Targeting Civilians in War

Targeting Civilians in War

Alexander B. Downes

Cornell University Press

ithaca and london

Contents

Acknowledgments

Introduction

1. Defining and Explaining Civilian Victimization

2. Statistical Tests: Civilian Victimization, Mass Killing, and Civilian Casualties in Interstate Wars

3. The Starvation Blockades of World War I: Britain and Germany

4. Strategic Bombing in World War II: The Firebombing of Japan and the Blitz

5. Guerrilla Warfare, Counterinsurgency, and Civilian Victimization: The Second Anglo-Boer War

6. Territorial Annexation and Civilian Victimization: The Founding of the State of Israel, 1947–49

7. Negative Cases: Why Civilian Victimization Doesn’t Happen

Conclusion

Notes

Acknowledgments

Over the course of researching and writing this book, I have benefited from the help and advice of many people. I owe a special debt of gratitude to John Mearsheimer. I arrived at the University of Chicago in the mid-1990s as a former classical musician nervously embarking on what I hoped would be a second career as a political scientist. John took a chance that someone who had until recently spent much of his time practicing the works of Mozart, Beethoven, and Brahms could become an international relations scholar. I am immensely grateful that he did. John read and commented on numerous drafts of this manuscript, giving feedback that was always helpful, and improved the final product immeasurably.

This book literally would not have been possible without the input of Robert Pape, who suggested the question to me in a brainstorming session in January 2001. What if you could show that democracies were more moral in the way that they fought wars? Now that would be really something, he mused. I came to a different conclusion regarding the impact of democratic war-fighting on civilians, but that in no way diminished Bob’s enthusiasm for the project as it progressed. I thank him for his support. Charles Glaser and Stathis Kalyvas helped sharpen my arguments and provided encouragement.

A few other individuals made special inspirational or practical contributions to this book. Benjamin Valentino, for example, was in many ways the trailblazer whose work on mass killing brought the study of civilian victimization into the mainstream of security studies. Ben met with me when I was starting the project and has provided helpful comments on my work as well as encouragement and advice along the way. I have been fortunate to have his path to follow. Ivan Arreguín-Toft generously shared his data on barbarism in asymmetric conflicts and has provided constant encouragement. Jasen Castillo, Kelly Greenhill, and Sebestian Rosato each offered not only extensive feedback on my ideas but also their friendship, support, and good humor, for which I am tremendously grateful. I thank Roger Haydon at Cornell University Press for his interest in and detailed comments on parts of the manuscript.

I also thank those people who read and commented on various versions of this project over the past several years: Laia Balcells, Barton Bernstein, Kathryn Cochran, Michael Desch, Matthew Fehrs, Martha Finnemore, Christopher Gelpi, Hein Goemans, Peter Gourevitch, Colin Kahl, Robert Keohane, Helen Kinsella, Gregory Koblentz, Matthew Kocher, Eric Mvukiyehe, Richard Price, Dan Reiter, Thomas Spragens, and Elisabeth Wood. I also thank the two anonymous reviewers for Cornell University Press whose comments were especially helpful. I apologize to anyone I may have forgotten.

Several institutions provided generous financial assistance that allowed me to conduct the research and writing for this book. I spent a fruitful year at the Olin Institute for Strategic Studies at Harvard University. I thank Stephen Rosen and Monica Duffy Toft for the opportunity to participate in Olin’s program and take advantage of Harvard’s many resources. I later received a year of support from the Center for International Security and Cooperation (CISAC) at Stanford University, where I profited from the guidance of Scott Sagan and Lynn Eden. Along the way I also received financial assistance from the Eisenhower Institute in the form of a Dwight D. Eisenhower/Clifford Roberts Graduate Fellowship, and the Harry Frank Guggenheim Foundation, which provided a dissertation fellowship. I also thank the Triangle Institute for Security Studies and the Arts and Sciences Council for Faculty Research at Duke University, both of which gave funds for research in the book’s final phases.

Parts of the introduction and chapters 1 through 3 previously appeared in my article Desperate Times, Desperate Measures: The Causes of Civilian Victimization in War, International Security 30, no. 4 (spring 2006): 152–95. Parts of chapter 2 appeared as Restraint or Propellant? Democracy and Civilian Fatalities in Interstate Wars, Journal of Conflict Resolution 51, no. 6 (December 2007): 872–904. Some of the material on the Second Anglo-Boer War in chapter 5 appeared in Draining the Sea by Filling the Graves: Investigating the Effectiveness of Indiscriminate Violence as a Counterinsurgency Strategy, Civil Wars 9, no. 4 (December 2007): 420–44. I thank these journals and their publishers for granting me permission to adapt the material in this book.

