more serious conflict.

    What makes a state more prone to conflict? Collective action results from internal and external pressures. External factors, such as state policy, geography, or a nearby conflict, remain out of a group’s control and tend to influence the likelihood of conflict. Internal factors provide the potential for action through characteristics such as cohesion, subordination, concentration, and size (Cohen 1997). Ethnic mobilization also depends on interethnic relations. The number of ethnic groups within the state, the relative dominance of each, and settlement patterns influence perceptions of group security (Reynal-Querol 2002a). Competition between groups advances conflict when political, social, or economic conditions reinforce group identity (Reynal-Querol 2002b). For example, differences in the socio-economic status of groups can lead to political or social exclusion. When a group feels its role does not reflect its potential influence it may act violently to secure representation (Ostby 2008). In turn, the government decides how to address grievances based on a group’s influence and its potential for mobilization.

    Relative grievances result when the structural faults of the government or policy threaten group security. While some ethnic inequalities are legacies from colonial division, others result from formal discrimination. For example, African-Americans in the United States faced legal discrimination until large-scale mobilizations during the Civil Rights era. Inequality requires the state to address the root causes of privilege and disadvantages and merely ushering in a more equitable policy may not resolve structural faults. For example, when economic equality increases, the likelihood of conflict does too. While equality reduces economic grievances, it also increases a group’s awareness of social and political inconsistencies (Besancon 2011). Policy addressing only immediate issues may fail to amend the more basic conditions contributing to inequality.

    Policy findings suggest that lower costs of insurrection promote conflict (Collier, Hoeffler, and Rohner 2006). Civil war develops when the potential gains of insurgency outweigh the costs. Lower costs of insurrection occur when resources become scarce due to poverty, slow growth rates, a weak central state, rough terrain, or a large population (Hegre and Sambanis 2006). Estimating the cost of civil war contains a political dimension too (Sambanis 2001). Given economic grievance, the more multiethnic and the less democratic a state is, the more likely it will devolve into civil war (Basedau 2011). Political systems can promote or discourage the equitable division of resources. For example, when addressing grievances over scarce resources, the transparency of democratic systems can support state claims to neutrality. Discontent over the division of goods may result in conflict when influence bears no relation to the share of resources a group recieves. In such cases, groups above and below state averages engage in action against the state (Cederman, Gleditsch, and Weidmann 2010). Because economic grievances and political access influence group security, the state benefits from ensuring ethnic parties access to resources. When the state ignores ethnic influences the resulting violence can reinforce division and undermine control (Fearon and Laitin 1996).

    One set of the literature considers the role of political systems in obtaining an equitable distribution of resources, particularly through democratic inclusion (Reynal-Querol 2002b). Arend Lijphart proposes that consociational representation is the most effective democratic model (1999). Lijphart examines two forms of democracy: majoritarian and consensus. Majoritarian democracy represents the bare majority will, while consensus democracy represents the will of the greatest number of people. According to Lijphart, consensus democracies avoid ethnic conflict because they include minority views in the political structure. By requiring proportional recruitment, coalition governance, and autonomy, the state encourages cooperation and limits competition; however, reducing political accommodation to a proportional formula can limit representation to ethnic interests (Brass 1991). Federalism presents an alternative. Rather than dictating ethnic representation, federalism encourages crosscutting affiliations. Devolving decision-making to regional or municipal governments substitutes competition between groups for competition within groups. Both cooperation and transferring decision-making restrain exclusive ethnopolitical affiliations.

    Empirically, the role of electoral systems in preventing conflict onset faces debate. In general, democratic systems limit political instability (Goldestone et al. 2010). The democratic solution faces debate, however, because democracies in transition show a tendency towards ethnic instability (Goldestone et al. 2010; Saideman et al. 2002). The transition to democracy can pronounce the scarcity of resources leading to concerns over existing or future distributions. Further, while the government type─presidential or parliamentary─seems insignificant, it appears that only fully formed democracies decrease the probability of civil war (Hegre et al. 2001). Fully formed democracies escape scarcity because, by definition, they possess institutional capacity, which to a degree reflects economic development and a greater supply of state resources. The role of the political system in managing conflict may depend on the availability of resources.

    Proportional systems alleviate tension possibly because they invest ethnic groups in the state through representation (Cohen 1997). In turn, the success of representation depends on whether policy incorporates minority views. Federal systems reduce conflict because minority groups possess more policy influence at the regional level (Bermeo 2002). Often ethnic groups settle for autonomy rather than assume full state responsibility for welfare, national security, or other services such as public health. Although decentralization improves the government’s responses to citizen needs, states avoid federalism because it provides a framework open to manipulation (Brancati 2006). Regional parties can exploit federal institutions to reinforce ethnic identity, engage in ethnic outbidding, and promote formal discrimination. While proportional and federal systems provide a forum for grievances, they may not alleviate all tensions.

    Beyond democracy, research suggests that the state’s ability to distribute resources equitably determines conflict onset. For example, ethnic minority dominance relates minimally to the onset of civil war, a finding that contradicts many power-sharing theories (Fearon, Laitin , and Kasara 2007). Additionally, minority gaps in attachment often relate to territorial or economic autonomy (Elkins and Sides 2007). The success of a political system depends, in part, on the share of resources it distributes to a region, or ethnic group. The guarantee of resources influences conflict management more than the nature of system. How states address economic grievances affects conflict onset (Aleman and Treisman 2005). For example, fiscal transfers redistribute wealth and decrease the incidence of protest, while federalism devolves economic decision-making and increases rebellion. The research suggests that the former strategy transfers wealth to amend ethnic imbalances and avoids the institutional faults of decentralization. A system’s ability to provide resources to constituent groups limits conflict and, in the absence of alternative avenues, violence provides recourse (Schock 1996). The importance of resource distribution is the focus of this study.

