Defining Deadly Ethnic Riots

Lucid Madness

Why Deadly Ethnic Riots Occur

A Theory of Ethnic Conflict

Tracking Ethnic Violence

Precipitating Sparks

The Role of Ethnic Cleavages

The Role of Economic Development

Outside Instigators

Stimulating Civic Engagement

Additional Questions

Conclusions and Recommendations

About the Report

SPECIAL REPORT 101

Lethal Ethnic Riots
Lessons from India and Beyond

Briefly . . .

• Because deadly ethnic riots are activities undertaken by crowds, understanding why these riots occur and how they unfold requires analysis of the dynamics of crowd behavior.
• Rioters display a mixture of lucid calculation and irrational passion in their behavior, carefully targeting their victims but finding emotional release in their killing.
• Ethnic riots are most likely to occur when four elements are present: ethnic antagonism, an emotional response to a precipitating event, a sense on the part of the rioters and the larger social group to which they belong that killing is justifiable, and the assessment by rioters that the risk of response from police is low. Policymakers can reduce the incidence of ethnic riots by increasing the risk of response from police.
• Sparks that lead to the outbreak of violence can come from events at the national, state, or local level, and can range from rumors of the killing of a sacred cow to destruction of a holy place broadcast to millions by radio or TV. Responses to those sparks may be as mild as increased ethnic tension or as virulent as deadly ethnic riots.
• In India, lethal ethnic riots have occurred primarily in 4 of India's 28 states, and primarily in urban, not rural settings. Eight cities in India alone account for a hugely disproportionate share of death—46 percent—resulting from communal riots.
• Indian communities in which there is little interaction among members of different ethnic groups are the most likely to engage in ethnic violence. Somewhat less vulnerable to ethnic violence are communities in which members of different ethnic groups interact together in the simple routines of neighborhood life. Communities least likely to suffer from ethnic violence are those in which civic associations—ranging from film clubs and trade unions to political parties—are present and provide the basis for sustained interaction across ethnic lines.
• Governments in areas where ethnic violence is a problem tend to act in politically strategic, not legally correct, ways. In India, community members are more able to control politicians seeking to gain political advantage from ethnic conflict in locations in which strong inter-ethnic civic associations are operating.

[ Back to top ]

Defining Deadly Ethnic Riots

Donald Horowitz noted that deadly ethnic riots are sometimes confused with other types of ethnic violence. He defined ethnic riots as intense, sudden, but not necessarily wholly unplanned, lethal attacks on the civilian members of one ethnic group by civilian members of another ethnic group, with the victims chosen because of their group membership. Such riots are vicious events that involve not just killing and maiming but also mutilations and other atrocities. These riots usually produce large numbers of deaths, even more refugees and internally displaced persons, and greater ethnic homogeneity in the area as a result of the violence that has occurred.

While such riots occur fairly frequently, they do not occur everywhere. Obvious examples include recurrent Hindu-Muslim riots in India; the Sri Lanka riots of 1983 that accelerated the insurrection of the Tamil Tigers; the Kuala Lampur riots of 1969; the anti-Ibo riots in Nigeria in 1966 that eventually precipitated the Biafra war; and, more recently, the riots between Christians and Muslims in the Moluccan Islands in Indonesia—the so-called "Spice Islands."

According to Horowitz, observers during the last 50 years have advanced various theories to explain the incidence of deadly ethnic riots. During the past four decades, such theories have focused overwhelmingly on explanations having to do with grievance, relative deprivation, the actions of malevolent politicians who incite followers, or the actions of people who have something palpable and rational to gain from ethnic riots. Such theories fall short, however, because they fail to explain the intensity of violence; it is difficult to believe that such violence results from mere manipulation and calculation. The literature on ethnic riots has under-emphasized the fact that, fundamentally, riots are activities undertaken by crowds. Understanding riots requires better analysis of the way in which crowds behave and the role of passion.

Several aspects of deadly ethnic riots require a focus on the dynamics of crowd behavior:

• The scale and explosiveness of such riots, and their seemingly disproportionate character in relation to the events that tend to precipitate them.
• Their brutality and viciousness.
• Their ability to attract participants, despite the apparent benefits of "free-riding" for individuals who may hate members of another group but have the option to let others do the killing for them. Despite the option of "free-riding," there is much voluntary, authentic participation in such riots.
• The mixture of impulsive and instrumental factors that produce deadly ethnic riots. Such riots usually represent a mixture of passion and calculation on the part of those who participate in them.

