Plausible deniability? An investigation of government and government-outsourced violence in refugee hosting areas

ABSTRACT

This paper examines the propensity for host governments and the groups they sponsor to engage in violence in areas that host refugee populations. Drawing on arguments that governments strategically delegate violence to affiliated groups for “plausible deniability” purposes, it argues that, due to concerns over self-settled refugees’ welfare burden as well as the concern that these refugees will choose to live in border areas that are more vulnerable to (or useful for) militant activity, host governments, and their proxies are likely to target violence in areas with more substantial refugee self-settlement. At the same time, it anticipates that host governments will “outsource” this violence to surrogate groups where sizable camp-settled populations are present, due to a heightened risk of suffering international audience costs. Findings from a large-N sample of countries in Africa provide some evidence of the hypothesized outsourcing effect. While the presence of sizable camps alongside large self-settled populations is associated with a reduction in the likelihood of violence by host governments, it significantly increases the likelihood of violence committed by host-aligned proxies.

Cet article examine la propension des gouvernements hôtes et des groupes qu’ils soutiennent à s’engager dans la violence dans les zones qui accueillent des populations de réfugiés. Cet article s’appuie sur des arguments affirmant que des gouvernements délèguent stratégiquement la violence à des groupes affiliés à des fins de « possibilité de déni plausible » (Carey, Colarsi et Mitchell en 2015; Salehyan en 2010) pour soutenir qu’en raison des préoccupations relatives à la charge liée au bien-être des réfugiés qui s’installent eux-mêmes, ainsi que de la préoccupation liée au fait que ces réfugiés choisiront de vivre dans des zones frontalières qui sont plus vulnérables (ou utiles) aux activités militantes, des gouvernements hôtes et leurs mandataires sont susceptibles de concentrer la violence sur les zones où les réfugiés qui se sont installés eux-mêmes sont substantiellement plus élevés. Dans le même temps, cet article prévoit que ces gouvernements hôtes « externaliseront » cette violence en ayant recours à des groupes subsidiaires dans les zones où des populations importantes de réfugiés sont installées dans des camps, car le risque d’en subir les coûts de l’audience internationale serait plus grand. Les résultats acquis auprès d’un grand échantillon N de pays africains fournissent certaines preuves de l’effet supposé de cette externalization. Bien que la présence de grands camps aux côtés d’importantes populations de réfugiés qui se sont installées elles-mêmes soit associée à une réduction de la probabilité de violence des gouvernements hôtes, elle accroît considérablement celle de violences commises par des mandataires en accord avec les gouvernements hôtes.

Este artículo examina la tendencia de los gobiernos anfitriones y los grupos que estos acogen a participar en actos de violencia en áreas que albergan poblaciones de refugiados. A partir de los argumentos de que los gobiernos delegan estratégicamente la violencia a grupos afiliados con fines de “negación plausible” (Carey, Colaresi, and Mitchell 2015; Salehyan 2010), se alega que, debido a las preocupaciones sobre la problemática del bienestar de los refugiados autoasentados, así como la preocupación de que estos refugiados elegirán vivir en áreas fronterizas que son más vulnerables a (o útiles para) la actividad militar, es probable que los gobiernos anfitriones y sus representantes dirijan la violencia hacia áreas con un autoasentamiento de refugiados más importante. Al su vez, se prevé que los gobiernos anfitriones “tercerizarán” esta violencia para reemplazar a los grupos que se encuentran en campamentos militares de gran tamaño, debido al mayor riesgo de sufrir costos en la esfera internacional. Los hallazgos de una muestra “N” grande de países de África proporcionan cierta evidencia del efecto hipotético de la tercerización. Si bien la presencia de campamentos militares de gran tamaño junto a grandes poblaciones autoasentadas se asocia con menos probabilidades de violencia por parte de los gobiernos anfitriones, aumenta significativamente la probabilidad de violencia cometida por representantes alineados con el anfitrión.