    Theory

    The state provides security guarantees and distributes scarce resources. The state provides two types of resources: private goods and policy. Private goods include privileges, favoritism, and other forms of patronage. Only supporters of the party in power receive private goods. Policy refers to any action aimed at public welfare, whether social security, privatization, or tax cuts. Policy can benefit groups exclusively, but affects all citizens regardless of political affiliation. Policy draws from shared public resources and consumption by one group decreases the resources available to other groups. Whether the state distributes private goods or policy depends on the size of the winning coalition.

    Government consists of a winning coalition representing the minimum number of people needed to attain and retain office (Mesquita et al. 2000). The winning coalition, in turn, comes from the selectorate, or those with the right to choose a government (Mesquita et al. 2000). Winning coalitions form because of the private goods they offer and the policies they propose to implement. The size of the winning coalition varies with the political system. For example, a democratic, two-party system contains a winning coalition equal to half the selectorate plus one vote. Alternatively, a proportional system requires parties to band together in larger, plural winning coalitions. In contrast, an authoritarian regime in all probability contains a small winning coalition because control depends on inducing loyalty. By focusing patronage on only a few members, an authoritarian coalition invests members in maintaining their privilege through state control. To retain support, regardless of coalition size, the least satisfied members of the coalition need to prefer the incumbent to any alternative candidate. In multiethnic states, satisfying coalition members requires the government to address ethnic interests.

    Because ethnic identity presents a unique medium for overcoming collective action barriers, coalitions consider ethnic interests when distributing resources. Groups demonstrate influence, in part, according to the political system in place. In systems that guarantee access, groups generally prefer to employ passive strategies that do not threaten state stability. Aggressive resistance strategies such as rebellion can lead to repression, especially if ethnic reform undermines coalition support. Whether the state accommodates ethnic interests depends on the group’s influence. By voting, lobbying, protesting, or rebelling based on ethnicity, a group increases the chance that distribution benefits them. For a winning coalition to stay in power it must convince ethnic parties that its distribution of goods benefits a group more than any other option.

    The state response to ethnic action relates to the nature of the state and the nature of the winning coalition. When access-related grievances affect regime support, the winning coalition should direct distribution toward ethnic interests. Representing ethnic interests takes the form of access to private goods or the opportunity to influence policy. If an ethnic group receives private goods or can influence policy, then it should support the government. Representation does not need to include a formal seat in government, however. Even without providing formal representation, governments can supply resources to an ethnic group. If the probability of an ethnic group accessing state resources is low, then the government should expect ethnic challenges. Alternatively, if the probability of accessing state resources is high, then the government should retain state control. Winning coalitions increase the probability of access either through policy or patronage. The nature of the coalition influences which option it selects. Larger coalitions accomodate ethnic interests through policy because private goods lose value when distributed to more members. When policy improves ethnic conditions, coalitions retain support. Small coalitions can address ethnic interests through patronage or specific ethnic policies, depending on the scarcity of private goods. Regardless, the winning coalition relies on reallocating private goods, implementing ethnically-oriented policy, or promising change to retain ethnic support.

    The review of literature thus far led to the development of Hypotheses 1 and 2 below:

    Hypothesis #1: The greater the access to state resources, the lower the likelihood of ethnic conflict.

    Hypothesis #2: The lower the access to state resources, the greater the likelihood of ethnic conflict.

    In practice, accommodation takes various forms, including a public-private divide in which the state distributes patronage to one group and directs policy at another. In Syria, minority domination over a large, politically-excluded majority should have led to ethnic conflict throughout the 1970s and 1980s. Instead, policies implemented after the Ba’thist revolution largely contained ethnic strife (Perthes 1992). Under Asad’s policies, minority elite benefited from positions in the military and bureaucracy. Minority elite relied on their positions to influence public industries and maintain a privileged status. The Ba’thist revolution also created a new Sunni elite who benefited from economic policies such as import substitution. The Asad policies managed tensions despite minority domination because the new elite benefited from and relied on state control in the public sector. By providing access without regard for ethnicity, the Asad state deepened its ties to the coalition’s elite support base (Perthes 1995; Hinnebusch 1982).

    Conflict occurred in Syria when scarcity limited the availability of resources. At the end of the 1970s, rapid growth policies began to reveal inequalities between the new elite and the general population. An Islamist movement challenged the Asad state’s sectarian rule on behalf of the disenfranchised, which included a large, rural Sunni base. The movement presented itself as the protectorate of the Sunni majority against the Alawite minority. Despite ethnic sentiments, the revolt was contained within a single region. Critically, Sunni elites remained impassive in support of the regime and their own privilege (Perthes 1995). During the revolt, Asad responded to changes in the influence of elite Sunni support by strengthening corporatist ties and introducing a new plural economy. While reform gave Sunni elite economic control, it limited that control by retaining a state stake in industry. The ongoing civil war again reveals the role of resource distribution in determining conflict onset. When the Syrian revolution began, the commercial centers retained their support for the Asad regime, while underclass Sunnis challenged the state. As sanctions limit the flow of resources into Syria, elite support has dwindled. Without current privilege, and diminishing prospects for retaining that privilege, the elite have begun to withdraw their support.