For excellent research assistance at Duke University, I thank Soo-Jung Choi, Max Entman, Chad Troop, but especially Katherine Jordan. Katherine was instrumental in helping me prepare the final version of the manuscript and saved me from innumerable errors. I cannot thank her enough. I alone am responsible for any mistakes that remain.

On a personal note, I thank my parents, Bryan and Sheri Downes, for their love and support through thick and thin. They have always been there to help celebrate my successes and to help pick me up when I have stumbled. As a teacher and mentor I only hope I can live up to my father’s example, demonstrated by the outpouring of affection and thanks from his former students when he retired in 2001.

Finally, I thank my wife, Tanya Schreiber, who has tirelessly endured many a furrowed brow these last several years. Not only has Tanya served as a sounding board and adviser for my work—I think she can explain my arguments better than I can—but she has coaxed me through the hard times and bad days when I thought I would never finish. Moreover, she has shown me by example how to persevere; her quiet courage in the face of adversity has inspired me more than she knows. I dedicate this book to her.

Introduction

War, as it is so often said, is hell. War is cruelty and you cannot refine it, wrote General William T. Sherman in justifying his decision to evict the inhabitants of Atlanta and burn the city during the American Civil War. One of the principal reasons why war is thought to be hell is the impact it has on innocent civilians, for in addition to consuming the lives of armed combatants, war also devours the lives of those who are not involved in the fighting. Over the past three centuries, for example, civilians (a term I use interchangeably with noncombatants) have comprised half of all war-related deaths. In the twentieth century alone, an estimated 43 million to 54 million noncombatants perished from war-related causes, accounting for between 50 percent and 62 percent of all deaths from warfare.¹

Even a partial list of the horrors over the past one hundred years is staggering.² Tens of thousands of civilians, for example, perished in concentration camps in Cuba, South Africa, and the Philippines at the turn of the last century, and the German war against the Herero people in Southwest Africa resulted in their near extinction. During World War I, the British-led blockade of the Central Powers caused widespread malnutrition and disease and played a large part in the nearly 1 million excess deaths among the civilian population of Germany and Austria-Hungary. Nazi Germany murdered 6 million European Jews and millions of others in Eastern Europe and the Soviet Union during World War II, while Allied strategic bombing killed 300,000 Germans and as many as 900,000 Japanese. United States strategic bombing in the Korean War killed hundreds of thousands of noncombatants, as did indiscriminate American and Soviet counterinsurgency tactics in Vietnam and Afghanistan, respectively. Civil conflicts resulted in massive civilian death tolls in numerous countries, such as China (1927–49), Spain (1936–39), Guatemala (1966–85), Nigeria (1967–70), Ethiopia (1974–91), and Angola (1975–2002), and the ongoing conflict in the Darfur region of Sudan has killed at least 200,000 and displaced more than ten times that number.³

The startling number of civilian casualties in modern wars is puzzling for two reasons. First, there is a widespread belief that killing innocent civilians is morally wrong. According to a recent International Red Cross survey of populations in war-torn societies, for example, "a striking 64 per cent say that combatants, when attacking to weaken the enemy, must attack only combatants and leave civilians alone." By contrast, a mere 3 percent accept the view that belligerents should be permitted to attack combatants and noncombatants alike.⁴ The past and present attitudes of Americans are similar: before World War II the American public resolutely opposed urban area bombing as counter to American humanitarian ideals, whereas hypothetical scenarios regarding an invasion of Iraq in 2003 showed that a majority of Americans consistently opposed going to war if it would result in thousands of Iraqi civilian casualties.⁵ Why are noncombatants frequently targeted despite the widespread belief that doing so is not only wrong but also illegal according to international law?

Second, killing civilians in war is widely believed to be bad strategy: it rarely helps the perpetrator achieve its goals and can even be counterproductive by strengthening an adversary’s will to resist. One recent study, for example, argues that terrorizing civilians in war is self-defeating: The nation or faction that resorts to warfare against civilians most quickly, most often, and most viciously is the nation or faction most likely to see its interests frustrated and, in many cases, its existence terminated.⁶ A leading analyst of interstate coercion agrees, arguing that punishment strategies aimed at an adversary’s civilian population—implemented with airpower, seapower, or economic sanctions—rarely extract meaningful concessions.⁷

Given the moral stigma attached to civilian victimization and its supposedly dubious effectiveness, why do governments nevertheless use military strategies that target enemy noncombatants?⁸ One school of thought identifies regime type as the key factor but is of two minds regarding its effect. According to some analysts, democracies—which adhere to domestic norms that proscribe killing innocent civilians, whether at home or abroad—are less likely to target civilians than nondemocracies, which are not so constrained.⁹ Studies of democratic institutions and war, however, imply just the opposite: democracies could be more likely to target noncombatants because the vulnerability of leaders to public opinion makes them wary of incurring heavy costs on the battlefield for fear of losing support at home.¹⁰ This fear could compel democratic elites to target noncombatants to avoid costs or win the war quickly.

