Incentives to kill: The organizational roots of onesided violence, G Schneider

Tags: civilians, article, authoritarian rule, civilian victimization, rebel groups, civilian population, principal agent model, formal model, explanations, armed groups, Eastern Congo, University of Konstanz, European Union Politics, Department of Politics & Management, military fatigues, Thomas Elbert, Deterrent Effect, University of Nottingham, Lord Resistance Army, conventional military, Neil Mitchell, Criminal activities, Organizational Roots, political violence, outcome variable, the military commander, political groups, economic incentives, principal agent, European Commission, interviews, activities, economic goal, Jeremy M., Gerald Schneider, Violation of Human Rights, Paul Huth, International Peace Research Institute, Armed Conflict, Center for International Development and Conflict Management, rebel leaders, armed group, victimization, rebel movement, material incentives, Benjamin A., International Organization, Minorities at Risk Project, Joakim Karlsen, Cornell University Press, Banholzer, Gerald und Bussmann, Minorities at Risk, Journal of Peace Research, Collective Violence, Lilli Banholzer, Human Rights, Cambridge University Press
Content: Incentives to Kill: The Organizational Roots of One-Sided Violence Gerald Schneider1 Department of Politics & Management University of Konstanz This article examines the decision making dilemma of opportunistic commanders who can rely on conventional military violence, terror against the civilian population or a mixture of both instruments to advance a particular political or economic goal. I analyze this decision making problem within the framework of a multi-task principal agent model where the principal sees civilian abuse as a complement to conventional military actions, while the combatants have an incentive to specialize in the lower-cost activity of soldering. The model shows that armed groups will kill and maim civilians if commanders offer them a sufficiently high incentive to commit repulsive acts or if the leaders lower the level of inhibition through initiation rites or similar inducements. I rely on group and individual-level data to demonstrate that leaders rely on various selective incentives to lure more or less unwilling soldiers into the usage of one-sided violence. The longitudinal analysis of political groups in the Middle East and North Africa first evinces in line with the model that the bounty criminal activities offer increases the risk of one-sided violence, whereas insurgent activities are only linked to the usage of conventional military means. Interviews with former combatants in Eastern Congo moreover suggests that soldiers who were rewarded with drugs or who did not receive sufficient food were more likely to be active in units involved in one-sided violence. In contrast to some recent finding on civilian victimization, organizational hierarchy rather than anarchy is associated with an increased incidence of acts against the civilian population. Paper prepared for presentation at the SGIR 7th Pan-European Conference on International Relations, Stockholm (Sweden). September 9-12, 2009. A preliminary version of this paper has been presented at the CSWC workshop on the usage of principal agent models in studies of civil war, University of Nottingham, April 29-30, 2010. I would like to thank Sabine Carey, Scott Gates, Neil Mitchell, Gudrun Шstby and the other participants for their helpful comments and the German Foundation for Peace Research, the European Commission and the University of Konstanz for financial support. The data sets used for the analyses will be made available upon the publication of the article. 1 Professor of Political Science, Executive Editor "European Union Politics" and Research Affiliate Centre for the Study of Civil War, Oslo. Address: International Relations Chair; Department of Politics & Management; University of Konstanz, Box 86, 78457 Konstanz, Germany, Tel: +49-7531-88-2608/3081, Fax: -2774, [email protected]
1 Introduction When the Lord Resistance Army entered several villages in the north-eastern part of the Democratic Republic of Congo in December 2009, "the group sent in relatively clean-cut soldiers dressed in proper military fatigues" (Gettleman 2010). The remarkable outer appearance of the dreaded LRA soldateska did, however, not alter the deadly mission of the fighters who left the unprotected area with dozens of kidnapped villagers after killing several hundred civilians on the spot. Such acts of what Eck and Hultman (2007, see also Schneider & Bussmann 2010) call "one-sided violence" are hard to understand, especially in the light of the relatively minor looting possibilities that raids by small groups on isolated villages offer.1 It is therefore not surprising that many observers portray the notoriously murderous Lord Resistance Army, and other violent groups that are active in the Great Lakes region, as killers without purpose, all too often paying the seemingly obligatory tribute to Conradґs novel "Heart of Darkness" (2000[1902]). The difficulties in grasping the motives behind the cruelty become evident in one eyewitness's interpretation of the massacres. According to Father Joseph Nzala, "we donґt know what their strategy really is, but they clearly like killing, like destroying things" (Plauth 2010). In this article, I will argue in line with a growing body of literature that we can often trace the mass killings of the sort perpetrated by the LRA to cunning calculations by opportunistic military commanders. According to the multi-task principal agent model of onesided violence developed in this article, these leaders see in the killing and maiming of civilians a complement to conventional battlefield activities. In this perspective, the simultaneous usage of both instruments will be most effective to reach a certain goal. The fighter, by contrast, conceives of one-sided violence rather as a supplement to the soldering for which he was hired and trained for. He generally ­ but not always ­ prefers to engage into this sort of relatively low-cost activity instead of the slaughtering of civilians which he considers, at least in the beginning o the military career, to be repulsive.2 The analytical starting point for my formal and empirical analysis of the extent to which the structure of an armed organization causes one-sided violence is Collier's (2000) infamous metaphor that rebel leaders are often nothing else than "quasi-criminals". Indeed, there is ample anecdotal and statistical evidence to suggest that economic incentives in the form of resource abundance were a key cause of several recent civil wars despite the refutation of the more general greed thesis advanced by Collier and Hoeffler (2004) by Fearon (2005, 2008), Besley and Perrson (2008) and others.3 Esteban et al. (2010) suggests in a formalized extension of this macro-level argument that groups rely on mass killings to gain or maintain access to profitable economic resources. The micro-level variant of this argument advanced in this article suggests that there are organizational roots of one-sided violence in addition to the motives that arise from the strategic contest between competing ethnic or social groups over these natural resources. I point out that military leaders will offer in the absence of a natural resource rent other selective incentives to turn soldiers into the executioners of their command to move against the civilian population. My Clausewitzian argument suggests that opportunistic commanders and, by extension, their armed groups rely on various means of violence, ranging from torture to terror and eventually war, if the goals of the leader cannot be reached through coercion or 1 According to the definition of Schneider and Bussmann (2010), instances of one-sided violence are lethal or harmful acts that an organized group, which can be either a rebel organization or government actor, directs against unarmed individuals. The aggression results in the immediate physical harming or death of more than one person. I will use the concepts of "civilian victimization" and "one-sided violence" interchangeably in this article. 2 As most commanders, their henchman and soldiers are male, I will employ male pronouns and if I write about an individual actor. 3 Schneider (2010) offers a survey of this literature.
