What the data says: India faces bigger threats from Jihad and Maoism
Rupa Subramanya Rupa Subramanya | 12 May, 2016
IN 2009, ACCORDING to WikiLeaks, Congress leader Rahul Gandhi had told the then US Ambassador to India, Timothy Roemer, that “the bigger threat [than Islamic terrorism] may be the growth of radicalised Hindu groups, which create religious tensions and political confrontations with the Muslim community.” We have no way of knowing on what basis Gandhi made this assessment, nor do we know quite what his fears were. He could perhaps have had in mind communal violence or perhaps even acts of terrorism.
Then Home Minister P Chidambaram was more explicit at a gathering of intelligence experts when in 2010 he warned of the danger of “saffron terrorism”. He further pointed to “many bomb blasts of the past”, again without giving any analytical basis for his claim or furnishing data to back it up.
Not to be outdone, Chidambaram’s successor Sushil Kumar Shinde raised the spectre of ‘saffron terrorism’ in 2013, pointing to a few alleged incidents, but without putting them in the larger context of all terror acts committed by all perpetrator groups in the same period. Almost a month later, after a backlash, Shinde apologised for his remarks.
Recently, the historian and commentator Ramachandra Guha argued that ‘Hindu fundamentalism’ is a greater danger to India than ‘Islamic terrorism’. Guha’s stated reason for saying so is that Hindus constitute an overwhelming majority of India’s population and therefore fundamentalists from their ranks pose a bigger threat. However, he presents no data or any serious evidence to support his thesis.
These are but a few examples of the types of claims around ‘saffron terror’ and Hindu extremism one hears from establishment politicians and commentators. It’s not surprising that politicians of a party which relies on minority votes make vague claims about ‘saffron terror’ without backing them up. What is less excusable is a scholar of Guha’s distinction making similar claims.
Do such claims stand up to scrutiny? Is ‘saffron terror’ a greater threat than Islamic terror in India? Such assertions need to be weighed against the evidence.
A large and as yet oddly unexploited dataset exists and is the basis for the analysis presented here. The Global Terrorism Database (GTD) is maintained by the US-based National Consortium for the Study of Terrorism and Responses to Terrorism (START). It is the largest and most scientifically compiled database on global terror and is being constantly updated. For India, there are close to 10,000 events identified between 1972 and 2014.
The demolition of the Babri mosque in 1992 shows up in the Global Terrorism Database as a terrorist act committed by a Hindu group
The GTD has a precise and rigorous definition of terrorism. A terrorist act is defined as ‘the threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation’. This excludes the legitimate use of violence by the state, and is confined to non-state actors who intentionally use violence or its threat to try to achieve a well-defined goal.
Each event is coded with a varying level of detail, which allows researchers to determine with a fair amount of accuracy the identity of the perpetrator group. Analysing this data will therefore let us test the data-free hypotheses put forth about the threat of ‘saffron terror’. It took me more than a month to code such a huge dataset, as I manually worked through each incident and where possible, fact-checked the coding in the database by looking at relevant reports of that alleged terror incident. While extremely time-consuming, this methodology is superior to using mechanical algorithms to code events as the judgement of the researcher is required when the facts are incomplete or uncertain.
Scholars like to think of terrorism and communal violence as two distinct categories, each with different causes and explanations, but in practice they’re not always easy to separate. Even with a rigorous definition of terrorism, there are grey areas. For instance, what is the status of a terror incident which then morphs into an episode of communal violence? Whether it is coded as ‘terrorism’ or not depends on whether the subsequent communal violence contained elements of it or not.
The assassination of Indira Gandhi by her Sikh bodyguards, for example, appears as it should in the GTD as a terrorist act. However, the anti-Sikh pogrom in Delhi orchestrated by senior Congress leaders does not figure in the database, as it did not contain elements of terrorism. As another example, the demolition of the Babri mosque in 1992 shows up in the GTD as a terrorist act committed by a Hindu group. But in this case, the riots which erupted in several Indian cities (including Bombay), orchestrated by various political parties, also show up because they featured elements of terrorism such as bomb throwing.
