Foreign Intervention in Civil Wars [1 ed.] 9781527500471, 9781443891479

Citation preview

Foreign Intervention in Civil Wars

Foreign Intervention in Civil Wars By

Jung-Yeop Woo

Foreign Intervention in Civil Wars By Jung-Yeop Woo This book first published 2017 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2017 by Jung-Yeop Woo All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-9147-9 ISBN (13): 978-1-4438-9147-9

TABLE OF CONTENTS

Chapter One ................................................................................................. 1 Foreign Military Intervention in Civil Wars Chapter Two .............................................................................................. 11 Concepts and Definitions Chapter Three ............................................................................................ 31 Literature Review Chapter Four .............................................................................................. 47 Theory Chapter Five .............................................................................................. 75 Frameworks of Analysis Chapter Six ................................................................................................ 91 Descriptive Statistics Chapter Seven.......................................................................................... 109 Analysis I: First Intervention Chapter Eight ........................................................................................... 131 Analysis II: Second Intervention Chapter Nine............................................................................................ 149 Conclusion References ............................................................................................... 151

CHAPTER ONE FOREIGN MILITARY INTERVENTION IN CIVIL WARS

1.1 Introduction Since the end of the Cold War, civil wars have received greater attention. The sheer number of international conflicts is not comparable to that of internal conflicts. As Gleditsch et al. (2002) note, among 225 general conflicts in the world from 1946 to 2001, 163 conflicts can be identified as internal conflicts. They (Gleditsch et al. 2002) argue that internal conflict has been the dominant form of conflict since World War II. As internal wars have become more serious both in intensity and duration than interstate wars, scholars and practitioners have tried to determine the influences of domestic, regional, and international factors on the initiation, duration, and termination of civil conflicts. Civil wars have become the most important subject for scholars of international relations, since the effects of those civil conflicts have not been confined to those states alone. International law, norms, and institutions have traditionally banned intervention in the internal affairs of states. The United Nations Charter specifically indicates that domestic issues are not the jurisdiction of the organization’s reach. Historically, when foreign states intervened in other states, they sought to provide extraordinary justification on the basis of security or to embed intervention in an organized system of competition, as during the Cold War. The legitimacy of intervention has always been the subject of heated debate. Let us consider the example of Vietnamese intervention in Cambodia in 1979. In early 1979, Vietnam militarily intervened in Cambodia and overthrew the Pol Pot regime. Vietnam did not try to gain legitimacy by claiming that in intervening they were exercising some right to humanitarian intervention (Akehurst 1984). Instead, Vietnam denied that its forces had entered Cambodia and said that Pol Pot had been overthrown by the Cambodian people.

2

Chapter One

There was a Security Council debate on this matter. The Soviet Union, which backed Vietnam, argued that Pol Pot had been overthrown by the Cambodian people and, thus, not by Vietnam. Vietnam’s allies—including Bulgaria, Cuba, Czechoslovakia, East Germany, Hungary, Mongolia, and Poland—supported this argument. Almost all of the other states that took part in the debate said that Vietnam had acted illegally by intervening in Cambodia’s internal affairs. Several of these states also mentioned that the Pol Pot regime’s massive human rights violations should be condemned. Nevertheless, those states did not attempt to support Vietnam’s intervention, instead arguing that those human rights violations did not give Vietnam the right to overthrow that regime. As Akehurst (1984, 97) notes, “Not a single state spoke in favor of the existence of a right of humanitarian intervention.” As seen in the above case, even though sovereignty and nonintervention in domestic affairs of states are the most fundamental principles of the current international system, many cases have been observed in which third-party states became militarily involved in civil wars. According to Reagan (2000, 21), who defines internal conflicts as “armed combat between groups within state boundaries in which there are at least 200 fatalities,” there have been 140 civil wars from WWII to 1994. Out of those 140 civil wars, 90 conflicts have included at least one instance of third-party involvement. Hermann and Kegley (1996, 440) also contend that foreign military intervention is “arguably the most frequent type of military force in use and under debate today.” It was not an overstatement when, in 1975, Little (1975) argued that intervention was already a major feature of the contemporary international system. Pickering and Kisangani (2006, 363) contend that “as major wars have declined in frequency over recent decades and the efficacy of costly and time-consuming economic sanctions is questioned, foreign military intervention seems to have become a sine qua non of modern statecraft.” In recent history, we have observed that intervention by third-party states has generated widespread discussion and intense debate in both the domestic political arena and the international one. Foreign military intervention, let alone the legal and ethical debate over humanitarian intervention, has brought the issue into the international and domestic political arena. Decisions by government leaders to militarily intervene in a foreign country’s civil war, such as the US’s decision to involve itself in Vietnam, have inspired sharp divisions not only among politicians but also among ordinary citizens. The Vietnam conflict set an example for both the

Foreign Military Intervention in Civil Wars

3

military and policymakers, demonstrating that they should avoid future foreign military quagmires. Turkey’s intervention in Cyprus in 1974 still has a lingering effect on Turkey’s diplomatic relationship with other EU member states. The European Commission has proposed that the 25 heads of state freeze discussion on 8 of the 35 agreements that Turkey and all the EU members have to sign unanimously as part of the accession process. Three of these frozen chapters directly involve Turkey’s relationship with Cyprus. Cyprus became an EU member, but part of its territory was still occupied by Turkish forces (Konstandaras 2006). Even though there are many discussions of humanitarian intervention, not all intervention has the goal of ending internal conflicts and resolving human rights disasters. Carment and Rowlands (1998, 572) point out that “civil war represents opportunities for some and challenges for others.” It is also agreed that even if external actors do not instigate internal wars (Heraclides 2001), they can play a significant role in the process of civil wars. As discussed by Carment and James (2003, 11), “Far from resolving internal disputes, third-party involvement frequently translates into interference-prolonging and even intensifying such conflicts.” One extreme example that illustrates the role of external forces in the process of civil wars is the case of the Democratic Republic of the Congo (formerly Zaire). The conflicts in the Democratic Republic of the Congo occurred first in 1996 and recurred in the context of the second Tutsi opposition in 1998. During the first Tutsi opposition, which opposed the corrupt Mobutu Sese Seko regime, eight regional states supported the rebel Tutsis. South Africa and the United States tried to mediate an early settlement by providing an orderly exit for Mobutu. The war between the rebels and the government ended in eight months. However, another internal conflict emerged before a year had passed. More serious and destructive domestic and regional consequences were brought about by the second war. This second opposition by the Congolese Tutsis involved seven regional states taking opposing sides. It is reported that there have been more than three million casualties due to direct combat or war-related effects. This case illustrates the internal conflicts that are exacerbated by intervening outside forces. Based on previous studies as well as the above case, it is obvious that third-party intervention can affect the outcome and the duration of civil wars. Thus, we can assume that the intervention of third parties in civil wars has meaningful consequences for the manner in which those wars evolve. However, it is not only the decision to intervene that we find academically and practically interesting. If every civil war featured some