A second explanation emphasizes the barbaric identity of the adversary: civilian victimization results from the belief that one is fighting an uncivilized enemy.¹¹ Sebastian Balfour, for example, in his history of Spain’s efforts to conquer northern Morocco, writes that European powers made a distinction…between the treatment of fellow Europeans and that of colonials who resisted European advance. The standards of warfare that could be applied to the colonial enemy were different because these opponents were not ‘fully civilized.’¹² The choice of strategy, in other words, depends on one’s view of the adversary: the laws of war apply only in wars against civilized opponents, not barbarians.

A third set of arguments focuses on military organizations. One variant contends that organizational culture—defined as the the pattern of assumptions, ideas, and beliefs that prescribes how a group should adapt to its external environment and manage its internal affairs—is what steers militaries toward or diverts them away from civilian victimization.¹³ Specifically, when a military’s organizational culture dictates a strategy that relies on the targeting of civilian populations, states will typically escalate to the intentional killing of noncombatants and sometimes even genocide during the course of a war.¹⁴ A different organization-centered argument maintains that militaries target civilians when a particular service branch believes doing so will serve its parochial interests. This is likely to be the case when multiple services are competing to make the largest contribution to victory so as to capture the lion’s share of postwar military spending, or when an organization is struggling to achieve independence as a separate service.¹⁵

Although these arguments are intuitively plausible, the evidence does not support them: as this book will show, democracies are somewhat more likely than nondemocracies to target civilians, but democracy alone is not the principal driving factor. Preexisting cultural differences between states (such as race or religion) that might engender perceptions of barbaric identity, moreover, fail to predict civilian victimization. Nor do organizational cultures or parochial interests reliably explain decisions to target noncombatants.

Based on an investigation of interstate and colonial wars over the last two centuries, this book identifies two other factors that are primarily responsible for civilian victimization. First, desperation to win and to save lives on one’s own side in costly, protracted wars of attrition causes belligerents to target enemy civilians. According to the desperation logic, states that are embroiled in costly and prolonged struggles become increasingly desperate to snatch victory from the jaws of defeat and reduce their own losses. States that find themselves in this situation target noncombatants because doing so allows them to keep fighting, reduce casualties, and possibly win the war by coercing the adversary to quit. In short, states—including democracies—tend to prize victory and preserving the lives of their own people above humanity in warfare: desperation overrides moral inhibitions against killing noncombatants.

Civilian victimization motivated by desperation in wars of attrition rarely intends to destroy the victim group.¹⁶ Rather, the targeting of noncombatants in these wars is primarily a coercive strategy: it is a means to persuade an enemy government (in a conventional war) or rebel group (in a guerrilla war) to accede to the coercer’s political or military demands. As a form of coercion, civilian victimization can follow the logic of punishment or the logic of denial.¹⁷ In the former, violence is directed at noncombatants in the hope that they will rise up and demand that their government end the war. In other words, civilian victimization as punishment targets the enemy civilian population’s will to resist. The punishment logic commonly underpins violence against noncombatants during sieges, naval blockades, and strategic bombing of urban areas.

In the denial logic, on the other hand, civilian victimization is meant to undermine an adversary’s military capability to resist. In counterguerrilla wars, for example, killing civilian supporters of the other side—or evacuating them from targeted areas—intimidates other civilians (or simply prevents them) from providing assistance to the rebels and hence undermines the insurgents’ logistical ability to continue fighting. Even some bombing campaigns have aimed to reduce enemy output of military goods by killing the civilian workforce. Civilian victimization as a coercive mechanism is typically a tactic of later resort, because coercion—whether of the punishment or denial variety—does not work quickly.¹⁸ States, therefore, frequently turn to coercion later, only after their initial strategy to win the war quickly has failed.