2 peaceful means alone. In this line of argument, the victimization of civilians is based at first sight on a simple opportunity cost calculation according to which potentially violent actors kill and harm unarmed individuals, if such acts are more rewarding in comparison to the usage of a more peaceful instrument. The potential benefits are for instance the gains from looting or the reduction of the costs that civilian opposition brings to the activities of the armed group. Yet, the division of labour that is typical for armed organizations renders these calculations more complicated as the soldiers face the risk to be prosecuted once the military campaign is over, while the link of the commanders to the atrocities has first to be painfully established in court or elsewhere. This asymmetry suggests that, in most instances, the preferences of the planner (the principal) and the executioner of the violent acts (the agent) differ. While the commanders often believe that a combination of military confrontations and of terror against civilians is effective for reaching a particular goal, soldiers have a preference in simple soldering as this activity is less risky because of the infrequent fighting in may conflicts and because of the small chance of public prosecution for purely military activities. Political science typically conceives of the choice between various instruments of violence as a reflection of power asymmetries. The realist school of thought contends that the militarily strong conflict party tries to face its enemy principally on the battlefield, while its weaker opponent considers guerrilla tactics or terror strategies. In some instances, it is nevertheless the militarily strong side which victimizes the civilians. This can either be a consequence of genocidal tendencies (Harff 2003) or of the impossibility to fight the weaker side with conventional means (Valentino et al. 2004). However, such explanations of mass killings and genocides do not consider why groups often use both conventional means and terror strategies simultaneously and why some groups completely specialize in one instrument of violence. The analysis of the Minorities at Risk Organizational Behavior (MAROB) dataset evinces that political groups have very different profiles with regard to these two ideal type forms of violence. This resource classifies 118 organizations that were acting on behalf of 22 ethnopolitical groups from 1980 to 2004 in the Middle East and North Africa (Minorities at Risk Project 2008). 38 of these organizations used violence against civilians occasionally throughout this period, eight of them frequently. Of the 879 organization years covered by the dataset, 236 of the instances involved one-sided violence, 131 organization years only insurgency tactics and in 50 organization years both activities occurred.4 An increasing number of studies contends in line with this descriptive evidence that the usage of violence against civilians and unarmed soldiers does not only result from power asymmetries, but also stems from the incentives that armed groups offer to commanders and soldiers. Resorting to the principal agent literature developed in industrial organization, this alternative view conceives of one-sided violence as the product of a lack of control by the commander or the predominant recruitment of soldiers whose primary motive is economic gain (e.g. Azam 2002, Weinstein 2005, 2007; Mitchell 2009). However, the existing empirical studies perceive terror strategies as largely inefficient, attributing the killing of civilians to the lack of control by the military commanders (Humphreys and Weinstein 2006a) or the recruitment of opportunistic killers instead of soldiers who are devoted to the long-term aspirations of the rebel organization (Weinstein 2005, 2007). In other words, most extant models conceive one-sided violence as either the exclusive goal of an organization or as the overall inefficient side-product of looting and related activities that are, by and large, detrimental to the political goals of the organization. The consequences of these theoretical assumptions are that the models portray the civilian victimization as either an end in itself or 4 The positive association with a gamma coefficient of 0.48 between insurgency and one-sided violence indeed suggests that there is tendency towards a contemporaneous usage of the two instruments of violence.