4

Chapter One

sort of foreign intervention, then it would not be of much interest to us academically. This issue became very interesting, however, because we have found that the propensity toward intervention by third-party states varies. That means that we also have observed many cases in which no third-party intervention was made. Noninvolvement by foreign states in certain civil wars has also brought about huge debates among scholars, practitioners, and even ordinary citizens. The existence of many failed and troubled states has required some sort of intervention and, particularly, intervention on humanitarian grounds. Those cases would be considered legitimate in the sense that no violation of the principle of sovereignty would be perceived as having occurred. However, we can see from history that there have been cases where no foreign involvement was attempted despite the voice of the international community. For example, in Rwanda in 1994, all of the superpowers and the neighboring countries tried their best to “avoid” becoming involved in the conflict. Power (2002) questions the lack of action by the Clinton administration, which was considered more committed to humanitarian intervention and to a moral foreign policy than any U.S. administration since World War II. As Romeo Dallaire (the Canadian general who led the 1993 UN peacekeeping mission in Rwanda) notes in his book (Dallaire 2003), the Rwandan genocide could have been easily stopped if foreign powers had intervened in a timely fashion. However, as Power (2002) notes, the United States not only did not intervene to stop the humanitarian disaster, but it actively prevented other willing powers from intervening and taking action. After Hutu soldiers murdered ten Belgian peacekeepers on April 7, 1994, the Belgian government made it clear to the United States that it would pull its troops out unless the UN presence in Rwanda was reinforced. The United States let Belgium pull out its troops. The United States reportedly also influenced the decision making process of the UN Security Council when it decided to cut the strength of the UN Assistance Mission for Rwanda (UNAMIR) from 2,500 to 500 just two weeks after the genocide began. We know what happened after that. More than 800,000 people died during the three months that followed. Along with this observation of intervention and non-intervention in the history, now we see variation in the dependent variable: intervention and non-intervention by a third party in civil wars. As we can clearly see from history, some civil wars saw several foreign countries become involved, whereas there was no foreign intervention in others. For example, in civil wars such as the notorious genocide in Rwanda in 1994 and more currently in Darfur, Sudan, foreign countries including major powers have

Foreign Military Intervention in Civil Wars

5

been unwilling to intervene despite a cry for intervention on humanitarian grounds on the part of the international community. It seems to us that sometimes the principle of sovereignty and non-intervention is more respected, even though humanitarian interventions in those brutal cases would be considered legitimate according to international law (Hoffmann 1996). These variations in intervention and non-intervention decisions tell us that the decision to intervene is not just based on legal or moral principles. It is more of a political decision, a cost-benefit analysis whose aim is to procure more benefits through certain actions. This study begins from this question of what factors lead third-party countries to intervene in other states’ internal wars. Why do some civil wars lead to interstate crisis while others do not? Among the various strategies of intervention that foreign states can choose, unilateral military interventions are the subject of this book. As Otte (1995) notes, military intervention is one of the higher-profile forms of intervention because “military intervention is, firstly, and most obviously, a deviation from the internationally acknowledged norm of non-intervention, but it is also a deviation from the normal pattern of the relations between the intervener and the target state” (5). Practically, military intervention by third-party states into civil wars is very risky business for the intervening states in that they have to bear a huge cost burden and suffer a great degree of risk. Pearson (1974) also agrees that for foreign states to militarily intervene in foreign states’ domestic conflicts entails considerable costs and risks. In the current world system, not all states can sustain those costs and risks. Thus, many previous studies on military intervention have focused on intervention by major powers. However, military intervention is not exclusive to these major powers. Dunér (1983) shows that minor or small powers also have employed military intervention as a foreign policy strategy. Thus, we also need to consider the behavior of minor powers with respect to their involvement in civil wars. As Regan (2000) notes, the question of why foreign states decide to unilaterally intervene in civil wars is still unanswered. The purpose of this book is to contribute to the civil war literature by expanding our knowledge of what factors lead third-party states to militarily intervene in foreign civil wars. Given that we have observed frequent military interventions by both major and minor powers, it is very important to understand why foreign states decide to intervene sometimes and not to intervene at other times. With regard to third-party intervention in civil wars, numerous literatures have been generated that discuss the issues surrounding the question of whether military intervention by a third-party is legitimate and

6

Chapter One

ethical. These literatures have inquired into the moral and legal implications of third-party intervention in civil wars and have tried to determine the criteria that can be used to judge whether the intervention or non-intervention is legitimate. However, in this book, I will not discuss these ethical and/or legal issues regarding military intervention except in considering them as basic theoretical background in my literature review section. Even though the discussion of moral and legal issue of military intervention is a completely separate academic inquiry from that of this book, a brief discussion is necessary because many military interventions have been made over the course of history in the name of humanitarianism. Also, military intervention has sometimes been called for but not sufficiently enacted for humanitarian purposes, as in Rwanda in 1994. In sum, the scope of inquiry of this book is the question of what factors lead to third-party intervention in civil wars by both major and minor powers. For this purpose, I propose that we consider four dimensions of civil war intervention. The first dimension to consider is the civil war itself. The characteristics of civil war are assumed to have some impact on a third party’s decision regarding intervention. Whether a given war is triggered by ideological issues (identity wars) or ethnic/religious issues (identity wars) is assumed to have an impact on the decision makers of foreign countries. Second, the characteristics of intervening states should be considered. Their domestic politics, including institutional settings, should be taken into account if we want to see how their decisions regarding intervention are made. As Cament et al. (2006) note, it is easier for those leaders with authoritarian power to make a decision regarding military intervention in foreign civil wars. Third, the relationship between the host country and the intervening country needs to be analyzed. Their formal alliances and ethnic ties are assumed to be factors in decision making around foreign intervention. Moreover, following the realist perspective, differences in military capability between the target country and the potential intervener will also have an impact on the decision making process. Finally, the relationship between the intervening states should be considered. This consideration is not only limited to the cases where there is more than one intervener. As we can infer from extended deterrence theory, the existence of another country backing one side in the civil war will yield a different decision on the part of a potential intervener. Previous studies, including Regan’s (2000), have neglected this dimension. In most previous studies about foreign intervention, all interveners have been treated equally. It is logically more reasonable to incorporate the interaction between the potential and actual interveners into a decision calculation model.

Foreign Military Intervention in Civil Wars

7

Decision makers have to consider the actions of other potential interveners because this will drastically change their cost-benefit analysis with regard to a possible intervention. Regan (2000) began his theory of intervention by proposing the decision making model of potential interveners. However, as he admits, his statistical analysis, which uses the conflict as the unit of analysis, does not capture the dynamics between potential interveners as occurs in his decision-theoretic framework. He notes, “There are conceptual problems in using the conflict as the unit of analysis because the emphasis of the empirical model shifts from the perspective of the individual decision maker to the aggregate case, asking in essence whether certain structural and contextual conditions increase the probability of an intervention” (2000, 52). In his next study with Lemke, he corrected this problem to some extent. Lemke and Regan (2004) explicitly relate Singer’s (1963) international influence model to intervention into civil wars. They address the problem caused by using the conflict as a unit of analysis by including the potential interveners in the data for civil wars waged between 1944 and 1994. However, they do not incorporate the sequence of interventions by different interveners. Their analysis does not distinguish between the original interveners, who intervene in the civil war first, and the counterinterveners, who intervene to counter the involvement of the first interveners. Those counter-interveners did not choose to intervene at the start; however, the first interveners’ entrance into the civil war then influences the calculations of the potential counter-interveners. The researchers also do not distinguish the first interveners from the second interveners, who intervene after the initial intervention has actually taken place. The cost and risks for the second interveners will be much less significant than those experienced by the first interveners. Thus, it is logically unacceptable to treat these different interveners equally in the analysis. As Carment and James (2003, 11) point out, whereas the instances of ethnic intervention accumulate, “Unfortunately, students of conflict management are only beginning to construct theories about third-party interventionism with a general range of application, most notably with respect to ethnic conflict.” Moreover, most current studies on ethnic interventionism have usually focused on the state where the conflict occurs (Carment and James 2003). However, those studies cannot explain why some states intervene and some states do not intervene in other states’ internal wars. This will be my contribution to the understanding of this phenomenon in international relations. By adding the relationship between potential and