[ Back to top ]

Lucid Madness

Deadly ethnic riots, according to Horowitz, are characterized not only by sadism, euphoria, and bestial slaughter but also by elements of prudence and foresight. Such riots usually include an orgy of killing that is punctuated by interludes of detached planning. Traps are laid for victims who are murdered brutally, and great precision is exercised to target individuals for violence. That is, members of Group A want to attack members of Group B but not members of Groups C or D, or other members of Group A. Accordingly, Group A conducts its investigations of prospective victims meticulously to make sure that mistakes are not made. If there is any doubt about the identity of a putative member of Group B, that individual is likely to be sent safely away. Crowds involved in deadly ethnic riots also tend to choose relatively safe locations in which to operate, and to move to other locations if the first one becomes unsafe.

The pervasive emotions that are present in these events are anxiety and hatred—emotions that cannot be fully explained by any rational analysis. Crowds participating in deadly ethnic riots tend to engage in a great deal of faulty reasoning, and in the magnification of the danger faced by the group represented by the crowd. Before the riot occurs, there are often false rumors of aggression—usually of events that have not occurred, or at least haven't occurred in the form that the rumor depicts them as having taken. Often these false rumors describe events that are exactly the sort of event that is about to be undertaken by the crowd itself. In Sri Lanka, for example, there were rumors that a Tamil army was on its way to invade the South, when no such army was on its way. A nearly identical rumor circulated in Northern Nigeria—that an Ibo army was on its way to the North—when no such army was mobilized.

Deadly ethnic riots, according to Horowitz, are characterized by a mixture of hyper-vigilance and circumspection. The rioters imagine themselves to be engaged in heroic acts of self-defense against life and death threats, such as an army that is on the march or the poisoning of the community's water supply. The rioters often over-estimate the dangers they face and misperceive the intentions and actions of their enemies.

Individuals participating in deadly ethnic riots, Horowitz argued, are indulging in angry, but pleasurable, violence. Many studies of the cathartic effect of aggression show that those who aggress feel better afterwards. Killing and degrading are the objectives, and these emotions cannot be enjoyed vicariously by "free riders." The crowd takes pleasure in over-doing violence. It often trades off the possibility of killing a larger number of persons for the more certain pleasure of killing a smaller number using the slower techniques of torture and mutilation. Peer pressure means that a little violence is likely to grow into a lot of violence. The larger the crowd is, the more likely it is to be brutal. The violence indulged in by ethnic rioters is similar to that practiced by some violent street criminals. When conventional norms are inoperative, sadists, bullies, and fighters become models for emulation and respect in ways that they are not in ordinary times.

Horowitz said that experiments show that anger can grow over time, be stored, redirected, and then released all at once. The memory of prior events can be unleashed by a current event to enhance the level of anger. Rioters connect today's provocative action by a hated ethnic group to yesterday's. They report experiencing lasting hostility and grievances from earlier incidences. In severely divided societies, there is plenty of accumulated anger, and the riot is one gateway for its release.

Because such violence is born of hatred, it always aims to degrade and destroy its victims and to produce ethnic homogenization. The slogans of rioters are revealing. A common slogan is "burn, don't loot"—that is, destroy victims, don't profit from them. "Drive out the Bengalis" was the slogan of rioters in Assam in 1960.

[ Back to top ]

Why Deadly Ethnic Riots Occur

Horowitz argued that the concatenation of four conditions best explains the occurrence of deadly ethnic riots:

• A hostile relationship between two ethnic groups, not necessarily an ancient enmity, that produces antipathy or hatred.
• A response to events—usually denominated as anger, but perhaps more accurately rendered as arousal, rage, outrage, or wrath—that strongly engages the emotions of one of these groups.
• A keenly felt sense of justification for killing.
• An assessment of the reduced risks of violence that reduces inhibition.

Ethnic Antipathy and Hatred

Horowitz noted that social scientists have studied prejudice but not in connection with riots. Antipathy—strong opposition, antagonism, or aversion directed against targets who are believed to possess certain threatening characteristics—must be present for deadly ethnic riots to occur. Horowitz argued that experiments show that anger is easier to evoke when there is prior antipathy toward the target, and that antipathy produces a more severe response than when it does not exist. A durable emotion, antipathy awaits invocation as events arise.