3 put the blame entirely on the plundering soldateska, downplaying the mischievous role that the planners of one-sided violence play. To overcome the limitations of the current literature, I use the multi-task principalagent model by MacDonald and Marx (2001) to identify some of the conditions under which armed groups will rely on one-sided violence. I particularly assume that soldiers and commanders disagree over the extent to which an armed group should specialize in terror strategies. One-sided violence in this model becomes more likely if a) the usage of one-sided violence renders the success of the overall military campaign more likely and b) if the incentives offered for conventional engagements loom not sufficiently large in the minds of the soldiers. To illustrate the empirical implications of the model, I rely on individual and grouplevel data. Both sorts of analyses strongly suggest that the often cynical incentives and disincentives that armed groups offer to their soldiers motivate the killing and maiming of civilians. I will in the following first discuss the theoretical and empirical literature in more detail, then move on to the sketch of a formal model and finally present the empirical evidence. The article concludes with a summary. Principal agent models of one-sided violence Most recent studies on one-sided violence explain civilian victimization as an outflow of the ups and downs on the battlefield (see Schneider et al. 2010). Interestingly, these accounts of mass killings place the blame for the atrocities on the planners of these acts through the assumption that one-sided violence is a deliberate choice. The principal agent explanations of civilian victimization share the rationality assumption of the strategic explanations. So far, these models, however, mainly account for the cruel acts of the soldiers, thereby largely ignoring the intentions and commands of the military or political leadership of the armed groups. The anti-soldier bias of the current principal agent models becomes obvious in a discussion of some key contributions; they either highlight the role of the recruitment process or the lack of hierarchy as causal mechanisms for one-sided violence. I will discuss the related contributions and some of their extensions in turn by distinguishing more broadly between incentive- and disincentive based principal agent explanations of civil war. Incentive logic: Weinsteinґs (2005, 2007) model of one-sided violence implicitly builds on Collier and Hoefflerґs (2004) distinction between "greed" and "grievance" as motives for joining armed groups and, by extension, as explanations of civil war onset. In their view, the abundance of natural resources to which "quasi-criminal" rebel leaders (Collier 2000) might gain access through the use of force is a root cause of internal wars, while political reasons such as inequality or discrimination play a minor role. Weinstein, by contrast, conceives of the presence of political and economic injustice as a barrier to civilian victimization as such grievances motivate individuals who can identify with the political aspirations of an oppressed group to join rebel movements. An abundance of material incentives, by contrast, draws soldiers who want to enrich themselves to a rebel movement. According to Weinstein (2005: 605) "...resource-rich rebel groups are overwhelmed by opportunistic joiners, while those waging war with limited economic endowments attract committed soldiers to their movements" (2005, p. 605). Drawing on a comparative case study of four armed groups (National Resistance Army, Renamo, Sendero Nacional, Sendero-Huallaga) and some broader econometric tests, Weinstein (2007: 309) concludes that "[r]esource wealth is associated with higher levels of violence". Weinstein indeed establishes a positive impact of contraband income of rebel organization on the number of combat-related deaths. Yet, in order to see whether the material incentives really exist for opportunistic joiners we would need micro-level evidence on the
4 level of compensation of soldiers who are active in resource-rich environments in comparison to resource-poor areas. This need is particularly urgent in view of the rich literature that links looting of basic needs such as food to the occurrence of violence. According to Azam (2002: 132), looting and violence against civilians are "the main activity of soldiers during a war". Obviously, looting often undermines the effort of the military commanders. As (Bazenguissa-Ganga 1999: :51) notes in his examination of violence in the civil war in Congo-Brazzaville, "when the fighters developed the habit of systematic looting, it was difficult to mount any sustained offensive". However, the inefficiency of economic plundering and the associated killing and maiming of civilians might simply indicate that soldiers often need to secure additional income and that the ensuing one-sided violence is an indication of this trend. As I argue further below, opportunistic leaders might manipulate such needs to turn simple soldiers into he killers and the tormentors of civilians. Disincentive explanations: The hierarchical organization that is typical for armed units can be designed both as barrier against the abuse of civilians and as a tool that facilitates commanders to lure soldiers into crimes against humanity. One mechanism that helps leaders to transform recruits and kidnapped soldiers into "killing machines" is to punish those who refuse to commit such acts and to reward the others who perpetrate atrocities. The training that soldiers receive ­ and the threats associated with it to move against non-conformers ­ is one instrument that helps commanders to lower the individual inhibition to kill and maim. In criminal organizations like the Mafia and in various armed units that were active since the end of the Cold War the first murder of another person often serves as an initiation rite for those the commanders have selected to become the key perpetrators of the organization. As one former child soldier active in the Sierra Leone slaughtering writes: "After that first week of going out on raids to kill people we deemed our rebel enemies or sympathizers of the rebels, our initiation was complete" (Beah 2007). The abuse of organizational hierarchy has, however, played a marginal role in recent attempts to