8

Chapter One

actual interveners into the analysis, I hope to expand the literature on this subject. According to the data collected by Sambanis (2004) and Regan (2000), there have been 151 civil wars in the world (Sambanis 2004) and 63 of these have spurred external interventions. Among those 63 interventions, counter-interventions have been observed in 32 civil wars. More than 1 out of every 5 civil wars taking place between 1944 and 2001 has included interventions on both sides. More than half of the cases of intervention are cases that include both intervention and counterintervention. Thus, without incorporating this important feature of external interventions, we cannot argue that we have fully understood the mechanisms of external intervention in civil wars. As Dunér (1985, 2) points out, it is worth mentioning two points. First, as already shown above, it is generally agreed upon among scholars that internal wars have been the predominant type of conflict since World War II. The frequency of conventional warfare has declined, whereas there has been an increase in internal wars fought within the boundaries of single states. Secondly, it is within civil or internal conflicts that most interventions occur.

1.2 Outline of the Book The goal of the remainder of this book is the development of a framework for understanding what leads to third-party intervention in civil wars and an analysis of intervention patterns that can enhance that understanding. Eight chapters follow this introduction. Chapter 2 discusses the concepts and definitions of civil wars and intervention in this book. As will be discussed, the concept and the definition of civil war have not been agreed upon among scholars. There is no standard concept or definition of civil war. Thus, it is important to clearly indicate which concept and definition of civil war is being used in this book. Also the concept and the definition of intervention in this book should be clarified. Various scholars have considered different forms of involvement under the same heading of intervention. Consequently, without clearly defined terms, any analysis put forth in this book will hardly make sense to the reader. Chapter 3 provides the general literature review. First, the literature regarding the moral and legal discussion of intervention will be reviewed. Second, literature on intervention in civil wars will be reviewed. I will focus on reviewing quantitative empirical studies of third-party intervention in civil wars.

Foreign Military Intervention in Civil Wars

9

Chapter 4 discusses the conceptual modeling of third-party intervention in civil wars. In this section, I will compare the simple model of thirdparty intervention, which is the theoretical conceptualization by Regan (2000), with my extended model of third-party intervention. Sequential characteristics of initial intervention and counter-intervention will be discussed and modeled. Chapter 5 develops a framework for the analysis put forth in this book. With the conceptual modeling developed in Chapter 4, how we approach the intervention issue will be discussed according to the four-dimension framework. Chapter 6 describes the historical evidence and the data that are used in this book. Historical evidence shows how my model and framework for analysis fits the study of the internationalization of civil wars. Chapters 7 and 8 examine the cases of first intervention and second intervention, respectively. The research design and analysis of the firstintervention and second-intervention cases will be discussed in each chapter. Chapter 9 summarizes the findings of this book and concludes with policy implications.

CHAPTER TWO CONCEPTS AND DEFINITIONS

In this chapter, I am going to review the important concepts and definitions necessary for the discussion in this book as well as the general study of intervention. As is the case in many other social sciences, there exists a difficulty in intervention studies with regard to the concepts and definitions used in these studies. For example, the definition of intervention varies according to the ideas of the different authors. Thus, it is important to clarify what definition of each term is used in this book.

2.1 Definitions of Civil Wars We have witnessed the rapid growth of empirical literatures on civil war in part due to the huge increase in the compilation of quantitative data sets. As Gleditsch et al. (2002) note, the Correlates of War project has been dominant for more than three decades as a reliable data source for both external and internal armed conflicts. They argue that one of the virtues of COW data is “its emphasis on strict and transparent operational procedures” (Gleditsch et al. 2002, 617). However, as Sambanis (2004) notes, there is no consensus on how to measure and operationalize civil war. The definition of civil wars varies according to the authors as their specific interests come into play in their studies. Here, I will discuss the several definitions and operationalizations of civil war. Starting from Singer and Small (1972), much literature on civil war has used the Correlates of War (COW) project data set. Small and Singer (1982, 210) define a civil war as “any armed conflict that involves (a) military action internal to the metropole, (b) the active participation of the national government, and (c) effective resistance by both sides.” They distinguish civil (internal or intrastate) war from interstate or extrastate (colonial and imperial) war based on the war’s being internal to the territory of a sovereign state and the government’s role as a combatant. Operationally, they distinguish civil war from other forms of internal armed conflict based on the idea that state violence should be sustained and reciprocated and that the war should exceed a certain numerical threshold in terms of deaths. They adopt the convention of 1,000 fatalities

12

Chapter Two

for an internal armed conflict to be categorized as a civil war. However, this idea has been criticized by several scholars, including Sambanis (2004), Fearon and Laitin (2003), Gleditsch et al. (2002), and Regan (2000), who suggest that the cumulative death criterion presents some problems. For example, the Northern Ireland conflict yielded more than 25 annual battle deaths every year during the period between1971 and 1993, as well as in 1998. That conflict has entailed more than 3,000 casualties in total, but it does not meet the COW threshold of more than 1,000 deaths in a single year. Fearon and Laitin (2003) try to solve the problem presented by the cumulative death threshold of 1,000 by applying a rule that at least 100 deaths must occur every year on average in an ongoing war and that the aggregate death should exceed 1,000. They present the following criteria (Fearon and Latin 2003, 76): (1) They involved fighting between agents of (or claimants to) a state and organized, nonstate groups who sought either to take control of a government, to take power in a region, or to use violence to change government policies. (2) The conflict killed at least 1,000 over its course, with a yearly average of at least 100. (3) At least 100 were killed on both sides (including civilians attacked by rebels). The last condition is intended to rule out massacres where there is no organized or effective opposition.

They also provide the following secondary criteria for dealing with other coding issues (Fearon and Latin 2003, 76): (4) The start year is the first year in which 100 were killed or in which a violent event occurred that has followed by a sequence of actions that came to satisfy the primary criteria. (5) If a main party to the conflict drops out, we code a new war start if the fighting continues (e.g., Somalia gets a new civil war after Siad Barre is defeated in 1991). (6) War ends are coded by observation of a victory, wholesale demobilization, truce, or peace agreement followed by at least two years of peace. (7) Involvement by foreign troops does not disqualify a case as a civil war for us, provided the other criteria are satisfied. (8) We code multiple wars in a country when distinct rebel groups with distinct objectives are fighting a coherent central state on distinct fronts with little or no explicit coordination. (9) If a state seeks to incorporate and govern territory that is not a recognized state, we consider it a “civil war” only if the fighting

Concepts and Definitions

13

continues after the state begins to govern the territory (thus, Indonesia/East Timor 1975, yes, and India/Hyderabad 1947, no).

