Ethnicization of Civil Wars as a Problem for an International Gendarmerie
Research proposal to Carnegie Corporation of New York
Program on “The Contending Norms of Self-Determination and the Sanctity of Existing International Borders”
James D. Fearon
David D. Laitin
Department of Political Science
Four trends in the post World War II era frame this proposal. First, there has been a clear trend of a decreasing number of inter-state and an increasing number of intra-state wars as a percentage of all wars being fought in any given year. Second, an increasing percentage of intra-state wars have been fought in the name of ethnic as opposed to class-based social groups. Of the 106 intra-state wars documented in the “State Failures” project, our analysis reveals that about two-thirds have been “ethnic” in organization or goals. Third, as illustrated by Figure 1, the increase in the percentage of intra-state wars is due to the difficulty of ending them rather than a sudden increase in their outbreak with the end of the Cold War. The data show that civil wars have broken out at a rate of about 2.3 per year in the period since the end of World War II, but they have ended at a rate of only about 1.7 per year. These wars fester and have devastating implications for human rights, political order, and future economic growth. In part because of these terrible consequences but more because of the end of the Cold War, there has been a slow but now clear trend toward intervention in these conflicts by an international gendarmerie interested in peaceful settlement rather than victory by one side or the other.
In seeking to ameliorate the humanitarian tragedies associated with this new modal form of warfare, success for these international gendarmeries has been elusive. Why is this so? Are there principles of intervention that might do better? The principal goal of our research under a Carnegie grant would be to develop criteria for intervention that would be humanitarian in purpose and efficient in operation. Equally important, such interventions must be strategically sensitive. That is to say, the principles of intervention must contain strategies of exit. Furthermore, as we shall elaborate later on, the principles of intervention should not in themselves contain incentives for people in peaceful countries to become rebels themselves.
Our present research, funded by the National Science Foundation, demonstrates that conventional interpretations of ethnicized civil wars are inadequate. Most of today’s interpretations are built on what is best called a “grievance” model. Minority groups are seen here to resent a range of humiliating policies, such as language laws that require them to learn the language of the dominant group in the society, economic discrimination, and rights to local governance. The problem with this superficially plausible approach is that once we add to our sample the very large number of minority groups who are not engaged in an ethnicized rebellion against the state, the data show no correlation between the degree of grievance and the likelihood of an ethnic rebellion. Most minority groups harbor grievances; very few are engaged in civil wars against state authority.
In fact, once subjected to statistical analysis with the Minorities at Risk (hereafter, MAR) data set that includes hundreds of minorities on all continents, the grievance model performs quite poorly. On a whole range of statistical models, controlling for country wealth, economic growth of the country from 1960-80, and the geographic position of the minority, we relied on the MAR and the State Failure data to test for factors that explain outbreaks of ethnic civil wars since 1980. In contrast to the expectations of the grievance model, we have found that:
(1) Measures of the degree of religious and cultural difference between minorities and dominant groups do not help predict whether the minority will be involved in significant violence with the state.
(2) A measure of the economic disparity between members of the minority and members of the dominant groups is only weakly related to the probability that the minority will be engaged in significant violence.
(3) Countries with significant religious minorities are no more likely to have civil wars than those without. It does not matter if we consider only minority-majority divisions along the “civilization” lines of the world religions, as suggested by the theses of Samuel Huntington, or if we include sectarian differences within world religions as well.
(4) Linguistically diverse countries are marginally, and not at all in some model specifications, more likely to have significant civil wars.
(5) Democracies or federal governments, both thought to reduce grievances, are no more or less likely to have civil wars.
(6) Discrimination and group grievances are more difficult to operationalize and measure than are aspects of cultural difference. Nonetheless, the coding for language grievances in the MAR data set are weakly but negatively related to ethnic rebellion.
More consistent with the empirical data that we have analyzed is an “insurgency” model. Rebellion is more likely in regions of poor countries in which economic growth is slow, where there is a high concentration of the ethnic population in a single region, and in which the region is in a relatively inaccessible part of the country. In this model, we picture rebels as young men who can derive far more rewards by joining an insurgency (and stealing from the state and from ordinary people) than by joining the legitimate work force. Rebels increasingly are able to procure military materiel and social support for their insurgencies to the extent that they frame their goals in ethnic terms. This is what we mean by the ethnicization of civil war. The rebels’ initial strategy is to draw in the armies of the central state who will seek to crush their rebellion. If the central state is drawn in, it is likely – given the low information central authorities have concerning their peripheral citizens and the limited training provided to soldiers in the armies of poor countries – that its army will kill innocent bystanders. The more the collateral damage, the easier it will be for rebel leaders to recruit new members, and the probability of a spiral of violence increases. This, by the way, is the time when grievances are persuasively articulated. They are taken by outsiders as the cause of the rebellion, when in fact they are a rhetoric for recruitment, often most vigorously articulated after the violence has spiraled.