KEYWORDS:

• Refugees
• violence against civilians
• pro-government militias
• armed groups

Notes

¹ “UNHCR Dismayed by New Deportation of Eritreans by Sudan.” UNHCR, October 18, 2011. Accessed April 2, 2020. Available at: https://www.unhcr.org/news/briefing/2011/10/4e9d47269/unhcr dismay-new-deportation-eritreans-sudan.html.

² Frushone, Joel. 2002. “The Forced Repatriation of Congolese Refugees Living in Rwanda.” U.S. Committee for Refugees, December 16. Accessed April 2, 2020. Available at: https://reliefweb.int/report/democratic-republic-congo/forced-repatriation-congolese-refugees-living-rwanda.

³ Pro-government militias are organized armed groups that align with governments, but are not considered part of the government’s conventional armed forces (Carey , Mitchell, and Lowe 2013, 250).

⁴ Böhmelt and Clayton (2018, 202) define auxiliary groups as “armed groups somewhat aligned with the state to perform specific security roles, but they are not part of the regular (i.e., army, navy, and air force) military’s command-and-control chain” (202).

⁵ Farah, Douglas. 2000. “For Refugees, Hazardous Haven in Guinea.” The Washington Post, November 6. Accessed March 27, 2020. Available at: https://www.washingtonpost.com/archive/politics/2000/11/06/for-refugees-hazardous-haven-in-guinea/f5b0d504-365f-44e2-b0fa-6c4984d301da/?utm_term=.0b775b3c6ba2. Accessed March 27, 2020.

⁶ “Kenya: End Abusive Round-Ups.” Human Rights Watch, May 12, 2014. Accessed March 27, 2020. Available from: https://www.hrw.org/news/2014/05/12/kenya-end-abusive-round-ups. Accessed March 27, 2020.

⁷ I’m grateful to an anonymous reviewer for pointing this out.

⁸ See the Appendix for the full list of countries included in the sample.

⁹ Theoretically, I expect that both refugee and local host civilians are vulnerable to civilian abuse where refugees are self-settled since these refugees “blend in” with local populations under those circumstances, as opposed to being spatially segregated in camps. At any rate, the GED does not include information on the identities of those targeted with OSV, so I am unable to differentiate between refugee and host community victims in the analyses.

¹⁰ Accordingly, UNHCR guidelines stipulate that there should be at least one official health center and at least one official food distribution site per 20,000 camp-settled refugees. See the UNHCR Emergency Handbook, available at https://emergency.unhcr.org.

¹¹ The full description of these data on refugees, including a discussion of data quality, is available at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/25952.

¹² To note, the correlation between “large camp” and “self-settled” is 0.07.

¹³ I thank an anonymous reviewer for pointing this out.

¹⁴ The road density and electric grid indicators both come from the x-Sub Dataset (Zhukov, Davenport, and Kostyuk 2019).

¹⁵ For locations with no large refugee encampment, a one-unit increase in self-settled refugees yields a change in log odds of experiencing one-sided violence of.26 and.05 for rival and autonomous PGMs, respectively. For locations that do have a large refugee encampment, in contrast, a one-unit increase in self-settled is associated with a change in log odds of.65 for rival PGM OSV and.52 for autonomous PGM OSV. The combined effect of hosting a self-settled population and a large refugee camp on opportunist and non-ethnic PGM OSV is not statistically significant.

¹⁶ In an attempt to triangulate evidence, I compared Gineste and Savun (2019) Political and Societal Violence by and Against Refugees (POSVAR) data to the OSV data used in the present study. Seventeen countries with locations that both hosted refugees and experienced host government or host proxy perpetrated one-sided violence are shown in POSVAR to have experienced direct anti-refugee violence by governments/proxies during the time period under observation (2000–2010). This adds confidence that there was anti-refugee violence observed in the countries in which the current analyses identify violence in refugee areas.

¹⁷ As others recognize, refugee encampment is not necessarily a desirable alternative to self-settlement. Aside from placing severe limitations on refugees’ rights, encampment often leads to its own security challenges, including widespread violence against women and children, human trafficking, and high crime rates (Jacobsen 2001, 13). Encampment has also been found to be associated with a higher prevalence of communal conflict (Fisk 2019).