The second mechanism that leads to civilian victimization is belligerents’ appetite for territorial conquest: when states seek to conquer and annex territory that is inhabited by enemy noncombatants. This scenario typically occurs in wars of territorial expansion or when hostilities break out between intermingled ethnic groups that claim the same territory as their homeland. In this circumstance, attacking enemy civilians often makes good strategic sense because it eliminates fifth columns that could rebel in an army’s rear area and also heads off potential revolts that might occur later on. Furthermore, purging an enemy’s civilian population reduces the likelihood that the adversary will attempt to reconquer the disputed territory in the future by removing a major reason for war: rescuing their national brethren trapped behind enemy borders. One’s claim to territory, moreover, is strengthened by facts on the ground, foremost among them the national character of the population. The Kosovo Albanians’ claim to self-rule in Kosovo, for example, is enormously strengthened and legitimized by the fact that more than 90 percent of the population in Kosovo is Albanian. Each of these factors makes it likely that when states try to seize territory from each other, they will seek to change the demographic situation in the conquered area by targeting civilians.

Civilian victimization produced by the appetite to annex territory from a neighbor—contrary to civilian victimization in protracted wars of attrition—tends to be a tactic of early rather than later resort. The reason is that enemy civilians are readily accessible to the invader and expelling or killing them pays real dividends by removing the threat of rebellions in the rear, eliminating the possibility that the opponent might try to rescue its people in the future, and solidifying the attacker’s claim to the land. Moreover, because it is not usually produced by desperation, civilian victimization for territorial reasons can occur when a belligerent is winning or doing well.

Democratic regime type by itself increases the likelihood that a state will victimize enemy noncombatants in warfare. At first glance, democracies appear to target civilians at a slightly higher rate than nondemocracies do. This difference, however, is accounted for by democracies’ being more likely than autocracies to use force against civilians in protracted wars of attrition. Outside of such wars, there is no difference in the propensity of the two regime types to target noncombatants. The evidence therefore indicates that once democratic leaders find themselves in costly or losing wars, something about democracy, such as the additional pressure generated by electoral institutions of accountability, may force leaders to take measures to reduce losses or deliver victory—including civilian victimization—even if it means violating the strictures of liberal norms. This demonstrates a potential dark side of democracy: how institutions designed to ensure domestic peace and tranquility may lead democracies to perpetrate injustices abroad.¹⁹ It also shows how the norms and institutions characteristic of democracies can contradict rather than reinforce each other in wartime.

Leaders need not be certain that a strategy of targeting civilians will succeed for civilian victimization to be a rational choice; they merely need to believe that it might contribute to victory (or stave off defeat) or lower their costs of fighting. Leaders might also perceive themselves as having no other option than to attack noncombatants to achieve their war aims. If civilian victimization offers even a small chance of reversing a grim situation, or delivering a state’s goals at a cost it can afford to pay, leaders may rationally take that chance. Once states become committed to victory, therefore, if the costs of fighting increase or the war begins to appear unwinnable, they tend to victimize civilians first before abandoning their goals. The fact that civilian victimization is often chosen in the most difficult circumstances also helps to explain why it has a relatively low success rate, at least in conventional interstate wars.

Understanding the causes of civilian victimization in warfare should be of interest to analysts in a variety of disciplines as well as policymakers. Scholarly interest in the question why—and under what circumstances—civilians become victims in war has increased in recent years, yet it remains understudied relative to other topics in security studies.²⁰ This is a sad state of affairs given the millions of innocent people who died in armed conflicts in the twentieth century alone. This question until recent years was investigated largely by genocide scholars, who found a striking correlation between war and genocide, but have not adequately explored how the dynamics of armed conflict sometimes lead to mass killing of noncombatants. Admirable exceptions to this trend exist in the study of guerrilla wars,²¹ but by their own admission these studies cannot explain the many instances of large-scale killing that occur in conventional wars.²² This book, therefore, aims to increase our understanding of this important phenomenon and contributes several causal mechanisms—protection of friendly forces, the need to win, and the threat posed by enemy civilians to securely holding contested territory—that transcend the guerrilla-conventional divide and explain civilian victimization in both types of conflicts.

This book has relevance for several debates that are of central concern to security studies.²³ First, as already mentioned, understanding the roots of civilian victimization may help explain the historical prevalence of punishment strategies despite their seeming inefficacy. The extant coercion literature equates inflicting pain on civilians with punishment and destroying military targets with denial. A closer examination of civilian victimization, however, shows that violence against noncombatants can be employed to sap the enemy’s will to resist or his capability to resist. Differentiating civilian victimization as punishment from civilian victimization as denial also helps explain the puzzle why some coercive strategies commonly described as punishment are more successful than others, an issue I explore further in the book’s conclusion.