    Similarly, in Jordan an economic downturn in the late 1980s led to protest and the reorganization of the economy. A largely Palestinian minority controls the private sector in Jordan while East Bank ethnic groups compose the public sector. Both groups rely on state resources and resist changes in the division of privilege (Mufti 1999). For example, during the 1980s, Gulf “rents” faltered allowing the IMF to impose austerity measures in return for aid. Austerity required a decrease in public spending on subsidies and welfare. While austerity benefited the private sector, it threatened East Bank groups who relied on public subsidies in key industries. The economic crisis eventually demanded that the regime remove subsidies which led to protests in the East Bank. To counter the effects of austerity, the government of Jordan called parliamentary elections, which effectively returned access to state resources to East Bank clients. The layout of electoral districts favored the rural East Bank over the Palestinian minority, suggesting access-related issues were more important than political motivations. Austerity and the subsequent economic liberalization and political manipulation preserved the interests of both ethnic bodies. The expansion of parliament expanded patronage networks beyond a few broad elites to the various subethnic divides within the trans-Jordanian community (Robinson 1998). It also opened up opportunities for the business elite to access state resources in the form of exclusive contracts, regulations, and customs agreements.

    By addressing access-related grievances, the regimes in Syria and Jordan maintained a winning coalition. In both cases civil conflict demonstrated the influence of the group and led to a reallocation of resources.

    The final hypothesis is as follows:

    Hypothesis #3: The lower the access to state resources, the higher the level of -protest.

    Research Design

    A multiethnic state contains two or more ethnic groups. Ethnic groups are identified by Minorities at Risk (MAR) criteria, which includes groups with a “shared language, religion, national or racial origin, common cultural practices, or attachment to a particular territory” (Minorities at Risk Project 2009). The demographic indicators and cultural characteristics of MAR classify minority ethnic groups through a broad criterion before limiting them by political importance. Groups must provide a basis for ethnic mobilization and either “suffer or benefit” from discriminatory treatment. Political significance includes either a formal role in government, or political opposition. Discrimination depends on the state’s relative treatment of a group. If state policy affects an ethnic group in an exclusive manner, then MAR includes that group. If policy affects a group neutrally, neither to their advantage or disadvantage, MAR excludes that group.

    The MAR sample includes all groups documented between 1950 and 2003. Starting the study at the end of the Second World War allows for the evaluation of the modern nation-state while controlling for post-war migration. Limiting the sample to the beginning of the Iraq War limits the influence of statebuilding endeavors while still providing a fifty-three year observation period. The sample contains 227 groups with between one and ten groups per state. In all, 110 states are sampled. The MAR’s exclusion of advantaged majorities should not affect results because minority communities face a greater risk of lacking of access to state resources. Advantaged groups possess privileged access to resources and by definition do not face discrimination. Additionally, policy affecting the majority of people should result in violence on a massive scale rather than ethnic conflict. Further, majority groups possess more political influence because, when responding to inequality, their numerical advantage increases the effect of ethnic politics or force on state stability.

    Ethnic conflict occurs between two or more ethnic groups over social, political, or economic standing (Basedau 2011; Besancon 2005). The definition of ethnic conflict here refers to violent internal conflict. The UCDP/PRIO Intrastate Armed Conflict Onset dataset codes conflict according to the degree of armed force, a minimum of 25 battle-deaths per year, the nature of the parties in conflict (an opposition organization and the state), and incompatibilities. I measure conflict onset over one-year, two-year, and five-year peace intervals. I consider only internal armed conflicts over government incompatibilities and exclude conflicts with a territorial component. Succession claims result when longstanding political, social, or economic discrimination altogether erode a group’s confidence in state neutrality. Ethnic parties and the coalition in power can instead settle conflict over government incompatibilities within the framework of the state. Excluding territorial conflict should better represent how government responds to ethnic influence as well.

    Additional conflict onset measures include the Polity IV dataset on Major Episodes of Political Violence ( Marshall and Jaggers 2002a; Marshall and Jaggers 2002b). I evaluate measures of civil violence, civil war, ethnic violence, ethnic war, and a combined measure. The Polity IV scores measure a conflict’s impact on the social system. No conflict is coded as 0; sporadic or expressive political violence is coded as 1; limited political violence, 2; serious political violence, 3; serious warfare, 4; substantial and prolonged warfare, 5; extensive warfare, 6; pervasive warfare, 7; technological warfare, 8; total warfare, 9; and extermination or annihilation, 10. For this study, I measured onset by all events in a country-year. I evaluate both civil conflict and ethnic conflict in order to determine both exclusive ethnic incidents and incidents with significant ethnic participation. Ethnic and civil violence include codes one through three, and ethnic and civil war include codes four through ten. The Polity IV scale considers violence to include systematic, organized violence resulting in at least 500 directly related deaths, with warfare including all institutionalized conflicts.