With the above criteria, Fearon and Laitin (2003) identify 127 civil wars from 1945 to 1999. However, their solution to the death threshold also comes with problems. Moreover, one of the significant differences between Fearon and Laitin’s (2003) work and other data sets is that Fearon and Laitin (2003) include extrasystemic wars. Those extrasystemic wars have completely different characteristics than do traditional internal conflicts. By using a high threshold of deaths, one can distinguish civil wars from riots, terrorism, and some coups. However, this high threshold of death will exclude certain cases that fulfill all other characteristics of civil war but have fewer than 1,000 casualties. Gleditsch et al. (2002, 618–19), as a part of the Uppsala Conflict Data Project, define an armed conflict as “a contested incompatibility that concerns government or territory or both where the use of armed force between two parties results in at least 25 battle-related deaths.” The main reason for them to employ this coding rule is to avoid the problem caused by the high threshold cumulative death criterion. Armed conflict can be divided into three subsets according to their intensity (Gleditsch et al. 2002, 619): x

Minor Armed Conflict: at least 25 battle-related deaths per year and fewer than 1,000 battle-related deaths during the course of the conflict. x Intermediate Armed Conflict: at least 25 battle-related deaths per year and an accumulated total of at least 1,000 deaths, but fewer than 1,000 in any given year. x War: at least 1,000 battle-related deaths per year.

They also distinguish conflicts by type (619): x x

Interstate armed conflict occurs between two or more states. Extrastate armed conflict occurs between a state and a non-state group outside its own territory. (In the COW project, extrastate war is subdivided between colonial war and imperial war, but this division is not used here.) x Internationalized internal armed conflict occurs between the government of a state and internal opposition groups with intervention from other states. x Internal armed conflict occurs between the government of a state and internal opposition groups without intervention from other states.

14

Chapter Two

The last two categories are used to define the term “internal armed conflict.” However, it is very difficult to compare their data set with other data sets that use the cumulative death criterion. If their coding rule is applied, some conflicts will be coded as distinct events that would be coded as one event if the cumulative death criterion were applied. This might be the reason why the correlation coefficient between their data set and other data sets is lower than in the case of any other pairs. To solve the high threshold problem, Regan (2000) relaxes this high threshold of death and defines intrastate conflict as “armed combat between groups within state boundaries in which there are at least 200 fatalities” (21). With the 1,000 death criteria, it is inevitable that more cases of civil war in large countries will be included in the data set, since 1,000 deaths is an absolute threshold. Other than this lower death threshold, Regan’s coding rule is almost identical to COW’s operational criteria. Based on this lower death threshold, Regan (2000) identifies 140 intrastate conflicts during the period between 1944 and 1994. Sambanis (2004) argues that there exist significant differences across civil war lists and that those differences are mainly caused by three questions: “What threshold of violence distinguishes civil war from other forms of internal armed conflict? How do we know when a civil war starts and ends? How can we distinguish between intrastate, interstate, and extrastate wars?” (815). Sambanis (2004) presents criteria for classifying an armed conflict as a civil war (829–31): (a) The war takes place within the territory of a state that is a member of the international system with a population of 500,000 or greater. (b) The parties are politically and militarily organized, and they have publicly stated political objectives. (c) The government (through its military or militias) must be a principal combatant. If there is no functioning government, then the party representing the government internationally and/or claiming the state domestically must be involved as a combatant. (d) The main insurgent organization(s) must be locally represented and must recruit locally. Additional external involvement and recruitment need not imply that the war is not intrastate. Insurgent groups may operate from neighboring countries, but they must also have some territorial control (bases) in the civil war country and/or the rebels must reside in the civil war country. (e) The start year of the war is the first year that the conflict causes at least 500 to 1,000 deaths. If the conflict has not caused 500 deaths or more in the first year, the war is coded as having started in that year only if cumulative deaths in the next 3 years reach 1,000.

Concepts and Definitions

15

(f) Throughout its duration, the conflict must be characterized by sustained violence, at least at the minor or intermediate level. There should be no 3-year period during which the conflict causes fewer than 500 deaths. (g) Throughout the war, the weaker party must be able to mount effective resistance. Effective resistance is measured by at least 100 deaths inflicted on the stronger party. A substantial number of these deaths must occur in the first year of the war. But if the violence becomes effectively one-sided, even if the aggregate effective-resistance threshold of 100 deaths has already been met, the civil war must be coded as having ended, and a politicide or other form of one-sided violence must be coded as having started. (h) A peace treaty that produces at least 6 months of peace marks an end to the war. (i) A decisive military victory by the rebels that produces a new regime should mark the end of the war. Because civil war is understood as an armed conflict against the government, continuing armed conflict against a new government implies a new civil war. If the government wins the war, a period of peace longer than 6 months must persist before we code a new war (see also criterion k). (j) A cease-fire, truce, or simply an end to fighting can also mark the end of a civil war if they result in at least 2 years of peace. The period of peace must be longer than what is required in the case of a peace agreement because we do not have clear signals of the parties’ intent to negotiate an agreement in the case of a truce/cease-fire. (k) If new parties enter the war over new issues, a new war onset should be coded, subject to the same operational criteria. If the same parties return to war over the same issues, we generally code the continuation of the old war, unless any of the above criteria for coding a war’s end apply for the period before the resurgence of fighting.

With these coding rules, Sambanis identifies 145 civil war onsets between 1945 and 1999. As seen in Table 2-1, these various coding rules yield various lists of civil wars. Sambanis (2004) notes that “disagreements over the coded year of onset and termination of civil war may matter for the inferences drawn when we analyze civil war onset, duration, or recurrence using different data sets” (831). These discrepancies between different data sets may also cause inferential problems in the study of third-party intervention in civil wars.

1991–94

1952–52 1992–95

1948–51

Azerbaijan

Bolivia Bosnia

Burma

1968–80 1983–95

1955–55

Argentina

Angola

1962–63 1992– 1975–91 1992–94

1944– 1997 Year 1978–92

Algeria

Country Afghanistan

COW (v 3.0)

Bolivia Bosnia Myanmar/ Burma

Azerbaijan Bangladesh

Argentina

Angola

Algeria

Sambanis (2004) Country Afghanistan

Table 2-1. Comparison of Civil War Lists

16

1948–88 1960–95

1948–51

1952–52 1992–95

1945– 1999 Year 1978–82 1992–96 1996–91 1962–63 1992– 1975–91 1992–94 1997– 2002 1994–99 1955–55 1975–77 1991–94 1974–97

Burma

Azerbaijan Bangladesh Belgium Bolivia Bosnia

Argentina

Angola

Algeria

Fearon and Laitin (2003) Country Afghanistan

Chapter Two

1948–

1955–55 1973–77 1992–94 1976–97 1956–61 1952–52 1992–95

1962–63 1992– 1975– 1992–

1945– 1999 Year 1978–92 1992–

Burma

Bolivia Bosnia

Azerbaijan Bangladesh

Angola

Algeria

Country Afghanistan

Regan (2000)