explain one-sided violence. Most contributions rather attribute the civilian victimization to a lack of effective command-and-control structures. Liberals hope for instance that armies, which act on behalf of a democratic government, engage less frequently in one-sided violence than the units that autocracies send to military engagements. Unfortunately, the hope that the sanctioning mechanisms used in democracies deter soldiers from one-sided violence only receives limited empirical support. According to Downes (2006, 2007, 2008), democratic leaders rely on this instrument to shorten a war; Morrow (2007), by contrast, contends that democracies are the responding rather than the initiating actors in the usage of civilian victimization. As democratic norms do not necessarily reach down to the level of the active soldiers, we need to identify organizational structures within army units that possibly inhibit the fighters from using force against the unarmed. Many armies have adopted codes of conduct that seek to limit civilian abuse by soldiers; some of these rules follow from the Laws of War that the states in questions have ratified. However, the effectiveness of such norms is quite limited, as the empirical evaluation of Bussmann and Schneider (2010) with regard to the activities of the International RED CROSS suggests. It is therefore doubtful that armed groups really enforce the codes of conduct in the absence of a willingness to control the agents or an effective sanctioning mechanism. Moreover, most of the victimization of civilians stems from rebel groups which do not necessarily believe that the Laws of War encroach upon them (Eck and Hultman 2007, Schneider et al. 2010) We could therefore expect in line with Humphreys and Weinstein (2006a) that the increasing heterogeneity of an armed unit undermines possible sanctioning possibilities and increases the risk of one-sided violence. The author provides empirical support for this thesis in a cross-sectional analysis of violence perpetrated in Sierra Leone: "Fighting units composed of individuals motivated by private goals, with high levels
5 of ethnic diversity, and weak mechanisms to maintain internal discipline commit the highest levels of abuse" (ibid: 444). Humphreys and Weinstein offer in an unpublished appendix to their article a simple formal model in which the soldiers have the possibility to extract money from the civilians (Humphreys and Weinstein 2006b). This reduction of one-sided violence to an encounter between the fighters and the soldiers, however, excludes the possibility that commanders might, to some extent, also favour civilian victimization. Moreover, such models put a disproportional amount of the responsibility of the cruel acts on the executioners of the violence instead of the masterminds of the civilian victimization. Mitchellґs (2009) informal discussion of leadership types allows us to identify some of the roles that commanding officers and politicians play in the maiming and killing of civilians. Referring to the lack of intervention by Count Tilly, a commander in the Thirty Yearsґ War, in the massive rapes occurring during the massacre of Magdeburg, Mitchell sees one-sided violence as an interaction between a willingness to harm and the opportunities the soldiers have in raping women and children: "...while the presence of rape may result from either a principal who won't control or of a principal who can't control, the absence of rape reveals a principal in control" (ibid. 50).5 In this view, it takes most often two actors - the principal who orders the atrocities and the agent who commits them ­ to account for those mass killings that we cannot attribute to the lack of control argument. He argues that "agents of atrocities" become active once leaders see their power threatened or of truly evil politicians with a genocidal agenda are in command.6 The multi-task model outlined below develops this argument further through the relationship between an opportunistic commander and a group of soldiers whose low-cost activity is soldering in comparison to one-sided violence. I accordingly take issue with the anti-rebel and anti-soldier bias inherent in many contributions to the civil war literature and the implied inefficiency of one-sided violence during war. My analysis highlights that both rebel and government units have the possibility to rely on a wide array of "instruments" of political violence to reach their goals. As most violence in internal conflicts is directed against civilians and not against enemy soldiers, we need to build a model that motivates both conventional military and one-sided violence simultaneously. One-Sided Violence in a Model of Adverse Specialization The emerging literature on the organizational roots of one-sided violence has as indicated so far put the blame almost entirely onto the fighters and has completely neglected the evil role that the military leaders of armed movements can play. To fill this gap, I will introduce a baseline model which assumes an opportunistic military commander who believes that only a mixture of conventional military tactics and one-sided violence will result in a successful campaign. The study of one-sided violence becomes interesting from an organizational perspective only if the preference with regard to this instrument differs between the principal and the agent. In the following model, the commander sees this form of violence as part of a military campaign, while the fighters prefer simple soldering which might include the occasional small-scale battle. As conventional military activities are thus in many conflicts 5 Mitchell (2009) draws on the historical treatise The History of the Thirty Yearsґ War in Germany by German playwright Friedrich Schiller who also wrote Wallenstein, a trilogy of dramas, on a notorious war entrepreneur of that era. 6 My argument turns this logic upside down as the absence of rape and other forms of one-sided violence can in my view also follow from a lack of inducements that possibly turn fighters into the killers and tormentors of civilians. The main difference between my explanation and the one of Mitchell and other students of one-sided violence is that I do not conceive of such acts as necessarily ineffective but rather as one of many forms of violence that can be used in the pursuit of a particular goal.