    To measure ethnic civil conflict onset across regime type, I evaluated protest levels by the Banks Cross-National Time-Series dataset (Banks 2010; Brancati 2006; Elkins and Sides 2007; Saideman et al 2002;). The Banks dataset includes nine degrees of civil conflict recorded by the number of incidents per country-year. Of the nine categories I tested three, general strikes (Domestic2), government crises (Domestic4), and anti-government demonstrations (Domestic8). The dataset comes from New York Times reports and may underrepresent incidents. Additionally, the protest data does not differentiate between ethnic and nonethnic protest, although it should test the effect of ethnic access to resources across regime type because of sample limitations. As opposed to other measures of intrastate conflict, protest addresses actions against the state, which, rather than various communal or ethnic groups, distributes resources.

    Although access to state resources can take the form of patronage or policy, no measure exists for the degree of resources extended to ethnic groups. Instead the Minorities at Risk dataset has proxies for access to state resources (Minorities at Risk 2009). The data set includes a measure of each group’s relative economic discrimination. The index measures economic discrimination on a five-point scale of the neglect and exclusion of a minority group, and the intention of policy. No discrimination was coded as 0; historical neglect with remedial policies, 1; historical neglect and no remedial policy, 2; social exclusion and neutral policy, 3; and restrictive, exclusionary, and repressive policy, 4. Within the state, I aggregated each group’s economic discrimination score to form a single, state-level proxy of access to state resources per country-year. I expected the aggregate score to reflect economic opportunity at the state-level. The index serves as a proxy measure because, the greater the access to state resources, the greater the economic opportunity, and the greater the economic opportunity, the lower the degree of group discrimination. As opposed to group measures, a state measure reflects conditions more generally. The aggregate should control for the relative influence of groups while weighting for individual conditions and cross-group differentials.

    Polity represents the nature of the state. I measure political systems by the Polity 2 spectrum, which captures the qualities of both autocratic and democratic governments (Reynal-Querol 2002b; Hegre and Sambanis 2006). The Polity IV measure accounts for five degrees of regime type and failed states on a -10 to +10 scale. Polity IV codes state on the competitiveness and openness of executive recruitment, constraints on the chief executive, the competitiveness of political participation, and the regulation of participation. I consider autocracies to include the autocracy category (-10 to -6), anocracies to include closed and open anocracy (-5 to 5), and democracies to include both standard and fully formed democracies (6 to 10). Anocracies are political systems somewhere between autocracies and democracies where elites compete for power within the system.

    In a sensitivity analysis of the effect of common variables on civil war onset, Hegre and Sambanis (2006) find robust relationships between population, previous civil war, rough terrain, bad neighbors, low rates of economic growth, and the likelihood of civil war onset. A large population amplifies resource scarcity and can affect the likelihood of conflict onset independent of the equity of access. I model population by the World Bank standards for every country-year (World Bank 2011). Similarly, changes in growth rates alter expectations over the distribution of goods and may demand selective distribution. To measure changes in GDP growth rates, I again draw from the World Bank date on the percentage of GDP growth. A previous civil war affects political instability by lowering the cost of insurrection. I include all previous civil wars under a dummy variable based on the civil war definitions of the Polity IV political violence dataset. Rough terrain also lowers the cost of insurrection because the opposition can avoid confrontation and selectively contest the state. I use the data from Fearon and Laitin (2003), which estimates the total percentage of mountainous terrain within a country. Neighbors in conflict create a spillover effect and limit regional access to resources. Again, I control for neighbors in conflict through the Polity IV dataset, which factors the combined civil and ethnic conflict and measures of surrounding countries into a single variable.

    This study controls for the robust variables as well as ethnic fractionalization. Studies disagree on the significance of ethnic fractionalization but the measure seems significant in public goods provision, an important component of this study (Easterly and Levine 1997). Ethnic fractionalization will be measured by ethnolinguistic fractionalization (Fearon and Laitin 2003).

    I analyzed the data using a logit regression model. The model predicts outcomes for non-linear variables, allowing me to study the predicated probability of an event given the stated control variables. The independent variable, economic discrimination, was measure using a proxy measure, created by an aggregate of the MAR economic discrimination index. I measure the dependent variable, ethnic conflict, by the Polity IV political violence index, the PRIO intrastate conflict index, and the Banks dataset. Additional controls include infant mortality, changes in the GDP growth rate, mountainous terrain, previous civil wars, bad neighbors, population, ethnic fractionalization and regime type. I report statistically significant variables at or below the .05 level and other values when necessary for analysis. I check all tests for robustness and controls for correlation. If no relationship between ethnic access to state resources and the likelihood of conflict exists, the test should show no statistical significance for any of the conflict measures. Differences between conflict measures should enhance each model’s explanatory power.

    Analysis

    To test whether greater access to state resources lessens the likelihood of ethnic conflict onset, I evaluated nine models. Three of the models came from the PRIO dataset and measure conflict with one-year, two-year, and five-year periods between observations. The PRIO dataset contains two additional variables measuring the causes of conflict. These variables distinguish between territorial and government disagreements. In addition, I used four measures from the Polity IV dataset to test civil and ethnic violence and war.

    Beginning with one-year intervals between events, I found no statistical significance for the relationship between economic discrimination and conflict onset (P>|.863|). Contrary to expectations, economic discrimination reduces conflict onset, suggesting that governments may promote peace by denying groups resources. Democracy, bad neighbors, ethnic fractionalization, changes in GDP growth, and large populations do influence conflict however.

    At two-year intervals, economic discrimination remains an insignificant predictor although its effect on conflict onset remains negative. In both models, the influence of economic discrimination appears inconsequential. For one-year and two-year models democracy and autocracy seem insignificant although both reduce conflict onset. In contrast, ethnic fractionalization, bad neighbors, changes in GDP growth rates, and large populations retain their importance. These results confirm robustness checks performed by Hegre and Sambanis (2006).