1968–80 1983–92

1948–

1946–46 1992–

1991– 1971–

1962–63 1993– 1975–91 1992–94

1944– 1994 Year 1978–92 1992–

1966–71 1980–88

1973–73 1946–50 1947–47 1956–59 1967–68

1948–49 1949–62 1984–

1997–97

1948–48

Chile China

Colombia

Congo

Costa Rica

1972–72 1988–88 1991–91 1993– 1970–75 1978–91 1993–97

Chad

Cambodia

Burundi

CongoBrazzaville CongoBrazzaville Costa Rica Croatia

Colombia

China

Central African Rep Chad

Cambodia

Burundi

1948–48 1992–95

1998–99

1993–97

1946–49 1947–47 1950–51 1956–59 1967–68 1948–66 1978–

1965–79 1980–94 1994–97

1996–97

1965–69 1972–72 1988–88 1991– 1970–75 1975–91

Costa Rica Croatia

Congo

Colombia

China

Chad

Central African Rep

Cambodia

Burundi

Concepts and Definitions

1948–48 1992–95

1998–99

1948–62 1963–

1946–50 1950–51 1956–59 1991–

1965– 1994–98

1996–97

1970–75 1978–92

1972–72 1988–88 1993–

Costa Rica

Colombia

China

Chad

Cambodia

Burundi

1948–48

1948–48 1949–62 1984–

1946–50 1947–47 1959– 1980–

1965–72 1978–82 1983–96 1989–

1970–75 1979–91

1972–72 1988–88

17

1979–92 1974–91 1976–77 1978–83 1978–91

1991–94

1944–45 1946–49

Georgia

Greece

1965–65

Dominican Republic

El Salvador Ethiopia

1958–59

Cuba

18

Greece

Georgia

Egypt El Salvador Ethiopia

1991–94

Djibouti Dominican Republic

1991–92 1992–94 1944–49

1994–97 1979–92 1974–91 1978–91 1976–88

1965–65

1958–59 1963–67 1974–74

Cuba Cyprus

Greece

1945–49

1992–94

1945–54 1947–48 1952–54 1953–56 1954–61 1955–60

France

Georgia

1979–92 1974–92 1997–

1965–65

1993–94

1958–59 1974–74

El Salvador Ethiopia

Dominican Republic

Djibouti

Cuba Cyprus

Chapter Two

1944–49 1983–83

Grenada

1981–81 1991–93

1985–89 1992– 1979–92 1960–64 1962–91 1977–85 1987–91 1992–

1965–65

1958–59 1963–64 1974–74 1974–74 1991–93

Greece

Gambia Georgia

Djibouti Dominican Republic Ecuador Egypt El Salvador Ethiopia

Cuba Cyprus

1954–54 1966–72 1970–71 1978–84

1985–

1950–50 1953–53 1956–60

1978–79 1981–82

1959–59 1961–63 1974–75 1985–93

Guatemala

India

Indonesia

Iran

Iraq

Iraq

Iran

Indonesia

1959–59 1961–70 1974–75 1985–96

1989– 1984–93 1989– 1990– 1946–48 1950– 1953–53 1956–60 1976–78 1975–99 1990–91 1999– 2002 1978–79 1979–84

1998–99 1991–95

Guinea-Bissau Haiti

India

1966–72 1978–94

Guatemala

Iran

Indonesia

India

Guinea-Bissau Haiti

Guatemala

Concepts and Definitions

1978–79 1979–93 1959–59 1961–74

1950–50 1953–53 1958–60 1965– 1975–99

1952– 1982–93 1989–

1998–99 1991–95

1968–96

Iraq

Iran

Indonesia

Hungary India

Guatemala

1956–59 1961–66 1974–74 1985–93

1978–79 1981–82

1975–

1950–50 1953–53 1956–60 1963– 1965–65 1975–

1956–56 1954–64 1985–

1954–54 1966–72 1972– 1978–84

19

1960–62 1963–73 1958–58 1975–90

Laos

1989–90 1992–95 1996–96

1979–92

Liberia

Mozambique

Lebanon

1970–70

1996–96

Jordan

20

Morocco

1975–91 1976–92 1973–89 1996–

Moldova

1991-92

Moldova Morocco/ Western Sahara Mozambique Namibia Nepal

1976–95 1997– 1945–46

Nepal Netherlands

1975–88

1992-92

1989-94

1989–96

1958–58 1975–90

1949–50 1960–73

1970–70

Mozambique

Mali

1990–95

Mali

Liberia

Lebanon

South Korea Laos

Jordan

1989–90 1992–97 1999–

1958–58 1975–91

1991–93 1987–97 2000– 1970–71 1963–67 1991–93 1948–49 1960–73

Liberia

Lebanon

Korea Laos

Jordan Kenya

Israel

Chapter Two

Mozambique

Morocco

Malaysia Mali Mauretania Mexico Moldova

Liberia

Lebanon

Laos

Jordan Kenya

Israel

1979–93

1975–

1948–62 1990– 1975– 1994– 1992–

1960–62 1963–73 1958–58 1975–88 1988–90 1989–90 1992–93

1970–70 1992–

1991– 1964–94

1978–79 1982–90

1967–70 1980–81 1984–84

1971–71 1973–77 1994–95

1947–47 1982–95 1950–52 1972–80 1972–92

1989–89 1994–96

Nicaragua

Nigeria

Pakistan

Paraguay Peru Philippines

Romania Russia

USSR

Russia

1994–96 1999– 1944–48 1944–47 1944–50

1947–47 1980–96 1950–52 1972–92 1971–

1988–98

1971–75 1971– 1973–77 1994–99

Oman Pakistan

Papua New Guinea Paraguay Peru Philippines

1967–70 1980–85

1978–79 1981–90

Nigeria

Nicaragua

Russia Russia Russia Russia Russia

Portugal

Paraguay Peru Philippines

Papua New Guinea

Pakistan

Nigeria

Nicaragua

Concepts and Definitions

1946–48 1946–50 1946–47 1946–48 1994–96

1947–47 1981–95 1946–52 1968– 1972–94 1961–75 1962–74 1964–74

1988–98

1971–71 1973–77 1993–99

1967–70

1978–79 1981–88

Romania Russia

Papua New Guinea Paraguay Peru Philippines

Oman Pakistan

Niger Nigeria

Nicaragua

1989–89 1994–

1947–47 1982– 1950–52 1972– 1972–

1988–91

1978–79 1982–90 1990– 1967–70 1967–70 1980–81 1984–84 1986– 1970–75 1971–71 1973–77

21

1982–97

1971–71

Somalia

Sri Lanka

Tajikstan Thailand Turkey

Sudan

1991–96

Sierra Leone

1992–97 1970–73 1991–

1983–

1987–89 1963–72

1983–

1963–64 1990–93 1994–94

Rwanda

22

Syria Tajikistan Thailand Turkey

Sudan

Sri Lanka

South Africa

Somalia

Senegal Sierra Leone

Rwanda

1971–71 1983– 2002 1987–89 1963–72 1983– 2002 1979–82 1992–97 1966–82 1984–99

1989–99 1991–96 1997– 2001 1988–91 1991– 1976–94

1944–48 1963–64 1990–93 1994–94

1992–97 1977–80 1984–99

Turkey

1983–

1987–89 1963–72

1983–

1971–71

1981–91 1991– 1983–94

1989– 1991–

1999– 1962–65 1990–

Tajikstan

Sudan

Sri Lanka

South Africa

Somalia

Senegal Sierra Leone

Russia Rwanda

Chapter Two

Tajikstan Thailand Turkey

Sudan

Spain Sri Lanka

South Africa

Somalia

Rwanda

1992–94 1965–85 1984–

1983–

1987–89 1963–72

1982–

1982–91 1991– 1970–94 1990– 1968– 1971–71

1963–64 1990–94 1991–92 1993–94

1960–65 1994–94

1948–48 1962–69 1986–86 1991–92

Vietnam Yemen

Yemen AR

1960–65 1993–93 1996–97

1972–79

Zaire

Zimbabwe

Yemen PR Yugoslavia

1966–66 1980–88 1996–

Uganda

Zimbabwe

Congo-Zaire

Yemen PR Yugoslavia

Yemen AR

Vietnam Yemen

United Kingdom

Uganda

1948–48 1962–70 1986–86 1991–91 1998–99 1960–65 1967–67 1977–78 1996–97 1998– 2001 1972–79 1983–87