Second, the treatment of noncombatants by democracies in war provides a new venue for testing arguments regarding the effect of liberal norms and democratic institutions originally devised to explain peace between democracies. I follow the lead of some democratic peace theorists and their critics by deducing further implications from these arguments and testing them against new evidence.²⁴ To the extent that the debate between proponents and detractors of democratic peace has bogged down owing to the limited time period in which data are available (1816 to the present), the small number of democracies, and the scarcity of interstate wars, this procedure may help move the debate forward by providing new evidence for or against different normative and institutional causal mechanisms.

Third, this book also supplements the literature on norms and force in international relations. To date, this literature has focused on explaining the rise of (and adherence to) normative restrictions on the use of force against noncombatants. In particular, scholars have studied the development of taboos against the use of certain weapons, such as nuclear weapons, chemical and biological munitions, and antipersonnel land mines.²⁵ By contrast, I look at the problem from the opposite perspective: which factors lead states to violate norms? Under what circumstances do norms protecting innocent people break down? When do seemingly civilized people revert to uncivilized methods of fighting?²⁶ More generally, the resort to civilian victimization may provide insights into the poorly understood question of escalation in war.²⁷

Finally, my findings regarding democracy and civilian victimization question the view that democracies’ responsiveness to the electorate and their relative restraint toward civilians in war renders them more vulnerable tocoercion by punishment, such as suicide terrorism. Some argue, for example, that terrorists are emboldened to strike democratic states because their publics have low thresholds of cost tolerance and high ability to affect state policy. Moreover, democracies are perceived to be relatively restrained in the level of force they will employ in response to terrorist attacks.²⁸ By contrast, I find little support for the view that democracies treat civilians better in interstate wars, or that democracies are more likely to be targeted for civilian victimization by autocracies.

From a practical perspective, greater knowledge of the causes of civilian victimization in warfare can also help answer policy questions. For one, such understanding could help policymakers assess the risk thatparticular conflictswill descend into brutalization of civilians and thus guide decisions about whether and how to intervene abroad in order to preventthe abuse of noncombatants. It should not come as a surprise to policymakers, for example, when theaters of battle characterized by extensive ethnic intermingling—such as Bosnia—result in major depredations against civilians. Moreover, curtailing civilian victimization will typically require more than airpower because bombing is rarely effective at stopping ethnic cleansing or preventing widely scattered counterinsurgent forces from operating.

Understanding the causes of civilian victimization, furthermore, can also help us determine whether the spread of democracy in the international system will lead to fewer noncombatant casualties in warfare. Democratization is often touted as a win-win situation: democracies respect human rights at home and refrain from fighting each other abroad. Consequently, spreading democracy has become the cornerstone of American foreign policy in the post–Cold War world. If democracies also fight their wars in accordance with the principle of noncombatant immunity, civilian victimization may become just a bad memory.

The study of civilian victimization is particularly relevant for the United States, which—owing to its preponderance of power, tremendous military technological advantages, and most recently the impetus provided by the open-ended war on terror begun by the George W. Bush administration—is the most militarily active state in the world. American power and technology lower the bar against using force by making it appear relatively easy and low cost, while fighting terrorism gives the United States a compelling reason to intervene abroad: to oust regimes that sponsor or harbor terrorists and to prevent the development of weapons of mass destruction by regimes that might pass them to terrorists.

Skeptics might argue that a liberal democracy like the United States in the twenty-first century would not resort to violence against civilians to achieve its objectives. When a quick show of force fails to cow or defeat an enemy expeditiously, however, American policymakers—as in past conflicts—will face pressures to escalate militarily, which may involve inflicting harm on civilians. The key factors, I argue, are the cost and protractedness of the war, not the regime type of the states involved. Some authoritarian states may be quicker to put civilians in the cross-hairs nowadays, but democracies historically have shown a striking tendency to target noncombatants or use indiscriminate tactics that lead to large-scale civilian deaths when faced with costly fighting or uncertain victory. Moreover, the cost-sensitivity of modern democracies like the United States places additional stress on policymakers to fight cost-free wars. Even if leaders are reluctant to kill civilians intentionally and openly, this pressure can still lead to noncombatant suffering in indirect ways, such as through the targeting of infrastructure and reliance on economic sanctions that adversely affect civilian health. I explore this possibility in chapter 7.

