To account for the worst incidents of communal violence in post-Partition India, it is essential to develop an understanding of the complex local factors that encourage rioting, rather than rely on the findings of a predictive, statistical model.
Is it possible to predict or explain ethnic and communal violence using quantitative models? Mainstream social science has a growing regard for model building and formal hypothesis testing. In South Asian studies, this trend is most evident in accounts of ethnic riots or communal violence. In a new paper titled “The Search for Order: Understanding Hindu-Muslim Violence in Post-Partition India,” LSE’s Stuart Corbridge, Nikhila Kalra and Kayoko Tatsumi challenge the model proposed by Steven Wilkinson in Votes and Violence: Electoral Competition and Ethnic Violence in India and examine whether the rise of Hindutva forces and Muslim political agency can be so easily dismissed in accounts of violence.
Wilkinson and Ashutosh Varshney (in Ethnic Conflict and Civic Life: Hindus and Muslims in India) have shown conclusively that the geography of communal violence in India is not random. Although more than 70 per cent of all Indians have lived in the countryside since 1950, the victims of Hindu-Muslim violence have almost always been town and city-dwellers. The authors state, “[Varshney and Wilkinson’s] are extraordinary findings, even if they chime with stories from beyond South Asia. What is contentious, however, is the claim, made by both Varshney and Wilkinson, that variations across space and time in communal violence in India can largely be explained—and even be predicted—in terms of one fundamental driver: the presence or absence of civic engagement across the communal divide within individual cities (for Varshney) or levels of competition between leading politicians at the state level for the votes of different ethnic (or religious) communities, most notably ahead of national or State Assembly elections (Wilkinson).”
To challenge the utility of quantitative models, the authors examine Wilkinson’s model against a data set of the 20 worst incidents of communal violence in India since 1950. The following excerpt from the article summarises its main findings:
We find that the Wilkinson model is consistent with some important key facts in our data set, most notably in terms of “percentage Muslims” in riot-affected towns and cities and overall levels of urbanisation. However, proximity to national or state elections is not found to be a strong driver of prolonged ethnic rioting. Nor is it the case that India’s worst instances of communal violence occurred mainly where there was direct electoral competition between only two (or certainly less than 3.5) effective political parties—the other main predictive variable in the Wilkinson model.… Our findings raise the uncomfortable prospect that predicting, and thus policing, communal violence in India is less achievable than Wilkinson might incline us to believe.
To explain the limitations of attempts to explain and predict ethnic violence within the framework of a quantitative model, the authors consider time inconsistencies, principal-agent problems, religiosity and the rise of Hindutva forces, and the homogenisation of riot events. For instance, to illustrate principal-agent problems and the consequences of the fact that political actors are obliged to have their instructions carried out by other agents (meaning that prescribed actions do not necessarily have the same outcomes in all places), the authors cite the example of rioting which hit Bijnor in 1990, when Malayam Singh Yadav’s Janata Dal party headed the Uttar Pradesh government.
The chief minister had successfully pursued a strategy of welding together an electoral alliance between Muslim and Scheduled Caste voters. He had every incentive to prevent anti-Muslim violence. In September 1990, however, the president of the BJP, L.K. Advani, launched a rath yatra designed to rally Hindu support along communal lines against the National Front government’s decision to enact the major provisions of the Mandal Commission report and thereby extend education and public-sector employment quotas to India’s Other Backward Classes (OBCs, including many Muslims). Given that Bijnor fell on the route of this yatra, which had already left a trail of blood in its wake, and which in its own terms was hugely imaginative, UP’s chief minister visited the town in order to urge peace; following the visit he also had several BJP and VHP workers arrested for inciting violence. In normal circumstances, this might have been an effective precautionary measure. In an atmosphere, however, of inflamed tensions Malayam Yadav Singh’s actions provoked a fierce backlash from elements in the Hindu majority community and this proved a catalyst for communal violence.
The article’s authors conclude that “the causes of riots and prolonged rioting in India are too diverse to be captured in a predictive, statistical model. The error terms matter too much and caution us against single-track prescriptions for dealing with ethnic violence in India (or, indeed, elsewhere).” To account for the worst incidents of communal violence in post-Partition India, the authors argue, it is essential to develop a close understanding of the complex local factors that encourage or discourage the emergence of riots. They ultimately caution against assuming that patterns of violence or their underlying causes can be grasped within a unifying model.
For more on this topic, see Stuart Corbridge, Nikhila Kalra and Kayoko Tatsumi, “The Search for Order: Understanding Hindu-Muslim Violence in Post-Partition India”, Pacific Affairs: Volume 85, No. 2 (June 2012)