1960–75 1994–94

1971–98

1966–66 1978–79 1981–87 1990–92 1995–

Zimbabwe

Dem. Rep. of Congo

Yemen PR Yugoslavia

Yemen AR

Vietnam Yemen

UK

Uganda

Concepts and Definitions

1972–79 1983–87

1960–65 1977–78 1996–97 1998–

Zimbabwe

Zaire

Yugoslavia

Yemen AR

Vietnam Yemen

1952–56 1969–99 1960–75 1994–94 1948–48 1962–69 1986–87 1991–91

UK

Uganda

1950–56

1981–87 1993–

1972–79 1980–88

1992–

1960–65 1967–67 1977–77 1978–79

1991–92

1960–65 1994–94 1986–86 1948–48 1962–67

1969–

1966–66 1971–72 1980–86 1986–88

23

Chapter Two

24

These significant disagreements among different civil war lists make it imperative to discover how the results differ if we use different lists of civil wars. As we have already seen, Regan (2000) identifies 140 intrastate conflicts, of which 90 included at least one instance of third-party intervention. That means that about 64 percent of intrastate conflict included external involvement. However, according to Gleditsch et al. (2002), there have been 163 internal conflicts, of which 32 have included external participation by other states and 131 have not. This figure represents only 19.6 percent of total internal conflicts from 1946 to 2001, which is very different from the previous figure. As seen in Table 2-2, the correlation coefficients of these widely used civil war data sets range from as low as 0.51 between Gleditsch et al. (2001) and Sambanis (2004) to as high as 0.80 between Fearon and Laitin (2003) and Sambanis (2004). Table 2-2. Correlations among Civil War Lists, 1960–1993 COW 2000

Gleditsch et al. (2001)

COW 2000 1.00 Gleditsch et 0.46 al. (2001) Fearon and Laitin 0.70 (2003) Regan 0.70 (1996) Sambanis 0.74 (2004) Source: Sambanis (2004, 832)

Fearon and Laitin (2003)

Regan (1996)

Sambanis (2004)

1.00 0.54

1.00

0.46

0.67

1.00

0.51

0.80

0.72

1.00

In this book, I am going to use Sambanis’s list of civil wars for my analysis for several reasons. First, it is more reasonable to exclude extrastate conflicts from civil war lists because the characteristics of extrastate conflicts are significantly different from those of traditionally defined civil wars. Based on this reasoning, Fearon and Laitin (2003) will be not used. Second, Sambanis (2004) is more suitable for the analysis in that Gleditch et al. (2001) subdivide conflicts into multiple small conflicts just because of their annual number of casualties. If the data by Gleditch et al. (2001) are used, it is almost impossible to determine when a particular civil war ended.

Concepts and Definitions

25

2.2 Definition of Intervention Several decades ago, Rosenau (1969, 152) was concerned with the fact that “there appears to be no agreement whatsoever on the phenomena designated by the term.” According to him, the problem of the multiple definitional options with regard to the concept of intervention, it is usually merge of several definitions, which ends up as “any action whereby one state has an impact upon the affairs of another” (153). He argues that “often intervention is defined in such a general way that it appears to be synonymous with imperialism, aggression, colonialism, neocolonialism, war, and other such gross terms that are used to designate the noncooperative interactions of nations” (153). Sometimes the definition of intervention is developed based on a normative sense of the issue. As Damrosch (1993) notes, “In the discipline of international law, ‘intervention’ generally means an improper interference by an outside power with the territorial integrity or political independence of a state.” Otte (1995) also defines the concept of intervention in a similar way. According to him, “Military intervention is, therefore, an act by states and it is concerned with their perceived national interests and foreign policy objectives. Its purpose is the same as that of all other form of foreign policy: it is intended to make the target state do what the intervener wants it to do but what it might otherwise not have done” (6). Bull (1984) also seems to agree with this definition. He argues that “a basic condition of any policy that can be called interventionary in this sense is that the intervener should be superior in power to the object of the intervention: it is only because the former is relatively strong and the latter relatively weak that the question arises of a form of interference that is dictatorial or coercive” (1). Thus, according to him, in most cases in the modern history of intervention, the interveners have been the great powers. However, these normative definitions of intervention are not appropriate for use in a systematic and quantitative study of intervention. Moreover, as Rosenau (1968) points out, these normative ways of defining intervention make it hardly possible to distinguish the concept from the notion of influence. As Regan (2000) notes, there exist various types of third-party intervention. There also exists a range of intervention strategies that may be employed by the interveners. Yoon (1997) argues that intervention “entails different levels of actions ranging from verbal statements, economic assistance or the withholding of economic assistance, initiation or increase of arms supply, deployment of advisers, deployment of combat personnel into a war zone, to actual military engagement in combat

Chapter Two

26

operation. Thus, its concept involves both nonmilitary and military activities” (585). Since the subject of this book is military intervention, I will proceed to discuss the definition of military intervention. Dunér (1983) defines intervention as “acts which affect one party’s military fighting or capacity to fight the opposing party” (60). Table 2-3. Instruments and Levels of Involvement I.

II.

III.

IV.

V.

VI.

Direct combat involvement a.

Invasion (regular)

b.

Specialist functions

Indirect combat involvement c.

Invasion (irregular)

d.

Shelling

Direct para-combat involvement e.

Advisory functions

f.

Arms supply

Indirect para-combat involvement g.

Military training

h.

Armed blockade

i.

Financial support

Direct supporting activities j.

Military warning

k.

Transport

l.

Base functions

Indirect supporting activities

Concepts and Definitions

27

However, it is not clear what kinds of actions should be included as military intervention. Military intervention can be categorized according to the form or instrument of involvement and the level of intervention, which Dunér (1983) determines based on the closeness or immediacy of the intervention on a battle situation. Using these criteria regarding the form of involvement and the level of intervention, he distinguishes between combat involvement, para-combat involvement and supporting activities. Each category can be dichotomized according to its degree of closeness (direct and indirect activities). Thus, his categorization of military intervention yields a six-echelon ladder that ranges from indirect, supportive intervention at the bottom to direct combat involvement at the top level. He presents brief definitions of each instrument (Dunér 1983, 60–61): Invasion refers to combat troops sent to the country in which the civil war is taking place. There are three alternative, operational criteria: at least 800 troops, 50 tanks or eight combat aircraft are sent in. the invasion units are either part of the intervener’s regular forces or irregular forces, for instance, mercenary troops under the intervener’s command. Specialist functions are fulfilled by specialized military personnel who are incorporated in the weapon systems of a party to the conflict. They may also be incorporated in some other intervener’s weapon systems, but personnel managing the intervener’s own weapon systems are excluded. At least 100 helicopter or airplane pilots must have been dispatched, or at least 200 tank operators. Shelling means heavy bombardment on at least 5 separate days over a period of 10 days in the country in which the civil war is taking place. While invasion refers to units stationed in the country, shelling is carried out by forces not stationed there (and means bombardment from neighboring territory, from the sea or from the air). Advisory functions refer to the control of radar systems, the running of military airports, the setting up of communication systems and the like. More than 100 advisors must have been dispatched by the intervener. Arms supply refers to the donation of weapons, the value of which must exceed $ 4 million (1975 monetary value). Weapons which are given away or for which only a token payment is required or which are loaned, are included, whereas commercial transactions are not relevant. Military instruction of the forces of a party to the conflict is military training. At least 800 soldiers must have been instructed; alternatively,

