BOOK SUMMARY Rivalry and Revenge. The Politics of Violence during Civil War Laia Balcells Duke University Introduction What explains violence against civilians in civil wars? Why do armed groups use violence in some places but not in neighboring places with similar characteristics? Why do they kill more civilians in some places than in others? More specifically, why do groups kill civilians in areas where they have full military control and their rivals have no military presence? The question of civilian victimization has been at the forefront of recent research on wars. To date, two types of explanations have emerged: a first generation of scholars considered prewar characteristics of countries. Following Clausewitz (1832/1968) and Schmitt (1976), violent conflicts were seen as the result of existing political cleavages, and violence as the consequence of these divisions. Recent empirical research has pointed instead to security concerns related to warfare, e.g., the military incentives of armed groups (Valentino et al. 2004; Kalyvas 2006; Downes 2006); to the survival incentives of civilians (Kalyvas 2006); or to the organizational characteristics of the groups (Mkandawire 2002; Weinstein 2006). These authors, who have used more systematic research methods than the previous generation of scholars, have been theoretically inspired by Mao Zedong's (1978) insight that war cannot be equated with politics because it has its own particular characteristics. The latter body of research has de-emphasized political variables despite the fact that civil wars are usually fought over political issues, e.g., demand for self-determination, regime, or leadership change. The tendency has been to assume that, even though politics matter at the outbreak of conflict, the internal dynamics of war are driven by factors that are not necessarily political. Summary of the Theory The book explores the puzzle of why we observe civilian victimization during civil wars that are fought along stable frontlines and in which armed groups have full military control of large areas of territory. Such civil wars are defined as conventional (Kalyvas and Balcells 2010), and they include the American and the Spanish civil wars, but also more recent civil wars such as the ones that took place in Bosnia, Ivory Coast, and Libya. The theoretical argument in this book is that armed groups target civilians who are strong supporters of the enemy, either to strengthen their control over territory they already occupy or to weaken the enemy in territory the enemy occupies. Violence takes two forms, direct or indirect, depending on the location of civilian supporters of the enemy. While direct violence occurs when enemy supporters are located in zones controlled by the armed group, indirect violence occurs when enemy supporters are located in zones controlled by
the adversary (provided the armed group possesses the military technology to carry out attacks in these regions). The key difference between the two forms of violence is that the armed group s own supporters can constrain direct violence in zones of control, whereas they cannot do so in zones of enemy control. Direct and indirect violence imply different strategies. When targeting enemy supporters behind enemy lines, the armed group aims to kill as many of them as possible, hence he targets locations with high concentrations of enemy supporters. In territory the armed group controls, in contrast, the group must take into account the preferences of its own supporters, who know the identity of the enemy s supporters and can choose whether or not to share that information with the group s militants. Group supporters are likely to collaborate with the armed group and identify enemy supporters if and only if it is in their own interest to do so, which is the case when eliminating enemy supporters can decisively shift the demographic balance and help them gain political control of the locality. Thus, direct violence is likely to occur where the balance between group supporters and enemy supporters is relatively even. Indeed, in places where the group s supporters are already predominant, violence is unnecessary, whereas in places where enemy supporters dominate, violence would have to be massive (hence too costly) to make a difference. The prediction is thus that indirect violence increases with rival supporters domination of a locality whereas direct violence increases with parity between supporters of the two rival groups. Violence against civilians is only likely to occur, however, where there have been high levels of pre-war mobilization along the cleavage line that divides the two groups, whether being ethnicity, religion, or ideology. This mobilization is what leads people to identify as strong supporters of one side or the other. In conflicts without meaningful pre-war mobilization, identities are less salient and armed groups do not devote resources to eliminate strong supporters of the rival groups. In other words, I argue that conventional civil wars without meaningful pre-war mobilization should not be the sites of mass violence against civilians. Empirical Strategy The dependent variable in this book is lethal violence against civilians during civil wars. I distinguish between direct and indirect violence, which are two forms of violence that are usually studied separately in the literature and for which I provide a common explanatory framework. The empirical strategy of the book is multi-method: I use quantitative methods in combination with qualitative analyses. Following a recent trend in political science (e.g., Wilkinson 2004; Kalyvas 2006, Christia 2012), the research design consists of systematically exploring intra-country variation (with large-n sub-national data) and combining it with additional secondary evidence from other cases in order to provide external validity. In the book, I combine insights from two novel large-n subnational datasets (i.e., Spain and Ivory Coast) with a crossnational test of implications and secondary evidence from other cases. In addition, for the case of Spain I use evidence collected from oral sources (i.e., 60 civil war testimonies) and from over a hundred published sources, including general history books, as well as regional and local studies.