Our examination of the MAR and State Failure data show that the conditions that favor insurgency much more reliably discriminate the countries and groups that have seen significant civil violence from those that have avoided such violence than the grievance model. Among the many observable implications of the insurgency model for which we have empirical support, we shall in this proposal highlight several.
(1) Poor countries have ineffective police forces, and should therefore, in an insurgency model, face more insurgencies. In support of this reasoning, our data show that per capita income in 1960 is a powerful predictor of which countries have civil wars after 1980, as is growth in per capita income from 1960 to 1980.
(2) Surveillance of people in countries that have large populations is more difficult than those that have smaller populations. Countries with large populations should therefore, in an insurgency model, more likely face civil war, and our data support this relationship.
(3) Political transitions weaken central states, and when transitions occur, insurgencies are more likely to follow. Our data give ambiguous results here, and further research is needed to evaluate this observable implication.
(4) Access to foreign financial, arms, and training support is a key resource for insurgents. Our data show a strong relationship between the active involvement of a foreign power in an ethnic group’s cause and the likelihood of an insurgency.
In sum, the cross-national and cross-group evidence support the view that ethnic civil wars are not “ethnic” in the sense of being driven by cultural differences between ethnic minorities and majorities, or grievances held by mistreated ethnic minorities (even though many minorities are indeed badly mistreated). There are simply too many culturally distinct and aggrieved ethnic minorities that are not and have not fought. Instead, “ethnic” civil wars appear to be driven more by a form of military conflict that can be harnessed to diverse political agendas, and happens at present to be harnessed in two-thirds of our cases to ethnic autonomy.
An insurgency model allows us to focus on two aspects of ethnicized rebellion missed in the grievance model, both of which have profound implications for an international gendarmerie. First, we focus on the variation in support within the ethnic group in whose name the rebellion is being fought. We thus reject the view that once an aggrieved group has an opportunity to fight for its autonomy, that there is a natural solidarity within that group in support for the rebellion. The conventional wisdom has it that rebels face a collective action problem, in getting all members of the group who have latent support for rebellion (but are afraid to be the first rebels, in the rational fear that no one else will follow) to join in. Our counterargument suggests that it is the members of the ethnic group opposed to a rebellion who face the real collective action problem. Anyone who seeks an agreement with the state that would avoid the collateral damage associated with counter insurgency is likely to be branded a traitor by the rebels, and face sanctions. Under such conditions it will be quite difficult for moderates to organize successfully.
Second, a perspective that highlights grievances ignores state actions that are most responsible for the escalation of civil wars. States give lists of minority members to extremists in the dominant ethnic group, encouraging them to burn down the homes and businesses of minorities. States bomb civilian populations in response to provocations of miniscule rebel bands. Their security apparatuses torture innocent civilians seeking to extract information from them about rebels. It is egregiously ineffective counter-insurgency rather than inappropriate policy that translates grievance into civil war. By the time an international gendarmerie enters into a civil war, in large part due to these ugly approaches by governments to put down rebellions, antagonisms can be as irreconcilable as theorists who had posited “ancient hatreds” as the cause of these wars would have predicted from the outset.
It follows from our perspective that civil wars can be avoided by more intelligent counterinsurgency coupled with economic policies to make the life of the insurgent relatively less attractive. By this, we do not mean giving states the technology of suppression that would allow them to marginalize their peripheral populations without limits. Rather, we mean the development of internationally monitored criteria for limiting states in their strategies of responding to protest and demands for autonomy and secession. We have not yet identified how a set of international norms for the suppression of insurgencies can meet the twin goals of effectiveness in dousing the insurgency and responsiveness to the rights and goals of the people in whose name the insurrection is being fought. Working out such criteria would be the first goal of our research.
It also follows from our perspective that international gendarmeries all too quickly adopt the interests of the insurgents over the interests of the people in whose name the insurgents are fighting. The gendarmerie finds itself at war with the state that is organizing the counter-insurgency and exacerbating the situation of the people who faced the humanitarian disaster that caught the conscience of the world. The questions raised here are numerous: What are the conditions that demand international military action? Under what organizational umbrella should such a military action be organized? How is it best to trade off solving the humanitarian disaster and planting the seeds for a post-war situation that solves the problem that caused the war? How can an international gendarmerie organize its withdrawal, without compromising the security of those people in whose name they entered? Answering this set of questions would be the second goal of our research.