6 less costly than the killing and maiming of civilians from the perspective of the individual fighter, he sees the latter instrument as a costly complement to the former one. Inversely, the military commander conceives of one-sided violence as a supplement. In his perspective, a military campaign is most effective if the soldiers commit time and effort to both conventional fighting and the killing and maiming of civilians.7 (to be extended) Empirical tests of the incentive-disincentive thesis This section offers two empirical tests, one at the group level and one at the individual level, of the adverse specialization hypothesis of one-sided violence. Both examinations lend support to the main thesis that the risk of one-sided violence grows with the relative unattractiveness of a classical military effort in comparison to terror strategies and that accordingly incentives and disincentives are key to understand the victimization of civilians. The analysis starts out with a longitudinal analysis of the risk of civilian victimization perpetrated by the groups that make part of the Minorities at Risk Organizational Behavior (MAROB) dataset. This data project of the Minorities at Risk Project (2008) allows us to analyse whether organizations used violence against civilians in a given period. I will compare two explanatory models against each other. The first one explains whether the group in a given year engaged into one-sided violence, the second one whether it used insurgent tactics. The one-sided violence measure is a dummy where the category 1 stands for the usage of attacks on civilians as minor or major strategy; the variable takes the value of zero if the group did not use this strategy. The empirical basis is the ORGST/ variable of the MAROB data set. The insurgency outcome variable, which refers to ORGST8 of the same source, is a dichotomous measure that amounts to 1 if a group used guerrilla tactics frequently or infrequently, zero otherwise. The explanatory variables, which are all lagged for one year, are also taken fro the MAROB data set. Criminal activities measures dichotomously, based on the variable CRIME, whether a group behaved criminally or not in a given year besides eventually participating in political violence. Advocacy authoritarian rule expresses whether a group spoke in favour of authoritarian forms of government (AUTHORG), while the variable government repression consists of three levels of state-sponsored violence against the organization, ranging from no violence over periodic to consistent lethal violence (STATEVIOLENCE). The two models also include past one-sided violence (Insurgency model) and past insurgent behaviour (Onesided violence model). Table 1 summarizes the results that I obtained through random effect longitudinal logistic regressions. The theoretical expectation is that criminal activities explain one-sided violence, but not guerrilla activities. Groups which pursue the ambition to set up authoritarian governments are more likely to engage into any sort of political violence as the lack of democratic norms goes hand in hand with an increasing tolerance for violent means as opportunistic leaders with dictatorial visions feel less inhibited than elites who consider to fight for an electoral office at some point. I furthermore expect, in line with the grievance literature, that insurgent behaviour is largely a response to government repression, while the same factor is of lesser importance for the explanation of one-sided violence. Table 1: Determining the risk of one-sided violence and insurgeny in the Middle East and North Africa (Random effect longitudinal logistic regressions) 7 Obviously, we could also model the opposite situation where a principled commander wishes to only use the soldier on the battle field, while the soldiers are thugs and prefer either a mixture of soldering and one-sided violence or only the usage of the latter instrument. In all circumstances, such a commander would under all circumstances not last long. I will discuss more realistic alternatives to this principal agent model in the conclusion of this article.
One-sided violence
Criminal activities t-1
Government repression t-1 2.35
Advocacy authoritarian rule t- 9.56
One-sided violence t-1
Insurgency t-1
Number of observations
Number of groups
Log likelihood
Note: Entries are odds ratios and standard errors in parantheses. xtlogit onesided
The evidence assembled in Table 1 clearly shows that various instruments of violence follow a different logic. While economic incentives like the ones stemming from criminal activities influence the risk of one-sided violence, they do not impact upon the probability that a group engages into insurgent behaviour. This effect of criminal activities on civilian victimization is quite large with an odds ratio of 5.45. Only the impact of advocacy authoritarian rule surpasses the one of criminal activities. Interestingly, the odds ratio of this concept is roughly the same across the two instruments of violence. This suggests that groups with an authoritarian ambition are more likely to rely on violence than an organization that rather opt for democratic forms of government, a relationship that is in line with the domestic politics variant of the democratic peace. The influence of this concept also suggests that leaders of an organization with authoritarian leanings do not discriminate much between the different instruments of power. Interestingly, past one-sided violence increases the risk of subsequent insurgent behaviour, while the opposite tendency does not hold and past guerrilla activities do not increase the risk of civilian victimization. In the following, I will supplement this group-level evidence in support of the specialization thesis with interview data from the province of South Kivu in the Democratic Republic of Congo.8 This former Belgian colony experienced at the turn of the millennium what Prunier (2009) describes as a "World War". In this conflict, one-sided violence was frequently used (Schneider et al. 2010). A research team of the University of Konstanz interviewed in the spring of 2009 96 members of different armed movements in the provincial capital of Bukavu and its surroundings. As some of the respondents were taking part in different groups, the overall sample of interviewed ex-combatants amounts to 139 cases. 66 of the interview partners belonged to various factions of the Maп Maп militia which was particularly active in this region. 22 former fighters were affiliated to the Rassemblement Congolais pour la Dйmocratie (RCD), and 14 respondents came from Les Forces Armйes de la Rйpublique Dйmocratique du Congo (FARDC). Further interview partners belonged at some point to the following organizations: L'Alliance des Forces Dйmocratiques pour la Libйration du Congo (11), Le Congrиs National pour la Dйfense du Peuple (9), Les Patriotes Rйsistants Congolais (8), Forces Dйmocratiques de Libйration du Rwanda/ Interahamwe militia (5). Three respondents had been members of other groups. The small number of FDLR/Interahamwe is not puzzling since the Congolese authorities sent many captured, freed or released fighters of