    Under five-year intervals, I found that economic discrimination gains importance. Additionally, only infant mortality rate, changes in GDP growth, and a large population remained significant. All three variables affect resource scarcity and support an economic explanation of conflict onset. It appears that when variables with less direct relation to resource distribution lose significance, economic discrimination gains importance. Economic discrimination also gains significance when the time between measures increases. Longer stability periods may allow citizens to consider issues under prioritized during conflict. Alternatively, economic factors may gain importance when increasing the time between observations because governments can amend political issues over the short term.

    To test the suggested relation between economic discrimination and the duration between observations, I tested eight-year and twenty-year intervals. In accordance with previous results, the statistical significance of economic discrimination and its effect on conflict onset increases, but not beyond the reporting threshold. In totality, the first tests demonstrated a negative relation between economic discrimination and conflict onset that was nearly significant only when extending the time between observations. Refer to Table 1.

    To clarify the results, I tested the effect of economic discrimination on conflict onset over particular incompatibilities. As in the initial tests, more economic discrimination reduced conflict onset over government incompatibility. While a narrower measure than conflict onset, government incompatibilities can still emerge from social of political grievances restricting the test. The test confirmed previous results with bad neighbors, infant mortality, ethnic fractionalization, changes in GDP growth, and rough terrain affecting conflict onset. Additionally, autocracies and democracies appear to reduce the likelihood of conflict onset over differences with government. Despite these results, the composite polity score (Polity2) increases conflict onset. Because the composite score accounts for all regime types, anocracies may have a unique relation with conflict onset.

    To determine the effect of anocracy, I squared the Polity2 variable. Secondary tests show that economic discrimination retained a negative relationhip, but lost significance. In contrast, anocracy assumes importance. Anocracy decreased conflict over government incompatibility but the magnitude of its effect appeared minor. I tested territorial incompatibilities for a comparison of the relationship between anocracies, economic discrimination, and conflict incompatibilities. Economic discrimination and anocracy assume an insignificant relation with territorial conflict suggesting a particular relationship between ethnic economic discrimination in anocracies and conflict onset. Refer to Table 2.

    Further narrowing the conflict measure to ethnic and civil conflict reveals a significant and positive relationship between economic discrimination and low-level violence. In particular, ethnic violence relates to economic discrimination. The magnitude of the effect suggests that any expansion of discrimination against minority groups results in ethnic violence. In broader civil situations, violence also follows discrimination, although the effect appears less important. Of particular interest, economic discrimination did not impact either civil war or ethnic war. The effect of economic discrimination on conflict appeared limited to lower level conflicts. In order to avoid warfare, the state may attempt to amend discrimination. Refer to Table 3.

    Whether greater access to state resource reduces the likelihood of ethnic conflict remains indeterminate. Initial conflict models suggest that more discrimination instead reduces conflict onset. A unique relation between economic discrimination and conflict onset appears in anocracies. Economic discrimination may allow anocracies to consolidate control over ethnic minorities when forming or transitioning. Alternatively, citizens in anocracies may lack the consistent expectations of citizens in autocracies or democracies. In autocracies, the expectation of repression can limit resistance despite discrimination. In democracies, inclusion guarantees some political recourse for discrimination; however, in anocracies, expectations depend on the coalition in power. The winning coalition in an anocracy may distribute resources according to autocratic goals or democratic values. In such situations minorities may accept temporary discrimination in order to ensure future inclusion. Ethnic groups do not prefer discrimination, but because the direction of transition hinges on the nature of coalition, security groups may choose to avoid violent conflict in the short term. In turn, anocracies may manipulate citizen expectations to ensure survival.

    The second set of tests determines the effect of regime type on the relationship between economic discrimination and protest levels. In order to evaluate low-level action, I tested anti-government protests, general strikes, and government crises. Without distinguishing regime type, the onset of anti-government protest had no relation with economic discrimination (P>|z| = .542). Distinguishing for regime type, in democracies, economic discrimination again relates minimally to protest levels. In autocracies, the effect of economic discrimination increases only slightly. Testing suggests that democracy increases the likelihood that economic discrimination leads to anti-government protests, but that autocracies increase the substance of that effect. Protests in democracies may relate not to cultural and political issues, which are already resolved, but to economic issues. Unionization or other forms of recourse limit conflict levels and the effect of economic discrimination. In autocracies grievances may relate more to cultural or political issues but political challenges threaten regime security. The threat of repression over political protest may lead groups to prefer to protest under the guise or protection of more general economic issues. As seen in Jordan, protest results when economic discrimination threatens group privileges or security. Contrary to previous results, economic discrimination in anocracies increases protest levels, although less so than in democracies and autocracies. Again, inconsistent governance may encourage limited action, reflected here in protest levels.

    Testing general strikes, defined as industrial protests aimed at disabling the state, produces similar findings. Without controlling for regime type, economic discrimination seems insignificant. Economic discrimination does have a more substantial effect on general strikes than anti-government protest. In part, general strikes reflect economic conditions. More economic discrimination should affect, and be manifested in, economic protest. Democracies increase the likelihood and magnitude of strike onset potentially due to worker freedom and protections. Autocracies produce similar results, but to a lesser degree than in democracies, again, possibly reflecting state control. Anocracies do retain their unique relation, limiting general strikes more than democracies or autocracies. Refer to Table 4.