In incidents of Pakistani terror in India, it is difficult to disentangle the role of Pakistani agents as against local Indian recruits
One unfortunate gap in the GTD which should be noted is that due to a transition from one keeper of the data to another, the 1993 record was lost. My analysis, therefore, does not pick up incidents of that year—such as the Bombay serial blasts—that represented the continued aftermath of the Babri mosque demolition.
In some cases, information that was unavailable at the time of an incident allowed me to more accurately code the perpetrator. For instance, on 29 May 2002, serial bombings occurred in Ahmedabad and were coded by the GTD team as having an ‘Unknown’ perpetrator, since no individual or group claimed responsibility. However, since then, four individuals were convicted of them by the Gujarat High Court and it was established that the motive was retribution for the communal violence that took place earlier that year sparked by the burning to death of Hindu pilgrims on a train in Godhra. Therefore, I’ve coded these incidents as caused by a Muslim perpetrator group.
So as to stack the deck against myself, I have furthermore coded several incidents which are still working through the legal system as ‘Hindu’ where the GTD refers to perpetrator group as ‘Unknown’. These include so-called saffron terror incidents such as the 2008 Malegaon blasts, where Hindu groups are suspected, but the case is still ongoing. Revelations at the time of writing cast serious doubt on whether the perpetrators were Hindu; but again, to be fair, I’ve retained the Hindu coding. What is more, the few incidents of caste-related violence of the type perpetrated by Bihar’s Ranvir Sena, which show up in the GTD and are coded as ‘Unknown’, have here been coded as ‘Hindu’, even in a few cases where it’s not clear whether the motivation was necessarily religious.
As we’ve seen in several incidents of allegedly caste-based violence, the motives may sometimes be economic and have nothing to do with the caste or religion of either perpetrators or victims. By coding the small number of such cases as ‘Hindu’, if anything, the number of incidents of Hindu terror would be slightly higher than the reality.
We should focus on the principal existential threats emanating from extreme left wing ideology, which fuels Maoist extremism
With these caveats in mind, the overall coding of the dataset reckons with the fact that the perpetrator group can be clearly identified by religion in some cases, and in other cases by highly specific, geographically well-defined insurgencies, which may or may not have anything to do with religion. Thus, following other research on terrorism within India, incidents within the regions of Jammu & Kashmir, the Maoist belt, and the Northeast are coded as such not by religious or any other identity of the perpetrator group. Another small group of incidents are coded as referring to the Tamil insurgency in Sri Lanka which spilled over into India.
By contrast, incidents that took place outside these well-defined geographical insurgencies are categorised by the religion of the perpetrator group—in particular, Hindu, Muslim or Sikh. A very small number of incidents fall into the category ‘Foreign’, such as terror strikes, which occurred in India but were unrelated to Indian issues. Two examples from 1972 were hijackings by the Popular Front for the Liberation of Palestine and the Japanese Red Army, both involving aircraft which had landed on or taken off from Indian soil.
Readers might wonder why there is no separate code for terrorist incidents in India emanating from Pakistan, the obvious example being the Mumbai attacks of 26/11. The difficulty is that in many incidents of presumed Pakistani terror in India, be it in J&K or elsewhere, it is often difficult if not impossible to disentangle the responsibility and role of Pakistani agents and mercenaries as against local Indian recruits and sympathisers. Thus, as mentioned, all incidents in J&K are coded by the geographic region, not as ‘Muslim’, even if some evidence exists in a particular incident of direct Pakistani or jihadi involvement. By contrast, the 26/11 attack is coded as ‘Muslim’, given that the motivation was religious. Recall that one of the targets was a Jewish cultural centre in Mumbai.
Lastly, where too little information was known to properly code an incident, or it was related to violence between political parties but unrelated to religion or identity, it was categorised as ‘Unknown, Other’.
Islamic terrorism is five times more likely to occur than Hindu terrorism, even excluding Jammu and Kashmir
Bearing all of the above in mind, here are the main findings, from a total of 9,069 incidents between 1972 and 2014: Maoist: 29 per cent; Northeast: 25 per cent; J&K: 21 per cent; Sikh: 13 per cent; Unknown/Other: 6 per cent; Sri Lankan: less than 1 per cent; Foreign: less than 1 per cent; Muslim: 3 per cent; Hindu: 0.6 per cent.
