In a recent research paper with my co-author Sriya Iyer (University of Cambridge), titled Religious riots and electoral politics in India, published in the Journal of Development Economics, I have found that in state elections held between 1981 and 2001, a Hindu-Muslim riot occurring in the year preceding the election, caused, on an average, a 5 percentage point increase in the BJP’s vote share in the election. In many constituencies a five-point swing is enough to win the seat and presumably, if constituencies could be targeted, then this could be a powerful, albeit deadly, electoral weapon. It needs to be stressed that the paper did not look for, nor did it find, any evidence to show that riots were being used as an electoral strategy. But it does show that if the party were to be led by a cold, ruthless person who would want to win elections at any cost, then there is enough incentive to instigate targeted riots for electoral gains.
Hindu-Muslim riots have a long history in India that precedes independence. After the gruesome violence of Partition, there was a period of relative calm. There were an average of 16 reported riots per year in the period 1950-1980. (All riot data is taken from news reports in The Times of India, originally compiled by Ashutosh Varshney and Steven Wilkinson, and then extended by us). Subsequently, the period that we study, 1981-2001, had on an average almost three times as many riots reported annually. This period also coincides with the rise of Hindutva politics and of the BJP as an electoral force in the country. While investigating the effect of riots on election results, it is important to keep in mind the possibility that both riots and an electoral swing in BJP’s favour may be caused by some unobservable underlying changes in society – like increased consolidation of the Hindu identity. Or the possibility that it is the upswing in voters’ preference for BJP in upcoming elections that is causing riots as a reaction. To isolate the causal effect of riots on election results, we identify a driver of riots that is independent of both the reasons mentioned above. (This is called the instrumental variable estimation technique). We posit that when an important Hindu festival falls on a Friday, it increases the probability of a riot happening. In fact, the data shows that it almost doubles the probability of a riot from 8 per cent to 15 per cent. We use this increase in probability of riots to find that riots do affect election results, independent of other causes, and on average they swing them towards the BJP.
It is challenging to isolate the mechanism through which the occurrence of a riot translates into a change in vote share. However, given the aggregate nature of the data, we find some patterns that point towards some likely explanations.
First we look for the possibility of change in voter participation because of riots. While we do not find an effect of riots on turnout ratio (the proportion of registered voters coming out to vote), we do find that in districts where the Muslim share of population is above the median, riots lead to a drop in the voter registration ratio (the ratio of voting age population that is registered to vote). While the precise reasons and implications for this are unclear, we can speculate that this could either be because Muslims themselves become reluctant to register (as sometimes electoral rolls are used to target Muslim households) or that the state machinery makes it more difficult for them to do so. The other possible mechanism is a change in voting preferences of the voters. We find that the effect of riots on elections is short-lived, in the sense that an election that happens three years after a riot does not see any appreciable effect. This lends weight to the explanation forwarded by political scientists, where temporary polarisation on religious lines leads Hindus to vote for the BJP rather than for the Congress or other regional parties, while Muslims vote for the Congress or other non-BJP parties. But since Hindus are in the majority in almost all constituencies, this polarisation ends up benefiting the BJP. Directly verifying this is non-trivial since data does not show voting behaviour by religion.
While the fact that riots help the BJP is not news to people who are aware of ground realities, but the extent and prevalence of the effect even today is deeply worrying, to say the least. Given the clear electoral incentives, instigating communal riots in targeted areas could well be one of the tactics used by the BJP in closely fought elections.
The writer is assistant professor, School of Liberal Studies, Azim Premji University