    There also seems to be a relation between economic discrimination and government crises. Testing indicates that greater levels of economic discrimination reduce government crises, defined here as any rapidly developing condition that threatens regime stability. Access to resources at the state-level may explain the relationship. When the state lacks resources to distribute, the government must choose how to allocate resources, yet retain state stability. Threats against the state may emerge when resource allocation reinforces existing policies of discrimination. Regimes may choose to discriminate against minorities in hopes of retaining a winning coalition, but in doing so, erode the coalition’s broader support. Significantly, regime type does not affect the relationship. Democracies decrease the effect of economic discrimination on government crises while autocracies and anocracies have a similar effect. Compared to other regimes, anocracies increase the role of economic discrimination by limiting crises, suggesting that groups will not consistently accept discrimination. Refer to Table 5.

    The findings here present inconclusive support for the relationship between economic discrimination and ethnic conflict. In general, more discrimination does not affect the likelihood of conflict onset. General findings suggest that economic discrimination may even reduce the likelihood of conflict onset. In a more specific test, greater levels of economic discrimination significantly increased the likelihood of ethnic violence onset. Additionally, economic discrimination seems to be a significant predictor of government crises. Tests of conflict over government incompatibilities support the proposition that governments may benefit from discriminating against ethnic minorities. It seems that in anocracies, the potential for future inclusion conditions actions against the state. The state may then employ economic discrimination to consolidate control or build institutional capacity, knowing that the promise of democracy deters action. Significantly, anocracies are less likely to face general strikes, anti-government protest, or governmental crises resulting from economic discrimination. Instead, democracies seem the most susceptible to civil conflict caused by economic discrimination. The expectation of equality may drive collective action in democracies, a factor potentially absent in anocracies and autocracies.

    The nature of the proxy may influence the studys conclusions. The economic discrimination variable potentially lacks the nuance necessary to evaluate changing conditions or the length of discrimination. A direct measure of an ethnic group’s access to state resources would better predict the role of access to state resources in multiethnic states. The proxy does suggest a relationship between conflict and discrimination. Future research should attempt to represent economic discrimination through alternative variables. A study of the economic discrimination variable at the group level may also clarify results.

    Conclusion

    Ethnic conflict undermines the stability of the state. Preventing ethnic conflict may require political representation, but only because political representation secures resources for a group. I have shown that an ethnic group’s access to resources can be measured by the degree of economic discrimination a group confronts. Because ethnic groups mobilize around mutual concerns, the state benefits from ensuring group access to resources, but perhaps more importantly economic opportunity. When the state fails to provide resources, conflict may result.

    A test of the effect of an aggregate economic discrimination measure on conflict onset revealed varied results. Greater levels of economic discrimination increased the likelihood of ethnic violence, and any changes in relative discrimination appear to result in ethnic action. Ethnic groups appear to prefer engaging in low-level conflict when expressing discontent in order to avoid threatening state stability. The relationship between anocracies and ethnic conflict requires future consideration. It seems that, particularly in anocracies, when a group relies on the state for inclusion, economic discrimination can limit conflict. Perhaps, if the possibility of future inclusion requires groups to remain passive then, despite economic discrimination, groups will prefer to limit actions. The measure of economic discrimination also calls for future consideration. A group-level analysis or new data on the degree of resources an ethnic group receives would clarify results. The results do provide future avenues for research and it is too early to disregard the impact of economic opportunity.

    The results suggest that regardless of the political system in place, the state’s ability to provide resources to ethnic groups limits violence. These results may explain, in part, the longevity of anocratic or authoritarian regimes in multiethnic states. Care should be taken when denying a state’s resources if a lack of resources can lead to ethnic conflict within the target state. Sanctions or other economic pressures can exacerbate tensions and undermine, not only regime stability, but also, social stability over the long-term. Stable societies should note the role of economic discrimination as well. Across all regime types, lower levels of economic discrimination decrease ethnic violence, a finding that should reflect either in political institutions or regime decisions. Regardless, economic discrimination does seem to relate to conflict onset, a finding that clarifies the role of political representation in multiethnic states.

