Lots of Buzz about Drones

The New York Times and the Wall Street Journal have been running a number of articles on the United States’ increasing reliance on drones to prosecute conflicts in Yemen, in Afghanistan, in Libya, and in general. The question of whether drone strikes help or hinder American interests has not yet been resolved, but it is likely to become even more relevant as Obama draws down American troops in Afghanistan and continues to authorize drone strikes in Libya against a sitting regime. Leaving aside questions of legality and morality for a moment, what do we know about the effectiveness of drone strikes?

Some research exists, but it is largely in the early stages. Patrick Johnston and Anoop Sarbahi have an interesting working paper on whether drone strikes in Pakistan have reduced terrorist attacks there. Here’s a graphic from their paper showing how drone strikes line up with terrorist attacks:

Using WITS data on terrorist attacks, they basically find that drone strikes between 2004 and 2010 have had a modest negative effect on the frequency and lethality of terrorist attacks in the Federally Administered Tribal Areas (FATA) in Pakistan, as well as a modest decline in suicide attacks and IED attacks. The study is rigorous and compelling.

One the surface, Johnston and Sarbahi’s piece suggests that drone strikes have worked so far in Pakistan. But there are three important caveats: (1) The study is limited to a single country over a relatively brief time span, when the drone program was just becoming a mainstay of the United States’ long-term overseas contingency operations strategy. It remains to be seen whether such trends play out in other contexts over time. (2) One never knows whether American drones accidentally killed more civilians than the terrorists would have: did the drone strikes ultimately hurt more Pakistanis than the deceased Al Qaeda affiliates would have if they survived? (3) We cannot know whether whether those civilians could have been spared if the United States (or Pakistan) had relied on human operations to neutralize these targets.

Many people bring up the objection that drone strikes aren’t really that precise, and civilians often get caught in the crossfire. In Pakistan, this has been an oft-cited source of anger directed at the United States, even though Pakistani military officials discreetly acknowledge that the drone strikes have helped reduce terrorism in Pakistan because they have dispatched high-profile Al Qaeda operatives. The New America Foundation keeps up-to-date information on drone strikes, locations, and fatalities (which you can access here). Take a look at this graphic from Peter Bergen and Katherine Tiedemann’s 2010 study, which is based on these data:

Through the beginning of 2010, a nontrivial number of victims of drone attacks were non-militant “others” (presumably civilians). 2008 was a particularly bad year. But reports indicate that the proportion of civilians killed by drones in Pakistan has declined in recent years, and if Johnston and Sarbahi are right, the regrettable civilian fatalities have not led more Pakistanis to retaliate against the United States than would have normally done so regardless. In fact, Christine Fair argues that many Pakistanis appreciate the drone program.

One skeptic is P.W. Singer, whose book Wired for War makes the claim that relying on drones undermines American interests in the long term. His Ted talk is worth watching:

Singer certainly seems dubious of the strategic effectiveness of the use of drones in or out of combat. Laying aside the problem of civilian casualties, he suggests that the use of drones simply angers insurgents or their constituents, who are offended that Americans “don’t want to fight us like real men.” Another important point is that the United States cannot maintain a monopoly on innovation for these war-fighting technologies, and that ultimately, the United States’ major military rivals may be the ones who perfect such technologies to the United States’ peril.

From the American military’s perspective, Singer expresses some concern that American troops have suffered increased rates of post-traumatic stress disorder because of the psychological disconnect between the “video game” platform and the experience of killing. He also cautions that because drones have lowered the costs of killing, Americans will be more cavalier about and quicker to support the use of deadly force abroad. Like his other points, this seems like an empirically testable proposition that political scientists might explore in the future.

But in the meantime, given that the Obama administration seems keen on using drones to manage terrorist suspects abroad, the field is wide open for researchers to collect more data to determine whether drone strikes actually do systematically decrease subsequent terrorist attacks in contexts like Yemen and Afghanistan. Whether drones are preferable to other types of air power in coercing state opponents like Qaddafi’s Libya is an entirely different question that also merits study. Are drones more buzz than bite? Some ambitious researcher should go find out.

One Response to Lots of Buzz about Drones

  1. Thomas June 23, 2011 at 8:40 am #

    I took a look at the Johnson/Sarbahi paper and found it interesting. They do a thorough econometric job and are rightly careful in interpreting their findings in light of selection bias (i.e. more terrorist activity may attract more drone strikes). I think they have a much bigger problem: data quality. I have talked to a few people familiar with the drone strike program who have seen or are familiar with the classified data drone strike data. It is my understanding that, while the Bergen and Tiedemann study does a thorough job of using “open source” news reports, it “misses” many strikes that went unreported. The U.S. does not announce its drone strikes and reports from the tribal region are notoriously sketchy. Moreover, there are seriously conflicting reports of the number of fatalities, who constitutes a high value targets, and whether that person actually died (see the Beitullah Mehsud death controversy http://en.wikipedia.org/wiki/Baitullah_Mehsud#Death_in_August_2009) in both sources used for their study (LWJ and NAF). Johnson/Sarbahi are asking important questions…and they should be commended for that. However until they acknowledge and find a way to plausibly model the censored nature of their drone strike data (I can’t really see how without making some pretty large assumptions) and take into account the measurement error in the fatalities/high value target data, I feel that the value added by their study is not much.