On August 8, 2015 a Pakistani terrorist was captured alive in Udhampur district of Jammu and Kashmir (J&K). Just the previous day a Kashmiri militant affiliated with Lashkar-e-Toiba (LeT) was killed in a gunbattle with Indian security forces. According to an article published in Scroll “the unstated government policy is to kill a local militant as soon as he is spotted. There is apparently no incentive in capturing an armed Kashmiri militant”.  The writer seemed to imply that armed forces find higher value in capturing foreign terrorists. The article went on to suggest that “a closer scrutiny of militant attacks in Jammu and Kashmir reveals that most of them target the state and its armed forces”.
In what follows I analyze the pattern of terrorism related violence in J&K, and the response of security forces. I tackle these issues using data from the South Asian Terrorism Portal (SATP), having set up the following two workable hypothesis:
H1: The ratio of Captured to Killed is higher among foreign terrorists than locals;
H2: The ratio of Civilian to Armed Forces casualties in terrorist violence is less than one
There are arguments in favor of the first hypothesis. Capturing a foreign terrorist, usually from Pakistan, can be higher value for potential intelligence gathering as well as scoring a diplomatic point against the neighbor. For example, Ajmal Kasab’s capture helped expose the role of Pakistani terror network in Mumbai suicide attacks. Even Pakistan had to grudgingly acknowledge the plot having being hatched from its soil, causing a major diplomatic embarrassment. Media coverage of a captured Kashmiri militant may also generate certain degree of empathy for Kashmiri separatism among the average Indian.
To test this conjecture I compiled all major incidents of terrorist violence for 2014-15 from SATP. The database provides information on the number of terrorists killed, however, it does not give details on the origin of those killed or the number of terrorists captured. I went through media reports for each of the thirty incidents recorded for 2014 and 2015 to collect this additional data.
Two outcomes were apparent. First, there were more fatalities among foreign terrorists than local militants (1.5:1). The proportion of foreign cadre fatalities’ was higher even if we accounted for their bigger share among the number of active militants. Second, not even a single militant was captured alive across these incidents. If anything, the security forces seem to have been following a zero tolerance approach towards terrorists, irrespective of their origin.
My analysis could however be biased if SATP only reported incidents where at least one fatality was observed. In that case the dataset would have excluded those incidents where a terrorist was captured but without involving any fatality.
To alleviate this concern I compare the series’ on foreign mercenaries killed and arrested with terrorist arrested, surrendered and killed in J&K. As per my hypothesis the ratio of foreign terrorists Captured to Killed should at least be higher than that of the aggregate. A comparison of the two series is presented in Graph 1.
The ratio of foreign terrorists Captured to Killed is far lower than the aggregate. Interestingly, the ratio of Captured to Killed (Total) also fell sharply as the share of foreign militants grew from mid 90s onwards. In other words, the security forces were much more ruthless against the foreign terrorists than their local counterparts in the nineties.
There is also a rationale in following such strategy. Killing a local militant further complicates the law and order situation, as apparent from the organization of mass protests which invariably accompany the dead militant’s funeral. The contrasting response in Kashmir valley to the hanging of Afzal Guru, a Kashmiri militant implicated in 2002 Parliament attacks, and Ajmal Kasab serves as a useful example. Kashmir valley erupted in protest in response to Afzal Guru’s hanging, leaving 3 people dead and 50 injured in its wake. Kasab’s hanging, on the other hand, barely evoked any response. In other words, the state does not have to worry about law and order disruptions if a foreign militant is eliminated.
I now turn our attention to the pattern of terrorist violence. Specifically, I look at the trend of Civilian to Armed Force casualties during terrorist attacks, as shown in Graph 2. For a better understanding I also compare the trend in J&K to that of terrorist attacks across India.
Three patterns emerge. First, more civilians on average were killed than security personnel in J&K. Throughout the Kashmir conflict civilians have been targeted specifically as well have ended up as collateral damage. According to the Human Rights Watch, “Throughout the conflict, militant organizations in Kashmir have committed grave abuses. The most serious of these have been the murders of hundreds of civilians, both Muslim and Hindu, who have been targeted because of their suspected support for the Indian government, or because they otherwise opposed the policies or practices of one or another of the militant groups”. In some instances civilian deaths have been recorded during ambushes on armed forces. For example, recently a handicapped civilian was killed when terrorists attacked a police post in Baramulla district of J&K.
Second, the ratio of Civilian to Security Forces casualties peaked in 2005 and has seen a steady decline since. The ratio also fell below one from 2009 onwards. Finally, the trend in J&K closely followed that of all India average till 2007 when it began to diverge. The terror strikes in J&K seem to have more precision than the Naxal attacks, the other major source of terrorist violence in India.
Overall we can say that terrorist attacks in J&K have caused significant Civilian to Armed Forces casualties, although the trend has improved in the last few years. Even when civilians were not targeted directly, the collateral damage has been too high to consider the specific targeting of security forces as a credible strategy.
This article had set out to understand the pattern of terrorism related violence in J&K, and how security respond. We can draw two main conclusions from my analysis. First, security forces have been more ruthless towards terrorists of foreign origin. This seems to be a low cost strategy in comparison to eliminating a local militant. Second, the civilian casualties have been significantly high to argue for terrorist attacks being specifically targeted against security personnel.