The U.S. military is geared to fight high-intensity conventional warfare and is not nearly as effective at fighting guerrilla insurgencies. America’s opponents may be able to exploit this weakness to deflect U.S. power, lengthen wars, and frustrate U.S. forces, which could lead the United States to target noncombatants. The guerrilla insurgency in South Vietnam, for example, proved deeply frustrating to American political and military leaders and did not fit into the high intensity conventional strategy the U.S. Army had developed to defeat a Soviet invasion of Western Europe. The attrition strategy implemented by U.S. commanders in Vietnam—which depended on racking up a high body count—demanded indiscriminate firepower, which killed large numbers of Vietnamese noncombatants.²⁹ More recently, American officials recognize that the open-ended war on terror and Islamic radicalism begun by the Bush administration and described by the president as the defining struggle of our time could take a generation or more to prosecute.³⁰ This conflict has already caused Bush and his advisers to sanction open-ended detention without trial of suspected militants, a redefinition of torture to maximize the legal techniques available to American interrogators to extract information from suspects, and the practice of extraordinary rendition, whereby terrorist suspects are turned over to third countries that are known to practice torture.³¹ Finally, large-scale conventional conflicts—such as a potential war with China over Taiwan—could also prove costly and difficult to win, perhaps leading to serious bombing campaigns that would kill many civilians.

This

book joins and builds on a small but growing literature dedicated to explaining why states and rebel groups target noncombatants. Two of the most prominent contributions are Benjamin Valentino’s Final Solutions: Mass Killing and Genocide in the 20th Century and Stathis Kalyvas’s The Logic of Violence in Civil Wars. Readers familiar with this literature might wonder how my work relates to theirs.

The degree of overlap between this volume and Kalyvas’s is relatively small. Kalyvas offers a theory of violence in civil (or, more accurately, guerrilla) wars in which each belligerent aims to secure control over—and hence collaboration and support from—the population. These conflicts are characterized by divided sovereignties, zones controlled to various degrees by incumbents and insurgents. The purpose of violence is to obtain the collaboration of the population in the various zones. In areas where each actor exerts monopoly control, little violence occurs since it is unnecessary. Violence is also generally absent in zones where control is up for grabs, because neither side possesses sufficient information to target civilians selectively. Indiscriminate violence under these circumstances is counterproductive because it alienates people and may cause them to support the adversary. Violence is most likely in zones where one side is locally dominant (although not hegemonic) but its control is slipping as the opponent makes inroads into the area. The declining belligerent targets actual and potentialdefectors to deter them from switching allegiances and thus maintain its grip on the population.³²

This book, by contrast, deals mostly with interstate wars, almost none of which are insurgencies. In conventional conflicts, belligerents employ indiscriminate violence against an enemy’s civilian population—to the extent they are able and the circumstances warrant—to gain coercive leverage over the adversary’s government. Belligerents also use indiscriminate violence in conventional conflicts to eliminate unwanted populations from conquered territory. In neither case is exerting control over—or seeking the collaboration of—enemy populations a relevant mechanism. The main point of overlap with Kalyvas is my chapter on counterinsurgency, but even here I identify a set of cases that defy Kalyvas’s logic of selective violence. Rather than becoming increasingly discriminate in their employment of violence over time, the British in South Africa moved in the opposite direction, interning entire populations as they became increasingly desperate to suppress the Boer insurgency. Indeed, selective violence in this conflict was ineffective at extracting meaningful collaboration from the civilian population. Similar trends are evident in other colonial wars, such as the Spanish-Cuban (1895–98), U.S.–Filipino (1899–1902), and Second Italo-Sanusi (1923–32, in Libya) wars.

The common ground between this book and Valentino’s is more extensive. Valentino seeks to explain incidences of large-scale (fifty thousand or more killed) government violence against noncombatants in the twentieth century, whether inflicted in the course of a war or not. He argues that the key to understanding these violent episodes lies not in regime type, moments of national crisis, or a society’s cleavage structure—factors stressed by previous studies of genocide—but rather in the goals of state leaders. Three types of objectives in particular can trigger mass killing. First, leaders seeking to achieve the radical communization of their societies, such as Joseph Stalin, Mao Zedong, and Pol Pot, have used mass killing to eliminate elements of the population that opposed this transformation or which did not fit into the communist view of society. Second, mass killing occurs when leaders want to expunge an ethnic group in order to implement racist or nationalist ideologies calling for the ethnic, national, or religious purification of the state or to resolve political or military conflicts between ethnic groups over the control of territory. Finally, leaders seeking to defeat guerrilla insurgencies use mass killing to eliminate the civilian support that fuels guerrilla movements.³³

There are important areas of synergy between this book and Valentino’s, but also several areas where we diverge. The most important similarity is that we both argue that state violence against noncombatants is largely the result of rational strategic calculations rather than emotion, dehumanization, or irrational hatred. For example, my research shows that states hope to win quick and decisive victories by force of arms and turn to targeting civilians only when their countermilitary strategies are frustrated. States are also sensitive to the possibility that civilian victimization may trigger enemy retaliation against their own population or prompt influential third parties to enter the war. Each of these factors sometimes deters states—at least temporarily—from opting for civilian victimization. In a similar vein, Valentino documents how leaders explore a variety of steps to achieve their aims without violence and turn to bloodshed only when other avenues are blocked. Another point of convergence is that we both generally downplay the role of cultural factors in producing violence. I contend, for example, that perceptions of an adversary’s identity are unrelated to the occurrence of civilian victimization, while Valentino argues against social cleavages and dehumanization.

