An empirical regularity has time and again confronted scholars of ethnic conflict. Despite ethnic diversity, some places — nations, regions, towns, villages — remain peaceful, whereas others with the same diversity experience frequent outbursts of violence. Similarly, some multiethnic societies, after maintaining a long record of peace, explode all of a sudden. Scholars have sought to explain these variations across time and space for some time now. On Hindu-Muslim conflict, mainly confined to urban India, my book, published some years ago, dealt with variation across space.1Ashutosh Varshney, Ethnic Conflict and Civic Life: Hindus and Muslims in India, New Haven: Yale University Press, 2003 A project since then has sought to understand whether the Indian argument is portable to non-Indian settings. In this note, I first summarize my India findings. Then I briefly dwell upon what we have learned from cities in Indonesia, Malaysia and Sri Lanka about factors leading to violence and peace.
Hindu-Muslim relations and urban India
To understand variations across space on Hindu-Muslim violence in India, my book went through all reported Hindu-Muslim riots in the country between1950-95.2The data set, put together in collaboration with Steven Wilkinson (Yale University), is available at ICPSR. http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/4342 Two results were crucial. First, the share of villages experiencing communal rioting was remarkably small. Between 1950-95, rural India, where two out of three Indians still live, accounted for a mere 3.6 per cent of the deaths in communal violence. Hindu-Muslim violence was primarily an urban phenomenon. Secondly, within urban India too, Hindu-Muslim riots were highly locally concentrated. Eight cities alone accounted for a hugely disproportionate share of communal violence in the country: roughly half of all urban deaths and 45 per cent of all deaths, urban as well as rural. As a group, however, these eight cities represented less than a fifth of India’s urban population (and only about 5-6 per cent of the country’s total population, both urban and rural). Eighty two per cent of urban population was not “riot-prone”.
In other words, India’s Hindu-Muslim violence was city‑specific. State and national politics were best viewed as providing the context within which the local mechanisms linked with violence, or peace, get activated. To explain communal violence, we needed thoroughly to investigate these local mechanisms.
Discovering the role of civil society
Following this reasoning, my book on India selected six cities — three riot-prone, and three peaceful — and arranged them in three pairs. Thus, each pair had a city where communal violence was endemic, and a city where it was rare, or entirely absent. Roughly similar Hindu-Muslim percentages in the city populations constituted the minimum control in each pair.3The cities were Aligarh and Calicut, Hyderabad and Lucknow, and Ahmedabad and Surat.
The comparison yielded a clear relationship between civil society and ethnocommunal violence. To be more specific, my argument focused on the intercommunal links (i.e., networks and organizations that integrate Hindus and Muslims), not intracommunal links (i.e., networks and organizations that are all Hindu or all Muslim). Robert Putnam calls the former bridging, and the latter bonding, social capital.4Robert Putnam, Bowling Alone: The Collapse and Revival of American Community, New York: Simon and Schuster, 2000.
These networks can be further separated into two types: organizational and quotidian. I call the first associational forms of civic engagement; and the second, everyday forms of civic engagement. Business associations, professional organizations of doctors, lawyers, teachers and students, reading clubs, film clubs, sports clubs, festival organizations, trade unions and political parties are some of the examples of the former. Everyday forms of engagement cover routine interactions of life such as: whether Hindu and Muslim families visit each other, eat together often enough, jointly participate in festivals, and allow their children to play together in the neighborhood. Both forms of engagement, if robust, promote peace: contrariwise, their absence or weakness opens up space for communal violence. Of the two, the associational forms turn out to be sturdier than everyday engagement, especially when confronted with the attempts by politicians to polarize ethnic communities. Vigorous associational life, if intercommunal, acts as a serious constraint over the polarizing strategies of political elites.
