This is a guest post by Christian Oswald, Melanie Sauter, Sigrid Weber, and Rob Williams. It is based on the article titled Under the Roof of Rebels: Civilian Targeting After Territorial Takeover in Sierra Leone, appearing in International Studies Quarterly.
How do rebels treat civilians after they take control of territory in civil wars? The study of rebel governance has gained traction in recent years, but often looks at this phenomenon from a static perspective. Empirically, civil wars are frequently characterized by changes in territorial control. In our research, we combine these two perspectives and look at civilian victimization immediately after rebels seize territory.
Conflict researchers usually assume that civilian victimization is most severe in areas of actively contested control: areas with ongoing battles between a government and rebels. These areas are seen as most violent for civilians because armed actors attack civilians to deter defection and to enforce compliance. Once one side is able to establish firm control over a given territory, scholars generally expect civilian suffering to become less intense.
Taking rebel governance into account, we look at how long this process of establishing firmer control takes. We argue that civilian victimization will remain high—or even rise immediately after a rebel territorial takeover—until a basic level of governance and civilian compliance is established. In newly captured territory, rebels likely have less local knowledge to identify who is on their side and who might challenge their rule. Local capacities to govern non-violently, for example through co-opting local dispute resolution mechanisms, might be very limited for rebel fighters that are not used to govern. We argue that this lack of knowledge and resources makes it more likely that rebels first intimidate the local population with increased civilian targeting until they can transition to more stable and peaceful control.
Rebel Violence Towards Civilians in Sierra Leone
To test our argument, we look at the Revolutionary United Front’s (RUF) actions during the civil war in Sierra Leone. The RUF controlled over half the country’s territory in 1998 and twice advanced into the capital of Freetown. This extremely violent rebel organization engaged in widespread and violent victimization of civilians. This included indiscriminate killings, abductions, and rape, which makes the RUF an extreme case of a non-state armed actor targeting civilians, in addition to the large areas it controlled during the conflict. Figure 1 shows different incidents of violence by geographic location and type of incident.
As the RUF committed many acts of violence against civilians and was not known for establishing governance institutions, we would expect levels of civilian victimization to remain consistently high, both before and after seizing territory, rather than peaking after territorial takeover and subsequently declining. Accordingly, we regard Sierra Leone as a good case to test our theory, because we would not expect to find support for it given the RUF’s patterns of behavior.
The Challenges of Analyzing Spatiotemporal Data
In order to understand patterns of civilian targeting by rebel groups, we need to look at individual incidents of violence, rather than aggregate trends. We do this by examining deliberate targeting of civilians by the RUF (between 1997 and 2001)—both lethal and nonlethal—to fully capture patterns of civilian victimization.
These types of event-level data offer many advantages to researchers because they list the geographic location and date of each incident, but they bring their own challenges. The biggest challenge we face is deciding how to aggregate individual attacks into geographic units as the results of statistical analyses can vary depending on this decision.
Using data from both UCDP GED and ACLED we apply matched wake analysis (MWA), a difference-in-differences design for spatiotemporal event data, to estimate the causal effect of territorial takeover on civilian victimization. MWA allows us to circumvent the problem of obtaining results due to the arbitrary choice of a particular space and time window by generating estimates from multiple different possible combinations. We also use matching to control for the effects of local factors such as distance to the nearest diamond mine, because diamond revenue was regarded as crucial in the Sierra Leone civil war.
Acquiring Control Can Mean More Violence
Our analysis points to a positive effect of rebel territorial takeover on subsequent civilian targeting after a battle between the government and the RUF. However, this effect is limited in its temporal and spatial scope to thirty-five days and up to four kilometers from the battle. Put differently, higher levels of civilian victimization occur following the assertion of rebel control over a contested territory, but only in the direct vicinity of the battles where rebels seized territory from the government. This fits our expectation that at locations where violent territorial transfer occurs, the interactions between rebels and civilians become even more violent.
Looking more closely at distance to diamond mines and the capital, it becomes clear that the amount of violence that accompanies rebel efforts to establish governance is a function of how far they are from the capital and how close they are to a diamond mine. In other words, rebel efforts to establish governance are more violent if the territory offers rents they can extract and if it is further away from the center of power, thus reducing the likelihood of government forces seeking to regain control quickly.
What Does This Mean for the Study of Civilian Violence in Civil Wars?
Given our extreme case selection, follow-up research should examine if other civil wars show similar violent transition periods after territorial takeover. Our research reveals that the establishment of rebel governance might be a violent process despite rebels’ long-term dependence on civilian support. Civilian life under newly established rebel governance might be more dangerous, but we have to study other cases of territorial takeovers to understand all mechanisms.
Our evidence for a violent transition to power also suggests that we should study rebel governance with a temporal perspective in mind. If rebel rulers and their behaviour towards civilians change over time, research on rebel governance should carefully compare how rebels rule during different stages of territorial control. When, and under what circumstances, this violent transition period turns into stable and nonviolent governance and the provision of public goods requires further research as well. This is also relevant for future research on patterns of civilian victimization.