There are a few differences, however. One concerns the dependent variable. This book focuses specifically on war, and the reasons that states target noncombatants in armed conflicts, whereas Valentino uses a numerical threshold of civilian fatalities to identify his case universe. One of Valentino’s three causal mechanisms, therefore—the collectivization of agriculture under communist regimes—plays no role in my work since it typically does not occur during wartime. A second difference is substantive: although many of the cases I examine fall into Valentino’s category of terrorist mass killing, he devotes little attention to these cases in his book.³⁴ My book can thus be seen as complementary to his in that it opens up and explores the dynamics of this category of cases. My other causal mechanism for civilian victimization—appetite for conquest—to some extent combines Valentino’s ethnic and territorial categories of mass killing because I look at cases where one state attempts to seize land from another, and particular ethnic groups may be viewed as obstacles to control over that territory. Although Valentino breaks decisively with previous genocide literature in many respects, he retains its emphasis—at least in the case studies that form the bulk of his book—on mass killing as a phenomenon of domestic politics.

A third difference is theoretical. My research places much less emphasis than Valentino’s on the role of individuals and their ideologies and beliefs. Valentino, in advocating a strategic perspective on mass killing, argues for the importance of individual leaders and their goals, depicting mass killing as a brutal strategy designed to accomplish leaders’ most important ideological or political objectives and counter what they see as their most dangerous threats.³⁵ The centrality of leaders fits well with the communist and ethnic cleansing scenarios, but less well with the third scenario, counterinsurgency. Here Valentino sheds his emphasis on leaders’ goals for the simple logic that targeting civilians makes good strategic sense for combating insurgencies. No particular ideology or goal is needed to explain mass killing in this situation; all one needs to posit is a preference for victory. Most of the cases in this book are of exactly this type: conflicts in which one or both belligerents simply seek to win and use increasingly violent instruments to coerce the adversary to concede. In other cases, states fight not only to win but to obtain territory, and civilian victimization arises out of the perception that some groups pose a threat to the assimilation of that territory.

The next chapter defines civilian victimization in greater detail, lays out the various types of civilian victimization, and then presents a series of theories to explain it, including my own desperation and territorial annexation arguments. The remainder of the book is devoted to empirical tests of the competing theories, starting with a statistical analysis of civilian victimization in interstate wars in chapter 2 and proceeding in chapters 3–7 through a series of case studies of blockade, strategic bombing, counterinsurgency, ethnic cleansing, and cases where civilian victimization did not happen. Iconclude with a discussion of the theoretical and policy implications of my findings.

[1]

Defining and Explaining Civilian Victimization

What is civilian victimization, and why do governments victimize civilians in warfare? I begin by defining the concept of civilian victimization, the phenomenon I seek to explain in this book. Civilian victimization is a military strategy chosen by political or military elites that targets and kills noncombatants intentionally or which fails to discriminate between combatants and noncombatants and thus kills large numbers of the latter. I then outline the common forms of civilian victimization and provide illustrative examples.

In the second part, I summarize three competing perspectives on civilian victimization: regime type, civilized-barbaric identity, and organization theory. The regime-type view maintains that domestic norms and institutions are the most important causes of civilian victimization. The civilized-barbaric identity argument, by contrast, contends that states kill an adversary’s noncombatants when they view the enemy as outside the community of civilized nations. Organization theory, finally, suggests that the cultures or parochial interests of military organizations are the prime causes of civilian victimization.

The third section presents my theory of civilian victimization, which identifies two crucial factors: the growing sense of desperation to win and to conserve on casualties that states experience in protracted wars of attrition, and the need to deal with potentially troublesome populations dwelling on land that an expansionist state seeks to annex. I discuss the logic underpinning these two factors and the differing implications they have for the timing of civilian victimization. I also explain why targeting civilians—even though it does not always work—is not an irrational gamble, but rather a calculated risk. I conclude with a short discussion of methods and cases.