If one breaks down the data into decades, the numbers reveal that the 1980s were dominated by Sikh terrorism and the 1990s were dominated by terror in J&K. In the 2000s, the Northeast overtakes J&K. Finally, from 2010 to 2014, we a see a big spike in Maoist terrorism—the largest during this phase, as indeed for the whole period under consideration.
These striking results do not bear out the fear-mongering and alarmist hypothesis of ‘saffron terror’. What is reconfirmed is that the Maoist conflict is the biggest threat that the nation faces, accounting for almost a third of total incidents. Maoist terrorism, as we know, is not motivated by religion but by the extremist left wing ideology of perpetrator groups such as the Communist Party of India (Maoist) whose stated goal is the destruction of the Indian republic. As against this, such violence tends to be confined to an admittedly large swathe of central and eastern India.
Likewise, the large share of incidents in the Northeast are not related directly to religion, but reflect the ambitions of various local insurgent groups in places like Manipur, Nagaland and Assam.
As noted before, the large share of terror incidents in J&K requires care in interpreting. While related to a geographically defined insurgency, namely by groups which believe that the state should be either independent of India or join Pakistan, it is also true that the major perpetrator groups in J&K do have an Islamist motivation. That accounts for incidents of violence directed at Kashmiri Pandits, many of whom have consequently been driven out of their ancestral homes— which some would argue amounts to ethnic cleansing.
It would be disingenuous to claim that the large share of J&K incidents are unrelated to Islamism, which clearly is a motivation both for perpetrators within the region and those they might be working with in Pakistan’s military and intelligence establishment.
If terrorist incidents in J&K were considered Islamic and added to the non-geographical ‘Muslim’ category, then almost a quarter of all terror incidents in this long period would be considered Muslim and be tied with terrorism in the Northeast as second only after Maoist terrorism.
Suppose, for the sake of argument, we exclude J&K from the analysis and assess the ‘saffron terror’ claims purely on incidents of Islamic terrorism outside that state—we would still find that incidents coded ‘Muslim’ were five times as many as incidents coded ‘Hindu’. Given that Hindus account for about 80 per cent of India’s population, according to the latest census, and Muslims for about 14 per cent, so that there are roughly five times as many Hindus as Muslims, the data tell us that the likelihood of a terror incident coded ‘Muslim’ is disproportionately higher than one coded ‘Hindu’ when you adjust for their respective population shares.
This means that in India, Islamic terrorism is five times more likely to occur than Hindu terrorism, even excluding J&K. These hard numbers fly in the face of glib assertions that ‘saffron terror’ rather than Islamism-motivated terrorism is a bigger threat.
As a test of whether these results are an artefact of my coding, compare this with a similar coding exercise conducted by a different research group, the South Asia Terrorism Portal (SATP), which comes up with broadly similar findings. Note that this dataset includes both acts of terror and low-intensity conflict and therefore has a larger number of events. For a total of about 20,000 events between 2005 and 2016, Maoist or Left Wing Extremism again tops the list at almost 35 per cent, with the Northeast at 30 per cent and J&K at 29 per cent. Islamist terrorism outside these regions was about 4.1 per cent and Hindu or saffron terrorism was about 0.6 per cent. While the methodology and time periods are different, these findings are broadly consistent with mine, which covers a much longer time period, but with a narrower definition of terrorism.
Far from pontificating about so-called ‘saffron terror’ and Hindu fundamentalism, we should focus on the principal existential threats to the Indian state emanating from extreme left wing ideology, which fuels Maoist extremism, as well as the jihadi Islamist ideology, which is at the root of terrorism in J&K and other such terrorism outside that state.
Sensible folk, those not wearing ideological and other blinkers, should be concerned by the story that the hard numbers tell us.
(Rupa Subramanya is a co-author, with Vivek Dehejia, of Indianomix: Making Sense of Modern India)
Artificial Intelligence Is Like Allopathy Rajeev Srinivasan
The City of Loss Somak Ghoshal
What A Song and Dance Kaveree Bamzai