The deadly ethnic riot is angry violence, and that explains its magnitude, its explosive character, and its ability to attract participants. Anger always opens the possibility of action greater than its cause. Antipathy contributes to the brutality and magnitude of the violence because to see events through the lens of antipathy is to magnify them. When antipathy becomes hatred, evil intentions become magnified.

Ethnic antipathy takes anger-producing events and converts them into acts of an entire group, not just of individuals. Antipathy produces a tendency toward generalization—with the anger focused on the whole ethnic group and away from the individuation of targets—so that ethnic antipathy makes possible indiscriminate violence against members of the same ethnic group. In choosing victims, the crowd does not care whether the member of the hated ethnic group is a good person or not, just that he or she is a member of the targeted ethnic group.

Ethnic hatred consists of at least four elements: a growing and even obsessive focus on the hated group to the neglect of others; a belief that the hated group possesses fixed characteristics and is likely to behave in certain ways; a compression of intra-group differences attributed to members of the hated group; and an active sense of repulsion toward the group and its members.

Response to Events

Brutal ethnic riots, according to Horowitz, are responses to proximate events. But riots occur because the rioters are responding to fundamental character traits as the rioters see them, and to the behavior that the rioters expect of their targets. Group antipathy never permits perception of a precipitant as a one-time event; it is always interpreted as part of a larger pattern of unacceptable behavior engaged in by the targeted ethnic group. In their convoluted reasoning process, rioters show themselves to be prodigious unifiers because they assiduously link events together in a single, unbounded chain, and they don't distinguish between their own violence and the violence to which they think they are responding.

Anger, as Aristotle pointed out, derives from fear. According to Horowitz, precipitating events impart proximity to those fears. The targets of antipathy are uniformly seen to be dangerous groups. The precipitating event confirms the hostile intent of the targets and demonstrates what the rioters perceive as great cohesion among members of the target group. The rioters do not distinguish a precipitant that comes from a fraction of a target group from a precipitant that is representative of the entire target group. Riots tend not to arise out of private quarrels or personal affronts.

Justification of Violence

According to Horowitz, rioters conceive of and justify violence in three ways: as the rightful response to the enormity of danger they think they face from the target group; as punishment of wrong doers when the government has failed to take appropriate action; or as an act of war. Evidence that rioters are thinking in terms of warfare—in which killing is permissible—stems from the fact that rioters often engage in pre-riot rituals that involve traditional martial practices.

Regardless of whether killing is carried out in the name of self-defense, punishment for wrong doing, or warfare, each killing is not considered singly but as part of an extended transaction in which victims and perpetrators change places. According to Horowitz, rioters say to themselves, "They are killing us, therefore we may kill them. The riot didn't start the moment we started to kill them. It started before that."

According to the logic of the rioter, the person who kills under these circumstances is relieved of responsibility for killing by virtue of the connection established between that killing and the conduct that precedes it. Thus, the killing is not subject to moral judgment apart from the entire sequence of events. Rioters don't think they are attacking; they think they are responding, as evidenced by the fact that rumors of aggression committed by members of the targeted ethnic groups are nearly universal in events that precede deadly ethnic riots.

Risk Aversion

A major theme of deadly ethnic riots is risk reduction, which reduces inhibitions to engage in violence. An impressively wide range of conditions affects the rioters' calculations of risk, including:

• Supernatural beliefs in invulnerability produced, in some crowds, by the undertaking of prior invulnerability rituals.
• Lack of credible opposing force by the victims or the police.
• Societal condemnation of the targeted ethnic group confirmed by the inaction of the state to protect that group.
• Inadequate police deployment.
• A variety of risk-reducing, tactical decisions taken by the rioters themselves.

Elements that reduce risks taken by rioters include the following:

• Creation of an overwhelming mass of rioters. Rioters attack their victims in crowds rather than singly.
• The common use of bladed weapons against unarmed civilians.
• The leadership of local fighters skilled in fighting.
• Selective targeting of one ethnic group to reduce the possibility that multiple opponents from untargeted ethnic groups will combine against the attackers. The care with which rioters go about choosing victims is also designed to reduce the possibility that rioters will kill members of their own group because that kind of killing would reduce intra-group support for the violence.
• The decision to attack close to home so that rioters can easily retreat if they meet unexpected resistance.
• The timing of attacks. Rioters choose moments when the targets are unprotected by the police and by social and political authorities, when rioters have little fear of retaliation or criminal punishment, and when compunctions about killing are inoperative. Rioters attack strong targets at weak moments. They don't attack vulnerable, weak third parties or scapegoats.

