People gather to look at burned vehicles at the site of a bomb explosion that rocked the busiest roundabout near the crowded Monday Market in the northeast Nigerian city of Maiduguri on July 1, 2014. (Stringer/AFP/Getty Images)

According to United Nations estimates, the Islamic State, formerly referred to as the Islamic State of Iraq and Syria (ISIS), has killed thousands of civilians – including children – in recent months. Many more civilians have been wounded and upward of half a million have been displaced by violence. The Islamic State is, of course, not alone in intentionally targeting civilians for violence. Groups such as the Lord’s Resistance Army in Uganda, Boko Haram in Nigeria and opposition fighters in South Sudan purposefully kill, maim and abduct thousands of civilians as they pursue their objectives, with disturbing frequency. Why? What motivates armed groups to commit atrocities against civilians?

Recent scholarship highlights a number of possible explanations for this bloodthirsty behavior, including a group’s depth of territorial control, access to lucrative resources, prewar political cleavages, and tactical and material losses. Most recent studies begin with the core assumption that such violence is inherently strategic. In other words, insurgent groups perpetuate anti-civilian violence as a means of achieving preferred conflict outcomes. But scholars have rarely assessed the impact of civilian targeting on a group’s ability to attain its strategic objectives, particularly its longer-term political goals.

Is civilian victimization in fact strategically rational?

To begin to answer this question, we conducted a systematic analysis of the relationship between civilian victimization and the likelihood of different conflict outcomes (such as victory or a negotiated settlement) for more than a hundred post-Cold War African insurgencies. The results, which were recently published in the British Journal of Political Science, strongly suggest that – at least up to a point – greater insurgent brutality actually increases the probability that the conflict ends in a negotiated settlement. Given that rebels are typically the weaker party in conflict bargaining, this finding would seem to suggest that rebels are more likely to achieve significant concessions from the government by increasing violence against the civilian population. Disturbingly, then, civilian victimization might be all too rational.

This might be because the costs associated with insurgent attacks on civilians are not evenly distributed: The costs of anti-civilian violence (and the disruptions and dissatisfaction it causes) fall more heavily on the state, even when the insurgents are responsible for the violence. A government’s successful counterinsurgency campaign is predicated on maintaining order and (re)establishing control over unstable areas. Conversely disorder, instability and the erosion of state control directly benefit insurgents. Thus, when insurgents rely on civilian victimization, they impose significant costs on the state while absorbing relatively fewer costs themselves. In addition, by attacking civilians, rebels send credible signals to the state about their willingness to continue fighting a long, costly and brutal war. Faced with asymmetric costs and this signal of a long and costly war to come, governments are increasingly likely to make concessions as groups increasingly resort to atrocious tactics.

Beyond a certain point, however, brutality undermines the bargaining process. Attacks on civilians can effectively shift bargaining power away from the government and toward insurgents. Initially, positive shifts in rebel bargaining power increase the likelihood that the group achieves policy concessions from the state. However, beyond a certain point, these shifts may embolden the rebels, encouraging them to reject government concessions as their belief in their own future victory increases. Our results thus imply a curvilinear relationship between civilian targeting and the probability of a negotiated settlement wherein violence initially increases and then diminishes the odds of successful settlement.

The inflection point appears to be at the rate of approximately 400 deaths per month. Below that point, civilian victimization increases the likelihood that the conflict ends in formal government concessions to the rebel group. Beyond that point, however, the likelihood of significant concessions declines. Further, there is at least preliminary evidence that the declining likelihood of negotiated settlement results from a corresponding increase in the odds of rebel victory. This result underscores the argument that violence can pay perverse dividends to those groups that employ it.

These findings are broadly consistent with Jakana Thomas’s recent study, also discussed on the Monkey Cage, which examined the impact of terrorism on insurgent-government bargaining. She finds that violent groups that more frequently relied on terrorism (a phenomenon closely related to civilian victimization) were more likely to receive policy concessions from the government. Other studies, however, have arrived at contrary conclusions regarding the role of terrorism in assisting groups in achieving their core political goals. For instance, a recent study by Max Abrahms finds that radical groups that employed terror tactics against civilian targets rather than more traditional guerrilla tactics were less likely to achieve major political goals. The difference in these findings may come from the use of different definitions of violence, different study samples, comparing different types of actors or employing different definitions of “success.” For example, our study examined all organized insurgent groups in post-Cold War Africa and evaluated the impact of variations in intentional civilian deaths on the manner in which the war terminated (rebel victory, rebel loss and peace agreements). The Abrahms study, by contrast, examined a sample of foreign terrorist organizations designated by the U.S. State Department, codes groups according to their primary target selection (civilians versus other targets), and relied on a subjective coding of success based on the extent to which the group achieved its primary political objectives. Key differences in these items are likely to lead to different conclusions, and scholars and policymakers should be careful to understand the relevance of these differences.

Nonetheless, this is a healthy debate. The development of this research agenda is important for both scholars and policymakers struggling to find ways to avert mass civilian deaths and suffering. First, understanding the general influence of civilian targeting on conflict bargaining and war outcomes should shed light on the motives for their use. For instance, the findings from our analyses support a strategic view of victimization by providing evidence that violence helps groups achieve their military or political objectives. While we believe that factors other than the desire to achieve strategic goals motivate civilian targeting, our results and other existing studies strongly suggest that it is at least a frequent motivator of such violence.

Second, understanding the perceived incentive structures of violent groups may contribute to the construction of policies that can dampen the costs of war to the civilian population. Particularly, if civilian targeting is explicitly intended to shape the bargaining process, policymakers should be aware that an unwillingness to negotiate with “terrorists and thugs” might increase the frequency of attacks on civilian targets rather than deter future attacks. Of course, policymakers often fear that concessions to terrorists will simply embolden future violence. However, our results suggest that regardless of bold declarations to the contrary, governments are already bargaining with the worst types of rebels. Moreover, these groups are likely to extract more concessions the more violent they become.

Reed M. Wood is an assistant professor in the School of Politics and Global Studies at Arizona State University. Jacob D. Kathman is an associate professor in the Department of Political Science at the University at Buffalo, The State University of New York.