Why should this be so? Two links connect civic life and ethnic conflict. First, prior and sustained contact between members of different communities allows communication between them to moderate tensions and preempt violence, when such tensions arise due to: a riot in a nearby city or state; distant violence, or desecration, reported in the press, or shown on television; rumors planted by politicians or groups in the city to arouse communal bitterness and passions; or a provocative act of communal mischief by the police, thugs or youth. All of these can be equated with sparks that do not necessarily turn into fires. In cities of thick interaction between different communities, peace committees at the time of tension emerge from below in various neighborhoods; the local administration does not have to impose such committees on the entire city from above. Because of mutual consent and voluntary involvement, the former is a better protector of peace than the latter. Such highly decentralized tension-managing organizations kill rumors, remove misunderstandings, and often police neighborhoods. If prior communication across communities does not exist, such organizations do not organically emerge from below. They are typically imposed from above, and the committees from above do not work well because their politician members, though inducted for purposes of peace, are normally already committed to polarization and violence for the sake of electoral benefit. Their presence on peace committees is often merely notional.
Second, in cities that have associational integration as well, not just everyday integration, the foundations of peace become stronger. In such settings, even those politicians who would, in theory, benefit from ethnic polarization find it hard to engender ethnic cleavages, arouse widespread bitterness, and instigate violence. Without a nexus between politicians and criminals, big riots and killings are highly improbable. If unions, business associations, middle class associations of doctors and lawyers, film clubs of poorer classes, and at least some political parties are integrated, even an otherwise mighty politician-criminal nexus is normally unable to rupture existing links. Everyday engagement in the neighborhoods may not be able to stand up to the marauding gangs protected by powerful politicians, but unions, associations and the integrated cadres of some political parties –those, unlike the polarizers, not interested in ethnic conflict – become bulwarks of peace in two ways: their local strength convinces those who would benefit from riots that engineering riots is beyond the realms of possibility, or even if violent cadres of polarizing parties and the thugs associated with them try, they do not normally succeed in instigating riots. Integrated organizations constitute a forbidding obstacle for even politically shielded gangs. When associational integration is available, the potential space for destructive and violent action simply shrinks.
Civic links across communities have a remarkable local or regional variation. Depending on how different communities are distributed in local businesses, middle class occupations, parties and labor markets, they tend to differ from place to place. As a result, even when the same organization is able to create tensions and violence in one city or region, it is unable to do so in another city or region where civic engagement crosses communal lines. Local and regional variation in ethnic violence, its uneven geographical spread, can thus be a function of civic engagement, which tends to vary locally or regionally.
This argument is diagrammatically presented in Figure 1. It builds upon the metaphor of ‘sparks’ (small clashes, tensions, rumors) and ‘fires’ (riots) to make the point about the role of civil society. Intercommunal ties between Hindus and Muslims, not intracommunal ties among the Hindus or among the Muslims, were a strong bulwark of communal peace. If towns and cities were organized only along intra-Hindu or intra-Muslim lines, the odds of riots (fires) breaking out, given sparks (tensions, rumors, small clashes), were very high. In Indian cities, bonding social capital was highly correlated with Hindu-Muslim violence, but bridging ties could put out sparks very effectively, not allowing them to disrupt the local equilibrium of peace. The local organs of the state – the police and administration – simply worked better at riot-prevention in integrated cities. Islamic or Hindu religiosity was not the principal reason for riots. Rise of religiosity was in evidence in both peaceful and violent cities, not simply in the latter. In facilitating or preventing riots, it was the types of civic linkages – bridging or bonding, integrated or segregated – that mattered most.
Figure 1: Civil Society and Ethnocommunal Violence
Extending the terrain
The purpose of a project I have undertaken since publishing my India book is to extend the comparative terrain and seek a multi-national and multi-city comparison. I have selected cities from Sri Lanka, Malaysia and Indonesia. In Sri Lanka, I am studying Tamil-Sinhala conflict; in Malaysia, Malay-Chinese conflict; in Indonesia, Muslim-Christian relations. Some emerging findings are worthy of note.
To begin with, like India, collective violence is highly locally concentrated in Indonesia. A mere 15 districts (kabupaten), holding 6.5 per cent of the country’s total population, accounted for 85.4 per cent of all deaths in collective violence between 1990-2003.5Ashutosh Varshney, Rizal Panggabean and Mohammed Zulfan Tadjoeddin, “Patterns of Collective Violence in Indonesia (1990-2003)”, Journal of East Asian Studies, winter 2008.In most parts of the country, Muslims and Christians live peacefully together.
Beyond Indonesia, too, ethnocommunal violence tends to be highly locally or regionally concentrated. This was as true in Malaysia in 1969 when the country experiences its worst riots since independence in 1957. In Sri Lanka, too, local concentrations were evident, whether in 1956 and 1958, or during 1977-83, the worst phase of rioting on the island.