It is clear that rioters tend to judge risks accurately, as evidenced by lopsided casualty counts. Horowitz commented that he does not know of a single case in which more rioters were killed than the people they targeted. The fact that deadly ethnic rioting is a low risk enterprise for rioters suggests that policymakers can deter rioters by increasing their risks.

Horowitz noted that social scientists have exerted more effort to explain why ethnic violence occurs than why peaceful relations exist among different ethnic groups. It is clear, however, that deadly ethnic riots cannot proceed without social support among members of the perpetrator group. This comment provided the link between Horowitz's remarks and Ashutosh Varshney's presentation, whose work in India provides evidence that civic engagement among members of different ethnic groups can deter those who seek to convert ethnic conflict into ethnic violence.

[ Back to top ]

A Theory of Ethnic Conflict

Ashutosh Varshney made two essential arguments, both focusing on the relationship between the structure of civic life and the presence or absence of ethnic violence:

• Inter-ethnic and intra-ethnic networks of civic engagement play very different roles in ethnic conflict. The former build bridges between ethnic communities while the latter reinforce ethnic boundaries and reduce positive communication and interaction between ethnic groups. Inter-ethnic networks can work as agents of peace. But if communities are organized only on intra-ethnic lines—where members of different ethnic groups do not mix together and interconnections among ethnic communities are limited or do not exist—ethnic violence is much more likely to break out.
• Civic engagement takes two different forms: organized civic networks and everyday civic networks. The former includes a wide variety of associational forms of engagement between different ethnic groups, as realized through business associations, professional organizations, reading clubs, film clubs, non-governmental organizations, trade unions, and political parties. By contrast, everyday forms of engagement include simple routine interactions of neighborhood life, such as families from different ethnic groups sharing meals together or visiting each other, participating in joint festivals, and permitting their children to play together. If robust, both forms of engagement promote ethnic peace. Of the two, however, associational forms are sturdier than everyday forms as bulwarks against ethnic violence. Vigorous associational or organizational life can act as a serious constraint against the polarizing strategies of political elites intent on manipulating ethnic conflict for their own political purposes.

Varshney proceeded to define key terms used in his analysis. Civil society or civic life is that part of our lives that exists between the state or government, on the one hand, and family life on the other. Civil society engages people in a whole range of public activities that are wholly or partly independent of the state. That is, while civil society may not be a non-political space in our lives, it is a space in which the state is not involved. In its non-state functions, civil society can cover both social and political activities—including soccer leagues, trade unions, and political parties. Varshney noted that political parties in a one-party system are not part of civil society but merely an appendage of the state. But political parties in a multi-party democratic system are part of civil society as well as the state. The parties running the state are part of the state while the parties outside the state are part of civil society.

Turning to his definition of the word "ethnic," Varshney noted that, in the past, social scientists concentrated on racial or linguistic characteristics of groups. Research conducted by Horowitz has led to the adoption of a broader definition in which ethnicity is determined by birth-based, ascriptive group identities, whether real or imagined. That is, race, language, religion, tribe, and caste all can be described as ethnic markers. Varshney argued that one reason why this broader usage has been widely accepted is that ethnic conflict—whether religious, linguistic, racial, or tribal—in different locations may have different intensity, passion, longevity, or relative intractability.

Varshney also argued that social scientists generally have failed to make an important distinction between ethnic violence and ethnic conflict. In any ethnically plural society that allows free expression of political demands, some ethnic conflict is more or less inevitable. If different ethnic groups exist along with freedom of organization and expression, there are likely to be conflicts over resources, identity, patronage, and policies. The real issue is whether ethnic conflict is violent or whether it is waged in the institutionalized channels of the state.

If ethnic protest is expressed through such institutions as parliaments, assemblies, or government ministries or through non-violent street demonstrations, it takes the form of ethnic conflict but not ethnic violence. Such institutionalized conflicts must be distinguished from situations when protests become riots and, in their most extreme forms, civil wars or pogroms. Accordingly, ethnic peace should be conceptualized not as an absence of conflict but an absence of violence. Efforts can be made to reduce the level of violence and to transform violence into mere conflict.