8 Details about these interviews can be found in the appendix to this article.
8 these movements back to neighbouring Rwanda. The average respondent was 21 years old. Not surprisingly, only five of the 96 interviewees were female; most of them had been sex slaves during their time in the armed groups. Most male respondents had combat experience. The dependent variable used indicates whether or not a respondent heard at any point of time the order that the group shall kill or harm civilians. Although this variable does not directly express the real involvement of an interview partner in the victimization of the unarmed, it at least expresses that the military unit to which he or she belonged took part in acts of one-sided violence. The logit models that I will present below cluster for the group membership of a respondent. The empirical models differentiate between four models. The incentive model contains drug reward and not enough food as explanatory concepts. The first variable refers to the question of whether or not a respondent received drugs as a reward. This dummy variable amounts to 1 if the interviewee received drugs in this capacity, 0 otherwise. The second independent variable used for this model yields the dichotomous responses to the question of whether or not a respondent received enough food. The motive model equally relies on a dummy variable as an explanatory concept. Revenge and politics is 1 if a respondent mentioned political motives or revenge as motives for joining the organization, 0 otherwise.9 Note that the number of soldiers who claimed to be politically motivated only amounts to 8 cases. None of these respondents had allegedly ever heard the command that civilians shall be attacked. The organizational model, finally, includes as a measure of hierarchy in the group in which an interviewee had been active the responses to the question of who carried out a punishment in the group. Hierarchy is 1 if the commander punished offenders, 0 otherwise. The theoretical expectation is for the incentive variables that they increase the risk of individual involvement in one-sided violence. The influence of political or revenge motivations is unclear. Kalyvas (2006) and others suggest that one-sided violence is often used as a means to settle old scores. The grievances that fighters mention as reasons to join an armed movement could lead to the same sort of ideological hatred that distinguishes between "us" and "them". However, these motives might have led the recruits to join an armed movement that fights the enemy conventionally and that does not target civilians. I therefore expect in line with Weinstein (2005, 2007) that fighters who mention political motives or revenge as the chief reason to join an armed movement are more principled and have a higher chance to withstand orders to kill or harm civilians. Humphreys and Weinstein (2006a) furthermore suggest that anarchy increases the risk of one-sided violence. Note, however, that their model does not address the role of the military commander who might see terror to be more useful in some circumstances.
Table 2:
The influence of personal attributes of former combatants in Congo on the involvement of their group in the harming and killing of civilians
Variable/model Drug Reward Not enough food Revenge and politics Hierarchy
Incentive 5.55 (2.12) 2.48 (0.86)
Motive 0.38 (0.25)
Organization 2.93
Overall model 5.51 (2.26) 2.04 (0.89) 0.41 0.34 2.57
9 All missing values for this variable received a coding of 0; this imputation covered 40 cases.
Log-likelihood -65.75
Corr. Classified 73.77%
Note: Entries are odds ratios and standard errors; robust variance estimates with standard errors adjusted for
membership of interviewed ex-combatants in one of eight armed groups.
The models classify the cases surprisingly correctly, given the rather crude measures that were used. The incentive model offers the largest improvement over the null model. Especially the drugs variable increases the odds that a fighter was involved in the killing and harming of civilians by a factor of five. The corresponding figures for the hierarchy and the food variables are associated with odds that are about half the size of the drugs indicator. However, the impact of these factors is still substantial and it suggests that many acts of onesided violence are indeed associated in the conflict in Eastern Congo with looting and that strong command and control lines rather than hierarchy are key to understanding why civilian victimization occurs. While the latter relationship offers single-conflict evidence in contradiction of the Humphrey and Weinstein soldateska-claim, we would need broader samples of cases to clearly show whether organizational anarchy or strong hierarchy are at the root of one-sided violence. The substantial effects that the incentive and disincentive explanations have on onesided violence are considerable. The baseline probability that an interviewee had heard of the order to victimize civilians is six per cent if the incentives did not exist and in the absence of hierarchy, but in the presence of revenge or political motives to join the organization. The risk of involvement in one-sided violence increases to 66 percent if the explanatory variables are reversed. The presence of the incentive variables and the absence of the motives alone increase the risk to 42 percent. Conclusion Civilian victimization is one if not the feature of recent wars. This article has tried to move beyond current principal agent models or other frameworks that try to explain this so-called one-sided violence. As the extant models only motivate this instrument of violence and therefore ignore the trade-off between military and terror strategies, I have employed the multitask principal agent framework by MacDonald and Marx (2001) to account for the bewildering diversity of armed movements which either specialize in one or in both activities simultaneously. If we assume the military commander to be interested in an encompassing effort, while the soldiers strive to specialize in the low-cost activity, a combination of military and terror strategies will only result from proper compensation for traditional soldering. In other words, the principal-agent perspective pioneered in studies of