The project is investigating four hypotheses: two civil society based, and two state based. In addition to the Indian hypothesis (civic integration), the second civil society based hypothesis is “self policing”. Self-policing is a mechanism of peace deductively proposed by Fearon and Laitin but yet to be fully empirically examined.6James Fearon and David Laitin,“Explaining Inter-Ethnic Cooperation”, American Political Science Review, December 1996.It means intra-ethnic, or intra-communal, policing of one’s own youth, who are typically earliest to strike, or strike back, at other groups. If exercised by elders, by an ethnic association, or by civic organizations such as churches, intra-ethnic policing may lead to the same result as inter-communal engagement does in India.
The third and fourth hypotheses concern the state. One must remain open to state-level mechanisms, especially (a) if a state develops capacities to intervene quickly and effectively as ethnic sparks emerge, preventing their transformation into fires; or (b) the state introduces public policies and implements them, bringing a sense of justice or fulfillment to an aggrieved community, reducing thereby the odds of the very emergence of sparks. It is hard to imagine that in political life, sparks can fully be prevented, but the idea is worth exploring nonetheless.
These remarks, of course, should not be construed to mean that in times of ethnic conflict, the state plays only the two roles identified above. States in different parts of the world are known to have played two other kinds of roles. Sometimes, states inflame riots, siding clearly with one of the two ethnic groups in contention;[7. In India’s Gujarat state in 2002, the state played this role. See Ashutosh Varshney, “Understanding Gujarat Violence”, Items and Issues, Newsletter of the Social Science Research Council, New York, Fall 2002.]and at other times, they simply tolerate riots, taking very little action even as the embers of violence burn ferociously.7 The awful riots in Gujarat state in 1969 are an example of such behavior. See Ashutosh Varshney, Ethnic Conflict and Civic Life, Chapter 10 and 11.The state can, thus, play three distinct roles: riot-inhibiting, riot-tolerating, and riot-inflaming. Needless to add, it is when the state plays the first role, normally constitutionally required, that it directly contributes to ethnic peace, not otherwise.
Sri Lanka and Malaysia
Some materials analyzed thus far clearly go in the direction of the Indian hypothesis; others do not. To illustrate this divergence, I will briefly summarize Sri Lankan materials first, and then turn to the emerging Malaysian results.
Of the three Sri Lankan cities in the project, Colombo, Kandy and Negombo, the last has never had Tamil-Sinhala rioting. Negombo is a mere hour and a quarter away from the heart of Colombo city, where rioting during 1956, 1958 and 1977-83 was frequent. In 1983, riots in Colombo, given the involvement of the state, took the form of gruesome pogroms, touching off violence in many parts of the island.8State involvement in the 1983 riots of Sri Lanka is widely accepted. Indeed, during her tenure, Chandrika Kumaratunga, President of Sri Lanka (1994-2005), apologized for the role of the state in 1983 violence. Also see Stanley Tambiah, Sri Lanka: Ethnic Fratricide and the Dismantling of Democracy, Chicago: University of Chicago Press, 1991. Even at that time, Negombo experienced some arson, but no riots, as before.
Negombo is exceptionally ethnically integrated. The most striking local institution is the Catholic Church. Like elsewhere in Sri Lanka, the Tamils and Sinhalese are ethnically distinct in Negombo, but unlike elsewhere, most Tamils and Sinhalese in that town share the Catholic faith. The Church brings the two ethnic groups together and has historically provided bridging social capital, which, as a result, has developed deep, enduring, organic roots. Moreover, the Church is supported by a whole array of integrated organizations – business, labor, middle class – in the town. Negombo, thus, supports the Indian hypothesis about the role of civic integration in promoting ethnic peace. Unfortunately, it is one of the very few towns, if not the only one, where integration between the two leading ethnic communities exists. Most Tamils and Sinhalese in Sri Lanka are Hindu and Buddhist respectively, not Christian. On the whole, religion does not play a bridging role in Sri Lanka; and other aspects of social life of the masses also do not.9While the intermingling of religions might explain Negombo’s enduring peace, it should be noted that such an account does not explain why there were no riots in Sri Lanka once the civil war began in 1983. Sri Lanka after 1983 has not become any more integrated than it was before. The role of the state – whether it was interested in riots after 1983 at all – remains a potential explanation. The city-level materials from Malaysia are also striking. Malaysia has not had Malay-Chinese rioting since 1969. While it is not yet unambiguously clear what explains Malay-Chinese peace since 1969, it is already obvious that the Indian hypothesis – civic integration as a foundation of ethnic peace – is not applicable. The Malay and the Chinese in Malaysia continue to be highly segregated, both in everyday life and in organizations.