 “I came to kill Indians, it’s fun: captured Pak militant Naved”, Hindustan Times (08.08.15)
 “Pulwama encounter: Let militant killed, Udhampur attack plotter may be holed up”, The Tribune (07.08.15)
 “Udhampur attack: Indian media is pushing for militarism instead of dialogue in Kashmir”, Scroll.in (08.08.15)
 “Pakistani lie nailed: Ajmal Kasab was a Pakistani, admits ex-FIA chief”, Znews (04.08.15)
 “Pakistan acknowledges role in Mumbai attacks, arrests main suspects”, The World Post (15.03.09)
 “Udhampur attack: Indian media is pushing for militarism instead of dialogue in Kashmir”, Scroll.in (08.08.15
 2015 data was updated till 9th August
 “Police Census: 104 militants active in Kashmir”, The Tribune (23.02.14)
 The two series’ are only available for 90s and hence I only use them for robustness check.
 “Guns ‘n’ poses: The new crop of militants in Kashmir”, The Indian Express (26.07.15)
 For example, many Kashmiris assembled and protested against Indian govt. at the funeral of Talib Hussein Shah
 “The hanging on Afzal Guru: How and execution is roiling Kashmir”, Time (15.02.13)
“Kasab’s hanging evokes no response in Jammu and Kashmir”, Times of India (22.11.12)
 “Behind the Kashmir conflict: Abuses by Indian security forces and militant groups continue”, HRW Reports (1999)
 “Civilian in wheelchair, cop killed in terror attack in Kashmir’s Sopore”, NDTV (18.08.15)
Partisan politics once again raised its specter in the aftermath of Yakub Memon’s hanging. As Pratap Bhanu Mehta alluded, a pitched battle is being fought over the sanctity of institutions. The danger then lies in “taking away from them a presumptive legitimacy will leave us unprotected in every respect”.
The Indian media has scrambled to play the vigilante, led by Indian Express’s first page news headline. The urge to fight the partisan battle is encouraging a race to the bottom of intellectual rigour. As a relatively serious researcher I balked at the evidence presented by Shoaib Daniyal (Scroll, August 1st 2015) against death penalty acting as a deterrence to crime. Using case studies from United States the author concluded that “there is a consensus on the redundancy of the death penalty in deterring crime.” Even ignoring the adage of one size does not fit all i.e. there may be structural differences in the experience of United States and India, I have to point out serious methodological deficiencies in the author’s claim.
1. Correlation is not causation
The author contends that the correlation between homicide rates and death penalty should be the holy grail for its deterrence effect. Correlation, however, does not imply causation and at best acts as an eyeball test. Correlation between two variables are quite often spurious and driven by other explanatory variable (s). We really need empirical evidence on causal effect of death penalty on homicide rates to arrive at any conclusion.
2. Literature evolves
The author uses one empirical study published in 1950s as evidence against any correlation between rates of homicide and death penalty. While I don’t expect the writer to have presented a thorough literature review, a look at recent empirical evidence would have been instructive. For example Dezhbakhsh and Shephard (2006) in a panel study of 50 states show almost 23 percent increase in murder rates during the period when United States Supreme Court imposed a moratorium on death penalty (1972-76), compared to the years before and after. The authors also find a significant variation across states in murder rates’ response to death penalty.
3. Brutalization effect does not capture state characteristics
The writer refers to Brutalization effect i.e. increase in murder rates with the death penalty as a “compelling” evidence against death penalty. He however does not account for two key issues- the problem of reverse causality i.e. states may impose death penalty in response to high homicide rates, which may explain why states without death penalty have lower murder rates than states with a death penalty. Second, there may be other institutional factors explaining the fall in death rates in North Carolina or California, after the removal of death penalty. For example Levitt (2001) in his seminal paper showed that legalized abortion in 1970s accounted to about 50 percent drop in crime in the 90s. In other words the association between death penalty and lowering of murder rates may be due to a third explanator.
The existing empirical evidence, as admitted by the writer, points towards some deterrence effect of death penalty in the United States. Majority of the empirical literature in the past decade show that death penalty has a deterrent effect. For example, out of 22 empirical studies on death penalty 14 report some deterrence, seven show no effect while one is inconclusive. One could debate on the efficacy of death penalty while factoring in moral considerations. For example, if an additional death sentence causes two fewer homicides we can debate on its utility to the society. However, to put together shoddy argumentation as critical scrutiny is a blatant disregard of the reader’s intellectual capacity.
 Dezhbakhsh, Hashem, and Joanna Shepherd, "The Deterrent Effect of Capital Punishment: Evidence from a Judicial Experiment," Economic Inquiry 14 (2006): 512-535.
 John J. Donohue & Steven D. Levitt, 2001. "The Impact Of Legalized Abortion On Crime," The Quarterly Journal of Economics, MIT Press, vol. 116(2), pages 379-420, May.
 Hashem Dezhbakhsh & Paul Rubin, 2007. "From the "Econometrics of Capital Punishment" to the "Capital Punishment" of Econometrics: On the Use and Abuse of Sensitivity Analysis," Emory Economics 0715, Department of Economics, Emory University (Atlanta).