industrial organization and increasingly applied to the study of armed conflict also holds promise for the analysis of one-sided violence. I believe, however, that these models have to differentiate more properly between different instruments of conflict. The model suggests that appropriate rewards and punishment can prevent the adverse specialization in civilian victimization. The empirical evidence assembled in favour of this incentive/disincentive offers broad support of this thesis. While criminal groups are more likely to target civilians, a reduced risk of one-sided violence stems from politically motivated movements. The group-level evidence gathered in a comparative analysis of violent behaviour by groups active in Northern Africa and the Middle East finds its individual-level counterpart in the analysis of interviews that we have collected among ex-combatants in the province of South Kivu in Eastern Congo. The examination of this data sets shows that the risk that a combatant or his armed group was involved in violence against civilians clearly mirrors the
10 incentive and disincentives these fighters face during their stay in the armed group. While rewards in the form of drugs and the opportunity to loot for food are clearly associated with a higher risk of one-sided violence, military hierarchy also fosters this probability. This evidence contradicts the finding of Humphreys and Weinstein (2006) who have advanced a formal account of the traditional argument that organizational anarchy is a root cause of onesided violence. It is clear that alternative causal mechanisms could be found to uncover the organizational roots of one-sided violence. Armies of all types often delegate the task to attack civilians to militias, thereby minimizing the risk to become implicated. The decisionmaking problem of the commander who wants to have some terror done as part of the overall military campaign does, however, not look too different from the multi-task problem that I have outlined in this article. Many leaders have at some point delegated the task of attacking civilians to groups of henchmen who have a low inhibition to kill and maim the unarmed. Hitlerґs Schutzstaffel (SS) and its early competitor Sturmabteilung (SA) or Milosevicґs Srpska dobrovoljacka garda with its enigmatic leader Arkan (Zeljko Raznatovi) are exemplary cases of agencies which have recruited a disproportionate number of executioners among social misfits and, interestingly, criminals who see a military campaign as a means to enrich themselves. On some occasions, leaders who pursued an eliminatory agenda had had a criminal background before deciding to become politicians. Key examples here are Josef Stalin, a Bank Robber, and his Georgian entourage as well as Saddam Hussein who committed a murder at the age of 19. While these leaders might have pursued ideological goals, they often also used one-sided violence with the goal to enrich themselves. The macro-perspective of mass-killings pioneered by Esteban et al. (2010) shows that the exploitation of natural resources is one of the key factors behind recent genocides and politicides. The micro- and meso-level interpretation of one-sided violence presented in this article supplements this analysis and shows that soldiers are driven into the usage of force with a mixture of incentives and disincentives that do not have anything to do with the spoils of war and violence that their leaders are seeking to obtain. References Arreguin-Toft, Ivan (2005). How the Weak Win Wars: A Theory of Asymmetric Conflict. Cambridge University Press, Cambridge and New York. Azam, Jean Paul (2002). Looting and Conflict between Ethno-Regional Groups: Lessons for State Formation in Africa. Journal of Conflict Resolution, 46, 131-153. Azam, Jean Paul (2006). On thugs and heroes: Why warlords victimize their own civilians. Economics of Governance, 7, 53-73. Azam, Jean Paul and Anke Hoeffler (2002). Violence Against Civilians in Civil Wars: Looting or Terror? Journal of Peace Research, 39, 461-485. Bazenguissa-Ganga, Remy (1999) The Spread of Political Violence in Congo-Brazaville. African Affairs (1999), 98, 37-54. Beah, James (2007) The Making, and Umaking, of a Child Soldier. New York Times Magazine ( 9C8B63&sec=&spon=&pagewanted=2, last consulted, July 28, 2010) Bussmann, Margit& Schneider, Gerald 2010. A Porous Humanitarian Shield: The Laws of War, the Red Cross, and the Killing of Civilians. Working paper, Universitдt Greifswald/Universitдt Konstanz. Collier, Paul & Anke Hoeffler (2004). Greed and Grievance in Civil War. Oxford Economic Papers 56, 563-595.
11 Collier, Paul (2000). Rebellion as a Quasi-Criminal Activity. Journal of Conflict Resolution 44, 839-853. Conrad, Joseph (2000) Heart of Darkness. London: Penguin (originally published in 1902) Downes, Alexander B. (2006). Desperate Times, Desperate Measures. International Security 30 , 152-195. Downes, Alexander B. (2007). Restraint or Propellant? Democracy and Civilian Fatalities in Interstate Wars. Journal of Conflict Resolution 51 , 872-904. Downes, Alexander B. (2008). Targeting Civilians in War. Ithaca: Cornell University Press. Eck, Kristine & Lisa, Hultman (2007). One-Sided Violence against Civilians in War: Insights from new Fatality Data. Journal of Peace Research 44(2): 233-246. Fortna, Virginia Page (2004). Does Peacekeeping Keep Peace? International Intervention and the Duration of Peace After Civil War. International Studies Quarterly, 48, 269­292 Gettleman, Jeffrey (2010) "Fleeing Rebels Kill Hundreds of Congolese". New York Times, March 27, 2010 (, last accessed April 12, 2010) Gutiйrrez Sanнn, Francisco (2008). Telling the Difference: Guerrillas and Paramilitaries in the Colombian War. Politics & Society, 36, 3-34. Harff, Barbara (2003). No Lessons Learned from the Holocaust? Assessing Risks of Genocide and Political Mass Murder since 1955. The American Political Science Review 97, 5773. Herreros, Francisco & Henar Criado (2009) Pre-emptive or Arbitrary: Two Forms of Lethal Violence in a Civil War. Journal of Conflict Resolution, 53, 419-445 Hultman, Lisa. (2009). The Power to Hurt in Civil War: The Strategic Aim of RENAMO Violence. Journal of South African Studies 35, 821-834. Hultman, Lisa. (2010). Keeping Peace or Spurring Violence? Unintended Effects