To put the matter in perspective, let us briefly compare the city of Kuala Lumpur (KL), the site of Malaysia’s worst Malay-Chinese riots in 1969 but enjoying Malay-Chinese peace ever since,10here was a small Malay-Indian riot, however, in 1998. with Calicut, one of the peaceful cities in the Indian study. (The sampling methodologies in the two surveys were roughly similar.)
Whereas nearly 83 per cent of sampled Hindus and Muslims in Calicut reported “eating together often”, in KL that proportion is 1.8 per cent, and if added to “eating together sometimes”, only 8 per cent. Nearly 90 per cent of the Calicut sample had reported that Hindu and Muslim children played together in the neighborhood, the proportion in KL is 15 per cent. About 84 per cent of Calicut Hindus and Muslims visited each other socially; only 20% of the KL Malay and Chinese report doing so.
In Calicut, a huge proportion of associations and civic organizations – for businessmen, labor, middle class professionals–were integrated. In KL, only 3 per cent said that the business organizations they had joined were mixed; only 2 per cent reported being in mixed labor organizations; a mere 2.3 per cent were in mixed middle-class professional organizations; and finally, only 1.8 per cent, 5.7 per cent, and 1.4 per cent, respectively, said that the NGOs, party organizations and neighborhood associations in which they had participated were mixed.
In short, both in quotidian and organizational life, KL is a highly segregated city. It turns out that this is true of the other Malaysian cities as well.
Despite remarkably little integration, Malay-Chinese peace since 1969 has had it longest run for over a century. Most observers argue that it is the capacity of the state to nip tensions in the bud, or its policy performance, that accounts for the long peace.11James Jesudason, “State Legitimacy, Minority Political Participation and Ethnic Conflict in Indonesia and Malaysia”, in Nat J. Colletta, Teck Ghee Lim and Anita Kelles-Viitanen, eds, Social Cohesion and Conflict Prevention in Asia, Washington DC: The World Bank, 2001.Further research will show whether this explanation is correct. Meanwhile, it should be noted that the Government of Malaysia does not think state-level measures alone can continue to ensure peace, even if they have by now. Considering Malaysia’s racial peace fragile, the government launched a new drive aimed at societal integration. Rukun Tetangga, neighborhood level committees, are the principal organizational vehicle for the drive. Expecting that such committees will bring the three races – the Malays, the Chinese, the Indians — together in everyday life, the government has sought to cover the entire country with them over time:
“National Unity Department director general Abdul Rashid Sahad said racial polarization in the workplace is becoming a worrying trend in the country. He said the tendency of workers to stick within their own community is gradually infiltrating workplaces, which is unhealthy for a multi-racial country like Malaysia. …
In an effort to achieve “zero racial conflict” this year, the department has approved grants for Rukun Tetangga and public education programs, particularly those targeting the urban population, on the sensitivities of various races. Rukun Tetangga (RT) is a neighborhood watch group run by residents in housing estates to combat crimes and to promote racial integration. Abdul Rashid said the government hopes to set up 3,000 RTs to serve 7.5 million people by the end of the year, and eventually increase this to 4,800 for 12 million people by 2010.”“12“Workers of the same race stick together,” The Straits Times, 1 February 2003.
Let me summarize the likely conclusions from emerging materials. First, civic integration remains a promising foundation of ethnic peace, especially in mixed societies. But it is not the only one. Second, in segregated societies, where for a whole variety of historical reasons common civic sites have not emerged, in-group policing may well be an important mechanism of peace. Third, in both kinds of societies, mixed and segregated, purposive state action can play a peaceful role. Solutions are not simply civil-society based. But state action, or state capacity, aimed at peace, cannot be easily legislated. Why this is so requires a historically anchored political understanding.