[ Back to top ]

Tracking the Incidence of Ethnic Violence

Observers of ethnic conflict are struck by two facts: first, some ethnically diverse communities manage to remain peaceful while others experience enduring patterns of violence. Second, some communities that have maintained a long record of ethnic peace eventually explode in violence. How does one account for such variations?

To answer this question, Varshney's research team collected data on the incidence of Hindu-Muslim violence throughout India between 1950 and 1995. They noticed, first that from 1950 to the mid-1970s, there is no discernable pattern in the number of deaths resulting from riots. From the mid-1970s through the mid-1990s, however, there was a clear upward trend in the number of riots and deaths nationwide.

Analyzing these statistics, Varshney found that the number of deaths from riots in rural India is very small—accounting for only four percent of riot deaths. This is despite the fact that, in 1950, 85 percent of Indians lived in rural areas. (This number had declined to 67 percent by 1995.) Riots in India, therefore, are primarily an urban phenomenon.

Varshney's team then examined whether ethnic riots are evenly spread throughout India by identifying cities according to three criteria: those that had experienced at least 50 deaths in 1950–95; those in which 50 deaths occurred in at least ten riots; and those in which the riots were spread out over five five-year periods. In short, Varshney sought to identify Indian cities in which riots that were relatively frequent and deadly had occurred over a long period of time.

Only eight cities—Bombay, Ahmedabad, Aligarh, Hyderabad, Meerut, Baroda, New Delhi, and Calcutta—qualified for inclusion in this group. Together, these cities accounted for 46 percent of all riot deaths in India between 1950 and 1995. This is despite the fact that the population of these cities accounted for only 18 percent of India's urban population, and a mere 6 percent of India's total population. These findings led Varshney to the conclusion that, even in urban India, riots are heavily concentrated in a few cities.

[ Back to top ]

Precipitating Sparks

The sparks that cause riots can come from events or developments at the national, state, or local level. What is revealing is how communities react in very different ways to the same sparks. When the mosque in Ayodhya was destroyed by Hindu militants in December 1992—an event watched by an estimated 200 to 300 million TV viewers all over India—cities throughout the country responded very differently, with some experiencing riots and others remaining peaceful. Varshney argued, therefore, that one must focus not only on the different sparks that precipitate violence but also on the local mechanisms that either extinguish these sparks or transform them into full-fledged conflagrations.

Pursuing this line of inquiry, Varshney's research team selected three towns from among the eight most riot-prone cities named above—Ahmedabad, Aligarh, and Hyderabad—and matched them to three relatively peaceful cities in which Hindu-Muslims proportions in the population were roughly the same. In this comparison Ahmedabad was compared to Surat (a city in which only one deadly riot occurred between 1950 and 1995); Aligarh was compared to Calicut (which hasn't had a single deadly riot); and Hyderabad was compared to Lucknow (the locus of only two riots, in 1924 and 1941, in which people died).

Systematic comparison of these six towns led Varshney to the conclusion that the presence or absence of pre-existing local networks of civic engagement between the Hindu-Muslim communities is the single most important predictor of whether a community will respond violently to ethnic provocations. Varshney found that those communities that are most peaceful had local associations that made their members aware of the dangers of ethnic violence and worked to suppress the violent and criminal elements interested in exploiting ethnic conflict. Local civic associations—including trade unions, professional organizations, and political parties whose members cut across ethnic lines—were effective in killing rumors, improving communication, and exerting pressure on violent elements at the local level. Such organizations, which stand much to lose from communal splits, worked hard to maintain their turf. Less peaceful were communities in which Hindus and Muslims interacted together in daily life but where associations bridging ethnic divides did not exist or were weak. The communities most vulnerable to ethnic violence were those in which Hindus and Muslims had little interaction, either on a neighborhood or associational level.

If politicians insist on polarizing Hindus and Muslims for electoral advantage—and many will do so if it will help them edge out their competition—they can tear the fabric of everyday engagement apart. In fact, according to Varshney, a nexus of politicians and criminals was in evidence in all of the eight riot-prone cities identified above. Organized gangs protected by politicians can easily disturb the peace at the neighborhood level, often causing migration from ethnically mixed neighborhoods to ethnically homogenous ones. Towns in which civic associations were strongly active across ethnic lines are most resistant to the polarizing activities of criminal gangs and politicians.

