Have riots in India declined over the past three years?


Last week, the National Crime Records Bureau (NCRB) published data on major crimes in India recorded by police departments across states till 2016. At first glance, the data suggests a sharp drop in riots in the country over the past three years. But this impression is partly a statistical artefact, caused by changes in how NCRB classifies riots data.

Mint analysis of the crime data suggests that while riots and civil disturbances have indeed declined in 2016, this followed a marginal rise in such incidents in 2014 and 2015. The rate of such incidents has declined in the past two years, after it rose marginally in 2014, and is now back to levels witnessed last in 2013. However, what is worrying is that the rate of convictions in riot cases has declined marginally even as the overall conviction rate has climbed up in the country.

The analysis is based on a reconstructed time series of riots data, taking into account the reclassification of crime categories by NCRB as well as adjusting for methodological inconsistencies in how NCRB has normalized population figures across years. As an earlier Plain Facts column had pointed out, the use of outdated population projections (based on 2001 census) and inconsistent approach across years renders a historical comparison of NCRB crime rates meaningless. To compound problems further, till 2014, NCRB provided aggregate data under three related crime heads—riots, unlawful assembly, and offences promoting enmity between groups—together under the category of “riots”. Since then, the NCRB has begun providing the data for the three sub-categories separately, without providing a back series of such sub-categorization.

The change in the format of data collection was part of a periodical review by the NCRB and was made after a technical committee under the ministry of home affairs suggested the revision in early 2014. “States were asked to provide data in the revised format from 2014. That’s why there is a sudden change in the incidence of riots in 2014,” said a person with direct knowledge of the matter, requesting anonymity.

According to experts, there is a thin line between the three categories and a segregation of data will enable better understanding of the type of crime.

“An unlawful assembly or an offence promoting enmity need not convert to a riot. They are stand-alone offences. Similarly, all riots need not be based on hate speech (under Section 153A). Each offence has a distinct characteristic. Also, a single person could also be booked for an offence promoting enmity between groups which therefore cannot be counted as a riot,” said Y.P. Singh, a former IPS officer and lawyer.

While the re-categorization may be helpful, the lack of past data for the disaggregated categories makes it difficult to compare the current number of riots to the past number. A simple comparison would be misleading since the data for the past years refers to a broader category. To circumvent this problem, we merge the three related categories for recent years and compare that with the riots figures for the past. We call this category “riots and civil disturbances”. Data for the past three years suggest that riots account for roughly 86% of incidents in the overall category. Applying the same ratio for the past years, we then estimate the data for riots in past years as well.

The reconstructed time series—for riots and for riots and civil disturbances —shows a decline in both in 2016 after a marginal rise in previous years. The rate of riots (based on population estimates derived from the 2011 and 2001 censuses) shows a marginal decline since 2014.

The data shows that the share of agrarian and caste-based riots has risen over the past three years even as the share of communal conflicts has declined. However, this categorization is likely to be a rough categorization given that most riots do not fall in any of the specified categories, and have been classified as “other rioting”. According to the NCRB, ‘other rioting’ includes rioting due to ‘previous enmity, family disputes, heated arguments, rioting due to power supply issues, rioting due to civil disputes and attacks on police etc.’ It is possible that some of those riots classified as ‘other rioting’ have a caste or communal angle to them.

Among states, Bihar, Uttar Pradesh and Maharashtra witnessed the highest number of riots and civil disturbances in 2016. While Bihar and Maharashtra figured in the top three states even in 2011, for Uttar Pradesh, the past five years have marked a sharp rise in such incidents. Himachal Pradesh, Delhi, and Punjab had the lowest number of such incidents.

Punjab’s case is slightly surprising, given that it has recorded no riots or civil disturbances at all for the past few years. Another surprise has been the sharp rise in the reported number of unlawful assemblies in Rajasthan, which has pushed up its overall number of riots and civil disturbances.

In terms of rate of riots and civil disturbances, Kerala tops the list, followed by Bihar and Haryana.

Haryana, Jharkhand, Bihar, Gujarat and Maharashtra ranked among the top five states in terms of the reported number of communal riots in 2016. Riots due to caste conflicts were reported to be the highest in Uttar Pradesh, Bihar, Tamil Nadu, Maharashtra, and Gujarat.

The most worrying part about the data on riots and civil disturbances is the declining rate of convictions in such cases. The conviction rate of riots and civil disturbances has declined steadily over the past three years to reach a 17-year low of 18.1% in 2016.

This decline comes at a time when the overall conviction rate for Indian Penal Code (IPC) crimes has been moving up. At 46.8% in 2016, the overall conviction rate for IPC crimes is at its highest level in India since the turn of the 21st century.