United Nations peacekeeping and terrorism: short-term risks and long-term benefits

ABSTRACT

Does the introduction of UN forces impact terrorism? We argue that at least initially, UN peacekeeping missions may significantly shift the local conflict bargaining process, creating incentives for terrorist and insurgent groups to increase their attacks against civilians. UN missions create a symbolic endpoint to initial negotiations, alter the balance of power between combatants, and may change the relationship between local combatants and the civilian population they rely on for support. We test this argument using monthly data from 12 African countries, analyzing the risk of terrorism at the local level. We find that the introduction of UN forces in an area significantly increases the short-term risk of terrorism, but longer missions in the country reduce this risk.

KEYWORDS:

• Civilian protection
• peacekeeping/peacebuilding
• terrorism
• United Nations

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ In many ways, these resolutions codified the strong civilian protection mandate first seen with the UNAMSIL mission. Authorized in 1999 and completing its mandate in 2005, UNAMSIL replaced an earlier observer mission in Sierra Leone with a significantly larger peacekeeping force. While the early mission was fairly weak and unsuccessful, the addition of well-trained British paratroopers helped stabilize the conflict and reduce violence against civilians (Hultman, Kathman, and Shannon 2013; Kreps 2010), making it a model for UN peacekeeping deployments since (Olonisakin 2008).

² From 2013 until 2016, the UN Mine Action Service (UNMAS) recorded 279 improvised explosive device (IEDs) attacks with 119 fatalities; over half have been MINUSMA personnel (Karlsrud 2017, 1219).

³ Cilliers (2003) argue that in Africa, peacekeepers and other humanitarian workers are “numerous, prominent, and vulnerable” (99). While peacekeepers are mostly military personnel, typically seen as “hard” targets and thus more difficult targets for terrorists, UN missions are comprised of multiple militaries, each with different equipment, training, and communications gear (Spearin 2011). Add potential language barriers and command structure differences, and this inherent coordination challenge offers security openings which may be exploited by terrorists.

⁴ While assassinations of business and religious leaders are most likely to result in regime change in all countries, in authoritarian states, regime change was only likely after the targeting of UN officials (Bell 2017, 167).

⁵ This logic runs counter to Li (2005), who argues that democracy has differing effects, with participation decreasing terrorism and executive constraints increasing its presence. However, Li’s study focused on transnational terrorism, when Young and Findley (2011) replicated Li’s study for domestic terrorism, the only democracy variable they found to be significant was executive constraints, which resulted in greater levels of terrorism.

⁶ See, for instance, Collier and Hoeffler (2004), Lujala (2009), and Ross (2004).

⁷ These correspond to the concepts of inside, outside, and latent spoilers (see Greenhill and Major 2007; Stedman 1997).

⁸ However, Wood and Kathman (2014) find that groups who use a moderate level of civilian targeting are more likely to reach a negotiated settlement with the government.

⁹ Elections are one such event, as groups may carry out civilian attacks to gain a political advantage (Autesserre 2009; Hafner-Burton, Hyde, and Jablonski 2013; Klopp and Zuern 2007).

¹⁰ UN peacekeeping missions extend the duration of ceasefires (Doyle and Sambanis 2000; Fortna 2004, 2008), prevent conflict reemergence (Gilligan and Sergenti 2008; Hultman, Kathman, and Shannon 2016), reduce battlefield deaths (Hultman, Kathman, and Shannon 2014), and increase the success of democratization following conflict (Doyle and Sambanis 2000, 2006).

¹¹ While focused on military interventions rather than UN peacekeeping, Kilcullen (2009) highlights how development projects which engage local populations, such as road construction in Afghanistan, can disrupt terrorist operations.

¹² Between January and October of 2015, at least 39 civilians were killed by SPLA soldiers and other armed groups around the Bentiu PoC camp (Arensen 2016, 32).

¹³ Past works find that larger UN missions comprised primarily of armed military or police forces are effective at protecting civilians from government and rebel attacks (Fjelde, Hultman, and Nilsson 2019; Hultman, Kathman, and Shannon 2013; Kathman and Wood 2016).

¹⁴ For a more extensive discussion of how credible commitment problems prevent the resolution of conflict, see Walter (2002).

¹⁵ The prevalence and diversity of conflict essentially controls “for several common correlates of victimization” and avoids “a sample of wars that would meaningfully bias our results” (Hultman, Kathman, and Shannon 2016, 238).

¹⁶ This dataset divides the globe into a series of cells of roughly 55 km per side.

¹⁷ This corresponds to a 1, 2, or 3 on the GTD variable specificity. We also present models in our online appendix relaxing this level of specificity.

¹⁸ We exclude attacks against entities like the United Nations.

¹⁹ As Hansen, Nemeth, and Mauslein (forthcoming, 8) point out, “while country-level studies essentially have the benefit of “aggregating up” events across a territory, leading to a higher variation in count, a more local analysis is dominated by essentially a dichotomous process.”

²⁰ To compare, all attacks has 957 cell-months with at least one terrorist attack, soft targets has 617 cell-months, and civilians has 384 cell-months.

²¹ We disaggregate our expansion variable to test the impact of specific expansion activities; these results can be found in the online appendix.

²² See Appendix Table A2 for a full list of activities we use to code the expansion and duration variables.

²³ This corresponds to location codes of 2 or 3.

²⁴ Similar to Kalyvas’s warning for civil war research, terrorism has an urban reporting bias (see Nemeth and Mauslein 2019). However, terrorism may be an urban form of violence. Urban environments offer high impact targets: cities are regional centers of population, economics, media, and infrastructure (Savitch 2008), while capitals contain a range of symbolic and valuable targets (Berrebi and Lakdawalla 2007; Nemeth, Mauslein, and Stapley 2014).

²⁵ As an important caveat, these studies examined African conflicts only. In his analysis of Bosnia, Costalli (2013) fails to find evidence that UN peacekeepers reduce violence during fighting, but did note they aid post-conflict stabilization.

²⁶ We also ran additional analyses in our online appendix using Fjelde, Hultman, and Nilsson (2019) local UN deployment size data.

²⁷ See online appendix.

²⁸ Models with a six-month time frame yield similar results (see online appendix).

²⁹ We used one-to-one nearest-neighbor matching with replacement.

³⁰ To test the predictive power of our model, we follow Ward, Greenhill, and Bakke (2010) and produce an ROC plot comparing models with and without our independent variables using the matched sample. The area under the curve of our model is larger than the baseline model (≥² = 8.74, p < .01). See Appendix Figure A1.

³¹ The exceptions were Models 1 and 2 (all attacks), where country duration was just outside of standard levels of significance.

³² A decade after the Rwanda mission, New York Times columnist Nicholas Kristof gave a scathing critique of the organization, stating that we should “ … drop any fantasy the U.N. is going to save the day as a genocide unfolds. In that mission, the U.N. is failing about as badly as the League of Nations did” (2006, 25).

³³ This link between larger observer missions and increased violence mirrors work by Hultman, Kathman, and Shannon (2013) and Kathman and Wood (2016).

³⁴ As Lyall and Wilson (2009, 96–97) state, this unit “sought to intimidate the local population and flush out insurgents by projecting ‘presence’ through armored shows of force.”