The Spanish Civil War is the main case study of the book. I chose this case for a number of reasons: first, the Spanish Civil War is, together with the US Civil War, a paradigmatic case of a conventional civil war. Second, using this civil war constitutes a dispute to the neglect of historical cases in the study of civil war violence, which risks generating wrong conceptualizations of the phenomenon. Third, the Spanish Civil War has a special relevance on its own because it was a crucial conflict in the West European interwar period, which led to the first open confrontation between the antagonistic ideologies of Fascism and Republicanism/Democracy (Lannon 2002) and it was a particularly severe conflict, with circa 800,000 estimated total deaths. Fourth, the Spanish case allows for the use of fine-grained data at a local level. Finally, I could have access to primary and secondary sources in Iberian languages, as well as to a network of local historical archives and to interviewees, which allowed me to put together several datasets. The book also uses the case of the recent Ivorian civil war to test the observable implications of the theory with a recent civil war that was fought along ethnic lines. I have built a provincial level dataset on violence during the Ivorian Civil War, which also includes data on electoral results, ethnic composition, natural resources, and geographical characteristics of the provinces. I have also analyzed the dynamics of violence among the neighborhoods of the capital of Ivory Coast, Abidjan. The results for the Ivorian Civil War are consistent with those obtained for the Spanish Civil War, showing that direct violence increases with levels of parity between rival groups. Although the Spanish and the Ivorian civil wars are very dissimilar, the comparison of the two cases yields valuable insights. Importantly, the combination of an old and ideological civil war (Spain) with a new and ethnic civil war (Ivory Coast) adds external validity to the theory in the book. Chapter Breakdown The book is divided in three parts. The first part (chapters 1 and 2) introduces the book and the underlying theory. The second part (chapters 3, 4, and 5) focuses on the Spanish Civil War, which constitutes the core empirical evidence in the book. The third part (chapters 6, 7, and 8) focuses on the Ivorian Civil War as well as on evidence from additional cases. This part also concludes the book. Chapter 1 introduces the book, explaining its motivating puzzle as well as its theoretical and empirical strategies. Chapter 2 develops a theory of violence against civilians in civil wars. The theory explores the puzzle of violence during conventional civil wars, and it incorporates political variables in a strategic approach to wartime violence. The theory thus combines insights from the first and second generations of scholars of violence during armed conflict. The theory also considers emotional factors (i.e., endogenous to the war), which are taken as complementary to exogenous political factors. The chapter includes a set of testable hypotheses on the determinants of direct and indirect violence, as well as a number of additional observable implications. Part II focuses on the Spanish Civil War, which is the main case study used in order to test the theory of the book. Chapter 3 introduces the Spanish Civil War and presents descriptive
patterns that should allow the reader to acquire a sense of the dynamics of violence that took place in this conflict more generally, as well as in the different regions that will be analyzed empirically. This chapter generates valuable descriptive inference (King, Keohane and Verba 1994) and develops a number of novel insights regarding this civil war. In particular, it shows that executions taking place in both Nationalist and Republican territories followed similar patterns with regards local level interactions among armed groups, political committees, and local civilians. In a nutshell, this chapter shows that similar dynamics of violence against civilians took place in both fighting sides of the Spanish Civil War, even though they were fundamentally different armed organizations. Chapter 4 presents a comprehensive test of the theory of direct violence against civilians with quantitative data from eight Spanish provinces, which comprise over 2,000 municipalities. In order to test the hypotheses, I apply a variety of statistical techniques that range from non-parametric to parametric tests, as well as spatial regression analyses. The results of these tests are supportive of the idea that prewar competition at the local level is explanatory of direct violence in the first stage of the war and that wartime factors (i.e., previous violence) gain explanatory relevance in subsequent stages. The results also indicate that domination is not the mechanism linking political alignments and violence, as would be argued from a pure Clausewitzian perspective. In other words, we do not observe greater levels of direct violence in locations with greater number of supporters of the enemy. Furthermore, the consistency and robustness (across armed groups and regions) of the variable measuring parity between blocs (i.e., competition) challenges the hypothesis according to which organizational factors account for levels of violence against civilians in conflict. In chapter 5, I test the determinants of indirect violence against civilians with quantitative data at the municipal level from the region of Catalonia during the Spanish Civil War. I also draw on a variety of econometric techniques, namely multivariate and spatial regressions, and several different specifications of the models. The results are very robust at showing that political variables account for bombings at the local level and that the relationship between support for a group and bombings (by the rival armed group) is monotonically positive. This finding derives from the motives of armed groups as well as from the nature of the production of indirect violence. Furthermore, endogenous-to-the-war factors are also found to play a significant role in explaining indirect violence, as demonstrated by the occurrence of retaliatory strikes in Catalonia in advanced phases of the civil war. Part III provides external validity to the results in the book. In Chapter 6 I test the theory with data from Ivory Coast, where there was a civil war (with some interruptions) between 2002 and 2011. The war in Ivory Coast was also conventional in type, although it was accompanied by much less civilian victimization as compared to the Spanish one. I argue that this was due to the lower degrees of political mobilization in this country, which led armed groups to devote lesser amount of resources to victimize strong supporters of the enemy group. This was particularly the case of the 2002-2007 period: the absence of widespread political mobilization preceding the war onset made the rearguard territories overall more secure for the armed groups. Mobilization was however higher by 2010-2011,
when the Presidential elections in Ivory Coast were followed by an intensive wave of intensive violence. I explore empirically the dynamics of violence in this post-electoral period, using the electoral results as a proxy for the distribution of support for each of the blocs in conflict, across provinces. Interestingly, in Ivory Coast the determinants of violence against civilians resemble very much those in Spain: local level parity between the two political blocs is a crucial variable explaining direct violence against civilians. The consistency in the results has further relevance given that the civil war in Ivory Coast has been labeled as an ethnic civil war, and not as an ideological one like the Spanish Civil War. Furthermore, the war in Ivory Coast would fit into the new wars category (Kaldor 2006) while the Spanish Civil War would fit the old wars one. Chapter 7 contains secondary evidence from other cases, such as Bosnia, Northern Ireland,a nd Colombia, which is overall consistent with the theory in the book. In addition, the chapter tests theoretical and empirical implications of the theory. Using a large-n dataset on civil wars in the contemporary world (covering the period 1944-2011) I check differences in levels of violence against civilians across and within types of wars, and I explain them through the lens of the theory in the book. Chapter 8 outlines the main implications of the findings in the book. The chapter also considers competing hypotheses. Finally, the chapter addresses a number of caveats, traces some implications of the book for comparative politics more generally, and it delineates avenues for further research on the topic of civilian victimization during armed conflict as well as on broader issues regarding human security, conflict and violence.