To fulfill these two research goals will require us to develop in four ways the theoretical and empirical work we have done up till now in new directions. First, we need to continue our work to improve the MAR and State Failure data sets. Although we have spent the past year, supported by a grant from the National Science Foundation and the hospitality of the Center for the Advanced Study in the Behavioral Sciences at Stanford, in doing just that, much more work needs to be done. Once we recognized that the grievance model was inadequate and moved toward an insurgency model, systematic coding of new variables became necessary. For example, the MAR data set has very impressionistic values for group dispersion within a country. Once we recognized the importance of this variable for the insurgency model, we began (and still haven’t finished) a recoding of all groups based on their population patterns in the countries in which they live, and in neighboring countries as well. The insurgency model also requires us to know something about the terrain in the group’s home area, if in fact it has one. The harsher the terrain and the easier to hide in it (e.g. in mountains or jungles), the insurgency model would predict the more likely the group would be involved in a violent conflict with the state. Our insurgency model also demands that we know whether the home area of the group has valuable resources that are easy to carry (diamonds or drugs, for example). The more such exportable resources are available to the group, the more likely, our model predicts, young men will become insurgents. Since the MAR project was constructed with a grievance model in mind, variables consequential for an insurgency model were omitted, or coded casually. We are therefore asking the Carnegie Corporation for continued data gathering support.
Second, and now getting on to the crucial issues of public policy, we will need to think more systematically about the strategic dilemmas of international intervention. For example, if an international military force comes in automatically to protect the citizens of a region in which states have killed insurgents, this would give incentives to potential rebels in other countries to become martyrs in order to procure international support. Such an intervention regime might have the unintentional consequence of inducing new rebellions. We need to develop a simple model of external intervention that makes predictions about the likely strategies of a range of relevant actors such that we do not advocate policies with perverse incentives.
Third, we need to consolidate the literature on a set of international interventions so that we can isolate the factors that ameliorated and exacerbated the violence. We plan to look at international intervention (and its absence) in Bosnia, Kosovo, Nagorno-Karabakh, Chechnya, Sri Lanka, Sudan, Somalia, Liberia, Sierra Leone and Haiti. One of us (Laitin) has already done this in his research on the international intervention in Somalia. In a paper he wrote on that intervention, he showed (a) the incentives in the fulfillment of the short term humanitarian goal to subvert the long term goal of ending the violent conflict, (b) the strength of the U.N. in highlighting human rights abuses, but its weakness in fulfilling chapter 7 (operational) assignments, and (c) the inadvertent role played by many international human rights NGOs in exacerbating the violence. Laitin has also done extensive interviewing in Azerbaijan on the Nagorno-Karabakh insurgency, and has emphasized in a recently published paper the cultivation of a future insurgency that is now taking place in the refugee camps. From the work already done on Somalia and Nagorno-Karabakh, supplemented by additional case studies, we will go back to our theories to see if international intervention had the consequences we predicted, and also whether non-intervention had the consequences the model predicted as well.
Fourth, we plan on doing four weeks of extensive interviewing, one week at each of four institutions that have played a crucial role in the organization of international gendarmeries: New York and the U. N.; Washington D.C. and the American Department of State; Brussels and NATO; and Moscow and the Russian Ministry of Foreign Affairs. Here we plan to interview both policy-makers (at the time of the interventions) and policy-planners (who are developing criteria for intervention) in order to expand our vision of what is feasible and what has been tried in past endeavors. This will be a new element for us in our collaboration; yet we are convinced that recommendations for international intervention without an insiders knowledge of what works and what doesn’t, and what is feasible in the realm of international military/police intervention, will have little value.
For the past five years, our work on ethnic violence has been principally basic research. We have not ventured into the world of policy or prescription. We are applying for a grant from Carnegie not only because we wish to complete successfully the data sets we have been nurturing, but also because we notice an absence of systematic thinking in policy communities about military intervention in civil wars that is based on realistic assessments of the expected returns from such intervention. We are convinced that our insurgency approach is more realistic than the grievance approach toward the causes of civil war, and that there are important implications for policy once civil wars are interpreted according to our model’s mechanics. Drawing out those implications through a series of articles written for the wider policy community will be the planned product of our grant.