    References

    • Aleman, Eduardo, and Daniel Treisman. 2005. “Fiscal Politics in “Ethnically-Mined” Developing, Federal Srares: Central Strategies and Secessionist Violence.” In Sustainable Peace: Power and Democracy After Civil Wars, eds. Philip Roeder and Donald Rothchild. Ithica: Cornell University Press, 173-216.
    • Banks, Arthur S. 2010. “Cross-National Time Series Data Archive, 1815-2005.” Databanks International ed. Jerusalem: Databanks International [Producer and Distributor]. http://www.databanksinternational.com/ (Accessed: 10 July, 2012).
    • Basedau, Matthias. 2011. “Managing Ethnic Conflict: The Menu of Institutional Engineeering.” GIGA Working Papers 171 (June): 4-29.
    • Bermeo, Nancy. 2002. “A New Look at Federalism: The Importance of Institutions.” Journal of Democracy 13 (April): 96-110.
    • Besancon, Marie L. 2005. “Relative Resources: Inequality in Ethnic Wars, Revolutions, and Genocides.” Journal of Peace Research 42 (July): 393-415.
    • Brancati, Dawn. 2006. “Decentralization: Fueling the Fire or Dampening the Flames of Ethnic Conflict and Secessionism?” International Organization 60 (Summer): 651-685.
    • Brass, Paul. 1991. Ethnicity and Nationalism: Theory and Comparison. New Delhi: Sage Publications.
    • Cederman, Lars-Erik, Kristian S. Gleditsch, and Nils B. Wiedmann. 2011. “Horizontal Inequalities and Ethno-Nationalist Civil War: A Global Comparison.” American Political Science Review 105 (August): 478-495.
    • Cohen, Frank. 1997. “Proportional Versus Majoritarian Ethnic Conflict Management in Democracies.” Comparative Political Studies 30 (October): 607-630.
    • Collier, Paul, Anke Hoeffler, and Dominic Rohner. 2006 “Beyond Greed and Grievance: Feasibility and Civil War.” CSAE Working Paper Series (August): 508-535.
    • Easterly, William, and Ross Levine. 1997. “Africa’s Growth Tragedy: Policies and Ethnic Divisions.” Quarterly Journal of Economics 112 (Novemeber): 1203-50.
    • Elkins, Zachary, and John Sides. 2007. “Can Institutions Build Unity in Multiethnic States?” American Political Science Review 101 (Novemeber): 693-708.
    • Fearon, James D., and David Laitin. 1996. “Explaining Interethnic Cooperation.” American Political Science Review 90 (December): 715-735.
    • Fearon, James D., and David Laitin. 2003. “Ethnicity, Insurgency, and Civil War.” American Political Science Review 97 (February): 75-90.
    • Fearon, James D., David Laitin, and Kimuli Kasara. 2007. “Ethnic Minority Rule and Civil War Onset.” American Political Science Review 101 (February): 187-193.
    • Goldestone, Jack A., Robert H. Bates, David L. Epstein, Ted Robert Gurr, Michael B. Lustick, Monty G. Marshall, Jay Ulfelder, and Mark Woodward. 2010. “A Global Model for Forecasting Political Instability.” American Journal of Political Science 54 (January): 190-208. http://www.cidcm.umd.edu/inscr/polity/index.htm (Accessed: 10 July, 2012).
    • Hechter, Michael, and Dina Okamoto. 2001. “Political Consequences of Minority Group Formation.” Annual Review of Political Science 4 (June): 189-215.
    • Hegre, Havard, Tanja Ellingsen, Scott Gates, and Nils Petter Gleditsch. 2001. “Toward a Democratic Civil Peace? Democracy, Political Change, and Civil War, 1816-1992.” The American Political Science Review 95 (March) 33-48.
    • Hegre, Havard, and Nicholas Sambanis. 2006. “Sensitivity Analysis of Empirical Results on Civil War Onset.” Journal of Conflict Resolution 50 (August): 508-535.
    • Hinnebusch, Raymond A. 1982. “Rural Politics in Ba’thist Syria: A Case Study in the Role of the Countryside in the Political Development of Arab Societies.” The Review of Politics 44 (January): 110-130.
    • Horowitz, David. 1985. Ethnic Groups in Conflict: Theories, Patterns, and Policy. Berkeley: California University Press.
    • Lijphart, Arend. 1999. Patterns of Democracy. New Haven: Yale University Press.
    • Marshall, Monty G. and Keith Jaggers. 2002a. “Polity IV Project: Political Regime Characteristics and Transitions, 1800-2010.” College Park, MD: Center for International Development and Conflict Management, University of Maryland.
    • Marshall, Monty G. and Keith Jaggers. 2002b. “Polity IV Project: Major Episodes of Political Violence, 1946-2008.” College Park, MD: Center for International Development and Conflict Management, University of Maryland.
    • Mesquita, Bruce Bueno de, James D. Morrow, Randolph M. Siverson, and Alastair Smith. 2000. “Political Institutions, Political Survival, and Policy Success.” In Governing for Prosperity, eds. Bruce Bueno de Mesquita and Hilton Root. New Haven: Yale University Press, 59-84.
    • Minorities at Risk Project. 2009. Minorities at Risk Dataset.” College Park, MD: Center for International Development and Conflict Management. http://www.cidcm.umd.edu/mar/ (Accessed: 10 July, 2012.)
    • Mufti, Malik. 1999. “Elite Bargains and the Onset of Political Liberalization in Jordan.” Comparative Politics Studies 32 (February): 100-129.
    • Ostby, Gudrun. 2008. “Horizontal Inequalities and Violent Civil Conflict.” Journal of Peace Research 45 (March): 143-162.
    • Perthes, Volker. 1992. “The Syrian Economy in the 1980’s.” Middle East Journal 46 (Winter): 37-58.
    • Perthes, Volker. 1995. The Political Economy of Syria Under Asad. New York: I.B. Tauris.
    • Posen, Barry. 1993. “The Security Dilemma and Ethnic Conflict.” Survival 35 (Spring): 465-483.
    • Reynal-Querol, Marta. 2002a. “Ethnicity, Political Systems, and Civil Wars.” Journal of Conflict Resolution 46 (February): 29-54.
    • Reynal-Querol, Marta. 2002b. “Political Systems, Stability, and Civil Wars.” Defense and Peace Economics 13 (April): 465-483.
    • Robinson, Glenn E. 1998. “Defensive Democratization in Jordan.” International Journal of Middle East Studies 30 (August): 387-410.
    • Saideman, Stephen M., David J. Lanoue, Michael Campenni, and Samuel Stanton. 2002. “Democritization, Political Institutions, and Ethnic Conflict: A Pooled Time-Series Analysis. 1985-1998.” Comparative Political Studies 35 (February): 103-129.
    • Sambanis, Nicholas. 2001. “Do Ethnic and Nonethnic Civil Wars Have the Same Causes?: A Theoretical and Empirical Inquiry (Part 1).” Journal of Conflict Resolution 45 (June): 259-282.
    • Schock, Kurt. 1996. “A Conjectural Model of Political Conflict: The Impact of Political Opportunities on the Relationship Between Economic Inequality and Violent Political Conflict.” Journal of Conflict Resolution 40 (March): 98-133.
    • World Bank. 2011. “World Development Indicators.” Washington, DC: World Bank http://data.worldbank.org/ (Accessed: 10 July 2012).