of Peace Operations on Violence against Civilians. Civil Wars. Humphreys, Macartan & Jeremy M. Weinstein (2006a). Handling and Manhandling Civilians in Civil War. American Political Science Research, 100, 429-447. Humphreys, Macartan & Jeremy M. Weinstein (2006b) Handling and Manhandling Civilians: Derivation of Hypotheses". Unpublished webappendix, Columbia University/Stanford University (, last consulted April 21, 2010) Humphreys Macartan & Jeremy M. Weinstein (2008). Who Fights? The Determinants of Participation in Civil War. American Journal of Political Science, 52,436-455. Kalyvas, Stathis N. (1999). Wanton and Senseless? The Logic of Massacre in Algeria. Rationality and Society, 11, 243-285. Kalyvas, Stathis N. (2006). The Logic of Violence in Civil War. New York: Cambridge University Press. Krain, Matthew (2005). International Intervention and the Severity of Genocides and Politicides. International Studies Quarterly, 49, 363-387. Kupperman, Alan J. (2001). The Limits of humanitarian interventions: Genocide in Rwanda. Washington D.C.: Brookings Institutions Press. Lyall, Jason (2009). Does Indiscriminate Repression Incite Insurgent Attacks? Evidence from Chechnya. Journal of Conflict Resolution 53, 331-362 Lyall, Jason (2010). Ethnicity and Violence in Civil War: Evidence from the Second Chechen War. American Political Science Review 104 (1): 1-20. MacDonald, Glenn; Marx, Leslie (2001). Adverse Specialzation. Journal of Political Economy 109/4: 864-899. Melander, Erik (2009). Selected To Go Where Murderers Lurk?: The Preventive Effect of Peacekeeping. Conflict Management and Peace Science. 26,389­406
12 Mennecke, Martin (2007). Punishing Genocidaires: A Deterrent Effect or Not? human rights Review, 8, 319-339 Minorities at Risk Project. 2008. "Minorities at Risk Organizational Behavior Dataset." College Park, MD: Center for International Development and Conflict Management. Mitchell, Neil J. 2009. Agents of Atrocity. Leaders, Followers, and the Violation of Human Rights in Civil War. New York and Houndsmills: Palgrave. Plaut, Martin 2010 "DR Congo rebel massacre of hundreds is uncovered". BBC News, March 28 (, last consulted April 21, 2010) Raleigh, Clionadh, Andrew. Linke, Hеvard Hegre and Joakim Karlsen (2011). "Introducing ACLED: An Armed Conflict Location and Event Dataset". Journal of Peace Research, in press. Raleigh, Clionadh, Hеvard Hegre, Joachim Carlsen & Christin Ormhaug. 2006. Armed Conflict Location and Event Dataset (ACLED) Codebook. Version 1-2006. Centre for the Study of Civil War, International Peace Research Institute, Oslo (PRIO) Schneider, Gerald und Bussmann, Margit (2010). Accounting for the Dynamics of One-Sided Violence: Introducing KOSVED, Working paper, Universitдt Konstanz.. Schneider, Gerald; Banholzer, Lilli; Haer, Roos (2010) Cainґs Choice: Causes of One-Sided Violence against Civilians. In Causes of War: An Introduction to Theories behind Warfare and Collective Violence, Tor Georg Jakobsen Ed. 2010. New York: Nova Science Publishers. Valentino, Benjamin A. (2004). Final Solutions: Mass Killing and Genocide in the 20th Century. Ithaca: Cornell University Press. Valentino, Benjamin A., Paul Huth, and Dylan Balck Lindsay (2004). Draining the Sea: Mass Killing and Guerilla Warfare. International organization, 58, 375-407. Valentino, Benjamin, Paul Huth, and Sarah Croco (2006). Covenants without the sword: international law and the protection of civilians in times of war. World Politics, 58, 339-77. Vinci, Anthony (2005) The Strategic Use of Fear by the Lord's Resistance Army. Small Wars & Insurgencies, 16, 360-381. Weinstein, Jeremy M. (2005). Resources and the Information Problem in Rebel Recruitment.Journal of Conflict Resolution, 49, 598-624. Weinstein, Jeremy M. (2007). Inside Rebellion: The Politics of Insurgent Violence. Cambridge: Cambridge University Press.
13 Appendix: Details on the Interviews Conducted in Eastern Congo, May 2009 The interviews took place in three so-called welcome centres: BVES (Bureau pour le Volontariat au Service de l'Enfance et de la Santй), LAV (Laissez l'Afrique Vivre) and CAPA (Centre d'Apprentissage Professionnel et Artisanal). These welcome centres provided support to soldiers who were mainly children at the time of their release or escape from different warring factions and their transfer to the United Nations Mission DR Congo (MONUC). All centres granted the Konstanz research team, which consisted of researchers from the Chair in Neuropsychology (Thomas Elbert) and the Chair in International Relations (Gerald Schneider) the permission to conduct the interviews after an introduction to the purpose of the study and a presentation of codebook. The group of interviewers drew on a joint questionnaire which the research groups had developed jointly before the travel of the interviewers to Eastern Congo. Local collaborators translated the questions into the mother tongue of the respondents which were either Swahili, Mashi, Lingala or Kinyarwanda. Lilli Banholzer conducted 20, Heike Riedke 18 and Roos van der Haer 96 interviews; a handful of questionnaires were filled by Thomas Elbert and Elisabeth Schauer. The research team conducted all interviews in places that provided as much privacy as possible so that the respondents did not feel any pressure to misrepresent their experience in the armed groups in which they were active. Some interviews took place in separate offices provided by the welcome centres, others were conducted in open air. The interviewers introduced the respondents before the start of the interview to the purpose of the study and told them that they could stop the process whenever they wished to do so and that they would not receive any rewards for participation. The interviewers also told them that they and other researchers involved in the project would treat all information as strictly confidential.

G Schneider

File: incentives-to-kill-the-organizational-roots-of-onesided-violence.pdf
Title: Hiring Criminals or Soldiers: The Organizational Decision to Kill Civilians
Author: G Schneider
Author: schneider
Published: Tue Aug 31 17:27:53 2010
Pages: 14
File size: 0.17 Mb

Zack's Alligator, 7 pages, 0.14 Mb

Quiet City, 2 pages, 0.21 Mb

Cultural criminology, 2 pages, 0.01 Mb

Poems, 1799, 93 pages, 0.13 Mb
Copyright © 2018