Drone strikes increase terrorism and anti-us sentiment in Pakistan
Rafat Mahmood (University of Western Australia and Pakistan Institute of Development Economics), Michael Jetter (University of Western Australia, IZA and CESifo) APRIL 2019 IZA Institute of Labor economics, “Military Intervention via Drone Strikes” Accessed 7/2/2019 *mw*
This paper introduces an empirical strategy to isolate the causal effects of drone strikes in Pakistan on subsequent terrorism, anti-US sentiment, and radicalization, employing wind as the key IV. We hypothesize that wind decreases the likelihood of the US military employing a drone strike, conditional on observable characteristics, whereas wind is otherwise orthogonal to terrorist activities. Both assumptions receive support in our sample of 4,018 days from 2006 to 2016. Results from 2SLS estimations suggest drone strikes increase the number of terror attacks in Pakistan in the upcoming days and weeks. This finding prevails in a host of alternative estimations and robustness checks. Extending the timeframe of subsequent terrorism, we find evidence indicating drone strikes do not just affect the timing of attacks (e.g., by moving forward planned attacks) but rather increase the total number of attacks. In terms of magnitude, one drone strike today causes over four additional terror attacks per day in the next seven days which implies drone strikes are responsible for 16 percent of all terror attacks in Pakistan. A back-of-the-envelope calculation suggests 2,964 people died from terror attacks because of drone strikes. We then explore mechanisms, distinguishing between insiders, i.e., those who already belong to terrorist organizations, and outsiders, i.e., regular Pakistanis. Specifically, we study anti-US sentiment in the major English-language newspaper in Pakistan, anti-US protests, and online searches for terms that may be indicative of radicalization (jihad, Taliban video, and Zarb-e-Momin). In line with the blowback hypothesis, results from 2SLS estimations suggest the general populace increasingly turns to anti-US and radical expressions after drone strikes as all these measures rise substantially because of drone strikes. It is important to put the results pertaining to Pakistani news and Google search behavior in context. We are not suggesting Google Trends as the perfect yardstick to measure radical attitudes – an online search for a radical term does not make a terrorist. Further, identifying more negative emotions and anger 33 in US-related articles does not necessarily prove anti-US attitudes. For instance, articles mentioning the US may systematically apply negative language to their enemies. However, the persistency with which we identify signs of radicalization and anti-US sentiment because of drone strikes in the general Pakistani populace is consistent with the hypothesis that drone strikes systematically turn Pakistanis toward radical groups and against the US. In fact, given a literacy rate of 58 percent (Government of Pakistan, 2017) and the hypothesis that the tendency to radicalize usually decreases with education in Pakistan (Fair et al., 2014), studying an English-language newspaper and online search behavior (requiring literacy and internet access) may actually present a lower bound estimate of anti-US sentiment. To our knowledge, this is the first empirical analysis that is able to isolate causal effects of drone strikes. Contrary to the current opinion in the US military which suggests drone strikes curb terrorism, we find evidence to the contrary: Drone strikes (i) lead to more terrorism, (ii) make the US more unpopular in Pakistan, and (iii)steer Pakistanis toward radical ideas. In other words, not only are insidersretaliating against the US but outsiders appear to change their attitudes. As a consequence, the pool of militants may grow, if anything. As the US military continues to build and expand its drone program (e.g., in Yemen), we hope our research provides useful insights into the underlying consequences.