28

Chapter Two more than 100 instructors from the intervener state must have been engaged in the training. Armed blockade refers to military surveillance, in order to control passage to and from the country, notably in controlling gun-running, transferring equipment, military personnel, etc. It is not the same thing as reconnaissance, since the use of armed force is implied. Financial support means monetary assistance, i.e., donations, to finance a party’s weapon purchases or the hiring of mercenaries. The value of this assistance must exceed $4 million (1976 monetary value). A military warning is a verbal threat to enter the internal conflict, backed up by a military move (involving at least the forces corresponding to an invasion). Whether a military warning is followed by subsequent invasion is an empirical question. Also the definition does not preclude invasion taking place before the warning. When one state assists another state in moving the latter’s forces to the battlefield that is transport. The transporting country must be actively involved, not only permitting transit through its territory. The services must refer to manpower, not weapons only. Involvement in terms of base functions refers to a host country admitting (or at least not protesting about) the concentration of the military personnel of a party to the conflict on its territory. This personnel must not be in transit only. It may be ready for combat or under training. In operational terms, there are there alternative criteria: at least 800 soldiers in the base country, 50 tanks belonging to the party, or more than 100 instructors training the party’s soldiers.

Pearson (1974b) defines foreign military intervention only as “the movement of troops or military forces by one independent country, or a group of countries in concert, across the border of another independent country (or colony of an independent country), or actions by troops already stationed in the target country” (261). In his study, he includes only intervention in which direct military action was undertaken and, thus, he excludes long-term relatively inactive encampment on bases. Pearson (1974a) argues that the definitions of intervention in this kind of study should refer to some form of overt and relatively easily detectable behavior that researchers can actually observe. Freedman (1994) similarly defines intervention as “the use of armed force to influence the character and course of a developing conflict which is neither taking place upon nor directly threatening national territory, and does not touch upon any

Concepts and Definitions

29

specific obligations to allies” (3). These definitions of foreign military intervention are very clear and avoid any ambiguity, but they are too narrow and constraining for this study. Regan’s (2000) definition of intervention is similar to those of these scholars. Reagan (2000) defines an act of intervention as “conventionbreaking military and/or economic activities in the internal affairs of a foreign country targeted at the authority structures of government with the aim of affecting the balance of power between the government and opposition forces” (10). Since the focus of this book is foreign military intervention, I have discarded the elements of his definition that refer to economic intervention. In his definition of military intervention, he includes military troops, supplies, intelligence, or aid that can serve to help the existing forces of either the government or the opposition (Regan 2000, 9). Thus, his data set includes all those cases as the cases of intervention. However, military instruction and training, providing intelligence, and arms supply will impose much less significant costs for the potential interveners than will their sending out uniformed military forces. Thus, based on this perspective, foreign military intervention is defined in this book as the active and overt involvement of a country in foreign civil wars through the deployment of troops or naval or air forces, as well as through the provision of logistical aid, military advisers, or military forces.

CHAPTER THREE LITERATURE REVIEW

3.1 Introduction One area of civil war studies to which much attention has been recently paid is the internationalization of civil wars. Scholars have studied various aspects of the internationalization of civil wars. Until recently, however, the literatures on intervention have been more concentrated on the moral and legal issues of intervention (Atack 2002; Bull 1984; Chesterman 2002; Cronin 2007; Damrosch 1993; Hoffmann 1984, 1996; Lillich 1973; Little 1975; Ramsbotham and Woodhouse 1996; Smith 1994; Walzer 1977; Welsh 2004). It is only recently that scholars have started to ask the question of when countries are likely to intervene in other states’ internal conflicts. Recent studies have found that third-party intervention is an important element to consider in explaining both foreign policy decision making and civil conflict resolution. Researchers have argued that intervention is a key explanatory variable with regard to civil conflict duration (Regan 2002) and outcome (DeRouen and Sobek 2004). More related to the subject of my book is the explanation that Regan (2000) and Lemke and Regan (2004) provide regarding the conditions under which civil wars include intervention. Since this book is about the systematic empirical analysis of foreign military intervention in civil wars, more attention will be paid to reviewing those studies that include quantitative analyses of intervention. Moral and legal issues of intervention will be reviewed as well in the next section.

3.2 Reviews on Non-intervention Norms As noted above, the vast majority of the literature on intervention (not only limited to intervention in civil wars) is about the moral and normative aspects of foreign intervention in the affairs of other sovereign states. Traditionally, there has been a strong interest in the legal and moral aspects of intervention (Dunér 1985). From the legal point of view, there is

32

Chapter Three

consensus around the idea that intervention, especially unilateral intervention by foreign states, is prohibited by the norms of the international community. As Bull (1984) and Hoffmann (1984) note, intervention is generally believed to be legally and morally wrong. Sovereign states or independent political communities have the right to demand that their spheres of jurisdiction be respected. If one accepts the principle of sovereignty as the core principle of international politics, this will lead one to conclude that intervention, defined as “an act aimed at influencing the domestic affairs of a state” (Hoffmann 1984, 11), has to be illegitimate. Thus, intervention in the affairs of sovereign states will be considered dictatorial interference, which interferes with that right. This kind of normative objection to intervention is not limited to those interventions based on national interest, as occurred frequently during the Cold War period. The issue of humanitarian intervention has also generated one of the most heated discussions in international relations among both theorists and practitioners (Welsh 2004b). Again, the main subject of the debate is the almost irreconcilable tension between the principle of state sovereignty, which is the most important principle of the United Nations system and international law, and the evolving international norms related to human rights and the use of force. The tradition of this non-intervention norm was developed only in the twentieth century. As Akehurst (1984) explains, in the nineteenth century there were no such rules of international law that forbade a state to mistreat its own nationals. If ill treatment occurred in another state, other foreign states often claimed the right to use force to prevent such mistreatment. This right can be interpreted as the right of humanitarian intervention. Historically, this kind of intervention was exercised on a number of occasions by European states, mainly to prevent Turkey from ill-treating its Christian subjects (Akehurst 1984, 95). However, according to Trachtenberg (1993, 25), traditional justifications for intervention, such as the goal of securing economic centers or natural resources, acting to restore order, or protecting one’s own nationals, “gradually lost legitimacy” after 1945. Since the adoption of the United Nations Charter in 1945, international law has developed a body of rules on human rights that forbids states to mistreat individuals, including their own nationals—but at the same time, it has developed a body of rules restricting the right of states to use force (Akehurst 1984). Article 2 (4) of Chapter 1 of the United Nations Charter describes the prohibition of the use of force, which can lead to the non-intervention norm. This Article clearly requires members of the United Nations to