Varshney also addressed the important question of whether civic associations fail because of ethnic violence or violence occurs because of the weakness or absence of such organizations. To answer this question, his research focused on the once highly integrated Indian town of Ahmedabad, Gandhi's adopted hometown. India's first trade union was established there and the town was also the home of a host of other effective civic organizations that developed during the anti-colonial struggle against British occupation of India. Varshney argued that Ahmedabad was untouched by riots during the partition period in 1947 because of the existence of strong civic organizations. By 1969, however, they had declined substantially for reasons that had nothing to do with tension between the town's Hindu and Muslim communities. In that year, a deadly riot that lasted five days and resulted in 633 deaths broke out in Ahmedabad after the spread of a rumor that a Muslim had killed a cow (considered sacred by Hindus). The lesson learned from Ahmedabad and other case studies, suggested Varshney, is that riots are more likely to occur in cities where civic associations are weak or non-existent. Such organizations may be frayed by violence, but the odds are that their robust existence will prevent tensions from turning into large riots.

In arguing his case, Varshney made an analogy to seismology. If civil society in any given community is organized on inter-ethnic associational lines, there is a good chance it can sustain ethnic earthquakes or shocks that rank quite high on the ethnic Richter scale—such as desecration of an holy space viewed on TV by millions of people. If, on the other hand, individual members of different ethnic groups within a community relate to each other only in the context of regular neighborhood interaction, earthquakes of a smaller intensity—such as police brutality against a particular ethnic group or defeat of an ethnic party in elections—can give rise to communal violence. If engagement among people within a community is only along intra-ethnic lines, then small tremors—such as unconfirmed rumors and sports victories and defeats—can unleash unimaginable torrents of violence. In short, a multi-ethnic society with few inter-connections across ethnic boundaries is very vulnerable to ethnic violence.

Varshney noted that his conclusions are probabilistic, not law-like, and that some exceptions to those generalizations are likely to exist. But the odds of these exceptions occurring, he argued, are low.

Varshney also argued that observers must have a realistic view of the role of the state in ethnic violence. Citing examples from India, Sri Lanka, and Malaysia, he argued that some states act in politically strategic, not legally correct, ways when addressing ethnic tension and violence. States should be more proactive than they often are in preventing ethnic violence. Instead, they often combine legal, moral, and political calculations in highly unpredictable ways, allowing ethnic violence to occur. A more realistic understanding of how states actually function, as opposed to how they should function, does not mean that citizens should cease to criticize and pressure states when they fail to protect lives threatened by ethnic violence, nor should citizens stop trying every constitutional means to hold states accountable for unlawful behavior. But while making such attempts, citizens should not assume that states will behave properly even when pressured to prevent violence.

If the goal is to reduce ethnic violence, argued Varshney, building integrated civic networks is a better bet than trying to change state behavior. Towns such as Surat in the state of Gujarat, where the lives of Hindus and Muslims are deeply intertwined, avoided violence altogether or contained violence at a very low level. If civic life at the local level is highly integrated and organized across ethnic lines, the government will find powerful partners in civil society to prevent or stem violence. It follows, according to Varshney, that citizen action and interventions have to take two forms. First, citizens need to continue to pressure the state when it violates its constitutional duty to prevent ethnic violence. At the same time, and more importantly, citizens need to build integrated civic structures in their own communities. The first strategy—pressure upon the state—is the primary one used by citizens, but it is insufficient on its own.

[ Back to top ]

The Role of Ethnic Cleavages

Horowitz noted that two of the communities in Varshney's study that remained peaceful—Lucknow and Calicut—are places where serious Hindu-Muslim cleavages do not exist, and there are cleavages of a different kind. In the case of Lucknow, there is a Sunni-Shia cleavage, and in Calicut there are caste cleavages among Hindus as well as the presence of various other religious groups, including a large Christian community. Calicut is a case of what Horowitz described as "multi-polar fluidity" as opposed to the bipolar cleavages of many communities where ethnic violence occurs. The question is whether the complex cleavage structure, as opposed to the high degree of civic engagement, explains the relative peace in Lucknow and in Calicut. Horowitz also suggested that while Varshney's work may explain the importance of civic engagement to resistance against violence, it does not provide a full-blown explanation of what causes violence.