Defining Civilian Victimization

Civilian victimization as I define it consists of two components: (1) it is a government-sanctioned military strategy that (2) intentionally targets and kills noncombatants or involves operations that will predictably kill large numbers of noncombatants. Civilian victimization violates the principles of noncombatant immunity and discrimination as enshrined in the Geneva Conventions and just-war theory, which require that belligerents must distinguish between combatants and noncombatants and refrain from taking aim at the latter.¹ Common forms of civilian victimization include aerial, naval, or artillery bombardment of civilians or civilian areas; sieges, naval blockades, or economic sanctions that deprive noncombatants of food; massacres; and forced movements or concentrations of populations that lead to widespread deaths. As with Benjamin Valentino’s definition of mass killing, civilian victimization is not limited to ‘direct’ methods of killing, such as execution, gassing, and bombing. It includes deaths caused by starvation, exposure, or disease resulting from the intentional confiscation, destruction, or blockade of the necessities of life. It also includes deaths caused by starvation, exhaustion, exposure, or disease during forced relocation or forced labor.²

Several aspects of civilian victimization bear elaboration. First, in accordance with the Geneva Conventions, I define combatants as all organized armed forces, groups and units which are under a command responsible to that Party for the conduct of its subordinates, as well as individuals involved in the construction of weapons.³ Noncombatants, by contrast, do not participate in armed conflict by fighting, carrying weapons, serving in the uniformed military or security services, or building weapons. Two general principles can be used to demarcate the line between combatants and noncombatants. One view, articulated by Michael Walzer, maintains that people have the right to be free from violence unless they forfeit that right by participating in military activity, such as taking up arms or working in a munitions factory. As Walzer puts it, No one can be threatened with war or warred against, unless through some act of his own he has surrendered or lost his rights.⁴ A second perspective focuses on the degree of threat an individual poses to the adversary to draw the combatant/noncombatant line. Only those individuals who actively participate in hostilities—by serving in the armed forces, for example—constitute an immediate threat of harm to the enemy and thus qualify as combatants.⁵ I argue that individuals who build bombs and other munitions also pose a significant threat of harm and should thus be included in the combatant category. If they are not, attacks on munitions factories (or targeted killings of terrorist bomb-makers, for that matter) would have to be classified as intentional targeting of civilians. This violates the common-sense understanding that such individuals are not owed the same degree of protection as those employed in other sectors of the economy, and that killing them is not morally wrong.⁶ Thus, whichever of these general principles one subscribes to, both point to an understanding of noncombatants as outside the realm of military activity either as fighters or makers of weapons.

Skeptics contend that nationalism and industrialization have eliminated the noncombatant category altogether because all citizens in modern states contribute to the war effort if only by going to work, paying taxes, or consenting to the use of force.⁷ This view clashes with the obvious fact that even in modern societies, there are many people who contribute little if anything to the war effort, and further that this relative disengagement (and absence of threat) makes a difference as to whether they may be killed.⁸ Many individuals work in sectors of the economy unrelated to the war effort; some, particularly children and the elderly, do not work at all. One study estimates that 75 percent of the population of an industrial country does not labor in war-related industries, and that even in industrial cities, 66 percent of the inhabitants are civilians.⁹

Second, civilian victimization is a government-sanctioned policy or strategy, as opposed to random or uncoordinated attacks by a few military units. Intentional attacks on civilians, as Christopher Browning has pointed out, can take the form of arbitrary explosions of violence or revenge inspired by battlefield frenzy, on the one hand, or can represent official government policy or standing operating procedure, on the other hand.¹⁰ Only the latter comprises civilian victimization as defined in this book: it consists of a government policy of sustained violence against a noncombatant population, rather than haphazard outbursts of brutality by frustrated troops. This does not mean, however, that civilian victimization must be initiated by civilian leaders; in fact, it is sometimes initiated by the military on the ground, but once political leaders become aware of the strategy and approve it—or decline to stop it—it becomes de facto government policy.¹¹

Third, civilian victimization consists of strategies that intentionally target civilian populations but also strategies that—although not purposefully aiming at noncombatants—nevertheless fail to distinguish between combatants and noncombatants. A strategy of civilian victimization, in other words, either targets civilians on purpose, or employs force so indiscriminately that it inflicts large amounts of damage and death on noncombatants. In certain cases, for example, belligerents openly declare or make statements of policy that designate noncombatants as the target of a strategy. British bombing policy in the Second World War, which charged Bomber Command with destroying urban areas in Germany to undermine the morale of the enemy civil population and in particular, of the industrial workers, falls into this category.¹² British policy clearly meant to kill German civilians in order to bring about the collapse of Germany’s will to fight. The 300,000 fatalities caused by this strategy were not a side-effect of