    Table 1: The Relationship Between Economic Discrimination and Conflict Onset

    Conflict Measure

    Conflict Onset:
    One Year

    Conflict Onset:
    Two Year

    Conflict Onset:
    Five Year

     

    Coef. (Std. Err.)

    Coef. (Std. Err.)

    Coef. (Std. Err.)

    Economic
    Discrimination

    -0.012
    (0.072)

    -0.007
    (0.081)

    -0.040
    (0.072)

    Ethnic
    Fractionalization

    1.50***
    (0.402)

    1.18*
    (0.424)

    0.723
    (0.460)

    Bad Neighbors

    0.053**
    (0.016)

    0.048**
    (0.018)

    0.031
    (0.021)

    Changes in GDP
    Growth Rates

    -0.050***
    (0.012)

    -0.057**
    (0.012)

    -0.067***
    (0.013)

    Autocracy

    -0.400
    (0.208)

    -0.283
    (0.226)

    -0.410
    (0.252)

    Democracy

    -0.695*
    (0.246)

    -0.510
    (0.270)

    -0.453
    (0.302)

    Population

    0.325***
    (0.063)

    0.320***
    (0.069)

    0.300***
    (0.079)

    Previous Civil War

    -0.440***
    (0.801)

    -0.279
    (0.795)

    -0.195
    (0.762)

    Significance levels reported by * .05, **.01, ***.000. Values are rounded to a three digit score. Controls include the above as well as rough terrain and infant mortality rate.

    Table 2: The Relationship Between Economic Discrimination and Conflict Onset Over Government Incompatibilities

    Conflict Measure

    Government
    Incompatibility

    Government
    Incompatibility

    Government
    Incompatibility

     

    Coef. (Std. Err.)

    Coef. (Std. Err.)

    Coef. (Std. Err.)

    Economic
    Discrimination

    -0.106*
    (0.050)

    -0.131**
    (0.046)

    -0.090
    (0.050)

    Ethnic
    Fractionalization

    1.15**
    (0.333)

    1.21***
    (0.330)

    1.10**
    (0.323)

    Bad Neighbors

    0.043**
    (0.013)

    0.043**
    (0.013)

    0.036**
    (0.014)

    Changes in GDP
    Growth Rates

    -0.044***
    (0.008)

    -0.048***
    (0.008)

    -0.041***
    (0.008)

    Rough Terrain

    0.347***
    (0.040)

    0.370***
    (0.049)

    0.372***
    (0.050)

    Polity2

     

    0.018
    (0.009)

     

    Polity22

       

    -.0021***
    (0.002)

    _cons

    -2.92***
    (0.398)

    -3.30***
    (0.374)

    -2.63***
    (0.426)

    Significance levels reported by * .05, **.01, ***.000. Values are rounded to a three digit score. Controls include the above as well as population, previous civil war, and infant mortality rate. Democracy and Autocracy controls tested during the first round.

    Table 3: The Relationship Between Economic Discrimination and Ethnic or Civil Violence

    Conflict Measure

    Ethnic Violence

    Civil Violence

     

    Coef. (Std. Err.)

    Coef. (Std. Err.)

    Economic Discrimination

    0.983***
    (0.308)

    0.541*
    (0.257)

    Significance levels reported by * .05, **.01, ***.000, Control variables not included in table. Values are rounded to a three digit score.

    Table 4: The Relationship Between Economic Discrimination and General Strikes Across Regime Type

    Conflict
    Measure

    General
    Strikes

    General
    Strikes

    General
    Strikes

    Regime
    Type

    Democracy

    Autocracy

    Anocracy

     

    Coef.
    (Std. Err.)

    Coef.
    (Std. Err.)

    Coef.
    (Std. Err.)

    Economic
    Discrimination

    0.134
    (0.110)

    0.124
    (0.121)

    0.115
    0.121

    Significance levels reported by * .05, **.01, ***.000, Control variables not included in table. Values are rounded to a three digit score.

    Table 5: The Relationship Between Economic Discrimination and Government Crises Across Regime Type

    Conflict
    Measure

    General
    Strikes

    General
    Strikes

    General
    Strikes

    Regime
    Type

    Democracy

    Autocracy

    Anocracy

     

    Coef.
    (Std. Err.)

    Coef.
    (Std. Err.)

    Coef.
    (Std. Err.)

    Economic
    Discrimination

    -0.143***
    (0.036)

    -0.140***
    (0.034)

    -0.148***
    0.036

    Significance levels reported by * .05, **.01, ***.000, Control variables not included in table. Values are rounded to a three digit score.