Varshney replied that the role of ethnic cleavages is an important issue that he will continue to investigate in his future comparative resevariables—cleavage structure and civic organizations—may not be as linked as some think they are.

[ Back to top ]

The Role of Economic Development

During the question and answer period, both speakers responded to a question about whether there is a correlation between economic prosperity and reduced levels of violence because of growing economic cohesion between ethnic communities. Varshney responded that, in the 1950s and 1960s, social scientists widely believed in modernization theory, which posited that, over time, economic development would lead to the obliteration or weakening of communal identities. Clearly that theory is wrong, he argued. But Varshney noted that there is some relationship between ethnic riots and prosperity because such riots are much less frequent in prosperous countries, where ethnic prejudice most often takes the form of hate crimes, not riots.

With respect to the Indian case, Varshney observed that the state of Gujarat has the highest economic growth rates in India—on the order of 8 or 9 percent per year, thereby doubling its citizens' income every seven or eight years over the past twenty years. The second richest state in India, Gujarat will soon become the richest state if the current rates of growth continue. Yet, despite this, Gujarat was the scene of gruesome Hindu-Muslim riots in 2002. Such evidence suggests that poor societies experiencing periods of prosperity and high growth rates may still suffer from ethnic riots and only become relatively invulnerable once they reach some threshold of affluence.

On this same question, Horowitz pointed out that statistical tests of a posited relationship between cross-national prosperity and ethnic conflict—not ethnic violence—do not confirm a connection between those variables. At the same time he noted that there has been a drastic decline in the incidence of ethnic riots in the West. But whether this is attributable to prosperity, or to a profound ideological shift in the post–World War II period, is very much an open question.

[ Back to top ]

Outside Instigators

In response to a question about whether outsiders instigated ethnic riots in Gujurat, Varshney noted that the victims he interviewed always said that the rioters came from outside the neighborhoods where the riots occurred. When he asked the victims how outsiders knew whose houses to attack, the victims suggested that perhaps neighborhood residents provided information about which houses belonged to members of the targeted ethnic group.

Horowitz responded to the question by noting that the organization of ethnic riots varies from one setting and time frame to another. Sometimes these riots are truly local and spontaneous in nature, and sometimes they are well-organized. The greater the degree of organization, the more likely it is that such riots will include people from outside the immediate community or neighborhood where the rioting occurs. When outsiders do participate, they tend not to come from very far away. Horowitz added that blaming outsiders for instigating riots may be a way to avoid local responsibility for these events.

[ Back to top ]

Stimulating Civic Engagement

Varshney responded affirmatively to a question about whether civic engagement can be promoted in post-conflict situations to reduce ethnic violence. He suggested that efforts to build inter-ethnic organizations in settings that have recently emerged from long-term, widespread violence, such as civil wars, are likely to fail in the short run, but can succeed in other vulnerable communities emerging from less sustained violence. In areas recovering from civil war, offering public goods—such as community computer centers, libraries, and TV and video centers—may help increase positive interaction among community members, although establishing effective inter-ethnic organizations that prevent violence may not be possible in the short run.

Varshney agreed with Horowitz that creators of inter-ethnic organizations must focus on issues of common concern to the communities in conflict. Varshney cited the example of the Indian city of Bhiwandi, where the building of inter-ethnic civic associations that worked on such problems as sanitation and electricity over a three-year period helped break a pattern of ethnic rioting. Significantly, these organizations did not focus explicitly on achieving Muslim-Christian harmony.

[ Back to top ]

Additional Questions

The discussion then turned to whether riots are more likely to occur in urban, rather than rural, areas. Horowitz affirmed that that is generally the case, first because the events that spark riots tend to occur in urban areas, and, second, because members of different ethnic groups often don't live in the same village or they live in separate parts of the village. He added that rural riots generally are much harder to control and often have much higher casualty rates. Varshney noted that it is true that Indian Muslims are proportionally more urban than Hindus, although in fact many more Muslims live in rural India than in urban areas. He rejected the notion, however, that the incidence of ethnic violence in India can be explained on the basis of national-level variables, such as the urban/rural distribution of any particular ethnic group. Such an explanation, he argued, cannot explain the actual distribution of violence in India.

In response to a question about when ethnic riots end and broader conflict begins, Horowitz explained that expulsion often takes place in the context of riots. Organized expulsion is different, however, and takes much more involvement of the state. He argued that the outbreak of warfare requires additional explanatory variables. With respect to secessionist warfare, for example, there must first exist a plausible territory to which a target group can retreat and to which the group has made a claim. Above all, the violence must be organized, as opposed to spontaneous, for secession to occur. With respect to civil wars in which secession is not an issue, leaders must have calculated that the opportunities emerging from war are superior to the costs of war.

Varshney argued that the main distinction between riots on the one hand and pogroms or civil wars on the other is that in the latter case the state takes sides. That is, when ethnic riots occur, there may be doubts about where the state stands, but the principle of state neutrality is still in effect.

In response to a question about the role of parliamentarians in Indian riots, Varshney argued that some are catalysts of violence while others promote peace. In short, they calculate what best serves their interests, but their behavior cannot explain why riots occur in some locations and not in others. They simply calculate what is in their electoral interest, given the local cleavages.

Asked if inter-religious dialogues held the key to ethnic peace in India, Varshney suggested that, with minor exceptions, Indian religious leaders show little willingness to engage in constructive dialogue. One interesting thing to note, however, is that Hindu-Muslim violence generally does not take place around Sufi tombs. In any case, both he and Horowitz argued that it appears that mere dialogue and inter-ethnic contact are insufficient to prevent the outbreak of violence.

Asked if rioters perceive riots to have diminishing returns, Horowitz argued that one cannot assume that the proclivity for violence peters out once violence occurs. In fact, some locations become habitual centers of ethnic violence. Karachi had that status in the 1980s and early '90s, when it appeared that the threshold for the onset of violence became greatly diminished, the violence itself became more organized and brutal, and the rioters became better armed over time.

[ Back to top ]

Conclusions and Recommendations

• Deadly ethnic riots are not random, unpredictable events. They are responses to certain conditions that can be understood, analyzed, and prevented.
• Governments can reduce the likelihood that ethnic riots will break out by increasing the perception by potential rioters that participating in riots is risky.
• While events at the national or regional level may spark ethnic violence in India, the response to those sparks—ranging from increased ethnic tension to deadly ethnic riots—occurs at the local level. Therefore, explanations of why some Indian communities respond violently to ethnic provocations expressed at the national level while others do not must be found in factors operating primarily at the local level.
• Inter-ethnic civic associations in India are very effective in the effort to prevent or reduce violence. They help dispel inflammatory rumors, identify and isolate rabble-rousers, hide and protect potential victims, and assist the police in crowd control and in their investigations. Inter-ethnic civic associations also provide pre-established networks of communication across ethnic lines that can prove invaluable during the chaotic circumstances that lead to ethnic riots.
• Civic associations in India are effective in stemming violence only if they promote the mutual interests of two ethnic groups in concrete ways. Although further research is needed, it appears that trade unions, professional groups, opposition political parties, and other civic associations that represent mutual political, economic, and social interests across ethnic lines are more effective in preventing violence than civil society groups whose primary focus is the promotion of inter-ethnic dialogue.
• The most common response by citizens seeking to prevent ethnic violence is to criticize and pressure the state in an effort to make it accountable for its failure to take appropriate measures to prevent violence. While this strategy is important, a different strategy that focuses on building strong inter-ethnic civic associations at the community level may ultimately prove more effective in reducing the outbreak of violence in India.

[ Back to top ]

About the Report

Donald L. Horowitz of Duke University and Ashutosh Varshney of the University of Michigan made presentations at the U.S. Institute of Peace on October 31, 2002, based on their research about ethnic violence. In both cases, this research was funded by grants made by the Institute's Grant Program.

Professor Horowitz's research was published in his book The Deadly Ethnic Riot (University of California Press, 2001), which examines 150 episodes of ethnic violence in some 50 countries and also presents evidence from some 50 control cases—that is, locations where riots might have occurred but did not. Professor Varshney's research resulted in the book Ethnic Conflict and Civil Life: Hindus and Muslims in India(Yale University Press, 2002), and focuses on why ethnic violence occurs between Hindus and Muslims in some cities in India and not in others.

Judy Barsalou, director of the Institute's Grant Program, prepared this report.

The views expressed in this report do not necessarily reflect views of the United States Institute of Peace, which does not advocate specific policies.

[ Back to top ]

For information on additional Special Reports, please e-mail Public Outreach, or call (202) 429-3832.