Literature Review

33

“refrain in their international relations from the threat or use of force against the territorial integrity or political independence of any State, or in any other manner inconsistent with the purpose of the United Nations” (Charter of The United Nations). Thus, legitimacy of intervention is dependent on how we interpret Article 2 (4). Stone (2006) interprets Article 2 (4) very narrowly. He argues that the use of force for a wide variety of purposes (for instance, as a sanction in response to a breach of international law) is not aimed at the territorial integrity or political independence of any state. Thus, he argues that the use of force in those cases is not inconsistent with the purpose of the United Nations (Stone 2006, 43). If we follow this interpretation, it is easy for us to view the use of force for purposes of humanitarian intervention as lawful. Reisman and McDougal (1973, 177) also argue that “since a humanitarian intervention ‘seeks’ neither a territorial change nor a challenge to the political independence of the State involved and is not only not inconsistent with the purposes of the United Nations but is rather in conformity with the most fundamental peremptory norms of the Charter, it is distortion to argue that it is precluded by Article 2 (4).” However, most scholars agree that violations of political independence are not limited to cases where a state is annexed or has a change of government imposed on it by another state, and that violations of territorial integrity are not limited to cases where a state is deprived of part of its territorial integrity. Thus, any instance of humanitarian intervention, however limited, constitutes a temporary violation of the target state’s political independence and territorial integrity if it is carried out against that state’s wishes. Based on this interpretation, most authors consider Article 2 (4) as imposing a total ban on the use of force in international relations except when another provision of the Charter expressly recognizes or creates an exception to that ban. With regard to intervention in civil wars, Gow (1994, 29–30) argues that “civil wars have been conventionally (to some extent prudentially) judged to be matters internal to the state and, therefore, subject to the general proscription of interference in domestic affairs.” There has been heated debate with regard to intervention in civil wars. Surprisingly, many debates at the United Nations have been similar to the research subjects of the scholars who have tried to determine the factors that prompt foreign intervention in ethnic civil wars. Davis and Moore (1997) and Moore and Davis (1998) find that shared ethnic ties between a group that controls a state and its kin who are disadvantaged in a

34

Chapter Three

neighboring state lead to international tension between the two states (Moore and Davis 1998, 89). With a view that is very much related to that put forth by Davis and Moore (1997) and Moore and Davis (1998), Woodwell (2004) also agrees about this relationship between shared ethnicity and international conflict. He argues that there exists a strong and significant increase in dyadic conflict when two states share an ethnic group and an ethnic majority exists in at least one of the states. Carment and James (2003) and Carment et al. (2006) also contend that the ethnic composition of two states plays a key role in determining the tendency toward third-party intervention in ethnic civil wars. Legal issues with regards to the intervention on the part of states with ethnic ties were vehemently debated by the members of the United Nations. There was a debate about the definition of aggression in the Sixth Committee of the General Assembly in 1954. At that time, delegates from Greece and the Netherlands argued in favor of a right of humanitarian intervention in cases where there were ethnic ties between the intervening state and a racial minority suffering persecution in another state, which is the exact research subject of Davis and Moore (1997) and Moore and Davis (1998). In that debate, delegates from Israel, China, and Panama disagreed with this position. After its involvement in the Greek civil conflicts in 1947, Yugoslavia tried to justify its support for the rebels in Greece. They argued that as Slavo-Macedonian minorities were persecuted by Greece, they could not be indifferent to that crisis. However, a Commission of Investigation appointed by the United Nations stated that such persecution did not absolve Yugoslavia of responsibility for its assistance to the rebels (Akehurst 1984, 95). Then, our next question will be whether there is any other provision of the United Nations Charter that could be interpreted as authorizing humanitarian intervention as an exception to the general prohibition on the use of force in Article 2 (4). Article 39 of Chapter VII (Action with respect to threats to the peace, breaches of the peace, and acts of aggression) states that “The Security Council shall determine the existence of any threat to the peace, breach of the peace, or act of aggression and shall make recommendations, or decide what measures shall be taken in accordance with Articles 41 and 42, to maintain or restore international peace and security.” Article 41 states: “The Security Council may decide what measures not involving the use of armed force are to be employed to give effect to its decisions, and it may call upon the Members of the United Nations to apply such measures. These may include complete or partial interruption of economic relations

Literature Review

35

and of rail, sea, air, postal, telegraphic, radio, and other means of communication, and the severance of diplomatic relations,” while Article 42 states: “Should the Security Council consider that measures provided for in Article 41 would be inadequate or have proved to be inadequate, it may take such action by air, sea, or land forces as may be necessary to maintain or restore international peace and security. Such action may include demonstrations, blockade, and other operations by air, sea, or land forces of Members of the United Nations.” Thus, under Chapter VII of the Charter, the Security Council has a very expansive degree of discretion in choosing to declare the existence of a threat to the peace and to take military or non-military action to remove that threat. The Security Council could therefore declare that violations of human rights in a particular country constituted a threat to the peace and could take military action (or authorize member states to take military action) to terminate such violations of human rights. With these Articles in the United Nations Charter, multilateral intervention authorized by the United Nations can be justified. According to Bull (1984), there is much support for the proposition that intervention may be justified when it is collectively authorized by the international community itself through an international organization, general or regional. Conversely, it means that unilateral intervention is still hard to justify. There is another case to consider with regard to the non-intervention norm. In 1960, shortly after the Congo became independent, Belgium deployed its troops in the Congo in order to protect Belgian interests and the lives of Belgian and other nationals. In the Security Council debates that followed the Belgian action, Belgium claimed that international law permitted the use of force for such purposes, and the Belgian claims were supported by Italy, the United Kingdom, France, and Argentina; Belgium, Italy, the United Kingdom, and France expressly described the Belgian action as a form of humanitarian intervention. The Congo, however, considered the Belgian actions to constitute an act of aggression because they had been carried out without the consent of the Congolese government; this view was supported by Tunisia, the Soviet Union, Poland, and Ecuador. In 1964, Belgium and the United States sent forces to the Congo to save the lives of hostages of many nationalities (including Belgian, United States, and Congolese nationals) from Congolese rebels. Statements by Belgium and the United States emphasized that the operation had been carried out with the consent of the Congolese government.

36

Chapter Three

Nigeria and Bolivia said that the action by Belgium and the United States was lawful only because it had been carried out with the consent of the Congolese government. So it is possible that they would have regarded the action as unlawful if the Congolese government had not given its consent. As seen in the above case, it is widely recognized by international lawyers and theorists that there are exceptions to the rule of non-intervention (Bull 1984). As in the case of the 1964 intervention in the Congo by Belgium and the United States, some instances of intervention are permissible when they are at the invitation of an incumbent government. Contrary to the principle of sovereignty at the expense of human rights in the United Nations, the Constitutive Act of the African Union (AU) allows for intervention without the consent of the target state. Articles 4 (h), 4 (j), and 23 (2) of the Constitutive Act of the African Union provide the legal basis for a “potentially unprecedented power of intervention” (Cillers and Sturman 2002, 1). Article 4 (h) provides for “the right to intervene in a Member State pursuant to a decision of the Assembly in respect of grave circumstances, namely: war crimes, genocide and crimes against humanity.” Article 4 (j) provides for “the right of Member States to request intervention from the Union in order to restore peace and security.” Article 23 (2) states that “any Member State that fails to comply with the decisions and policies of the Union may be subjected to other sanctions, such as the denial of transport and communications links with other Member States, and other measure of a political and economic nature to be determined by the Assembly.” Thus, at least on paper, “sovereignty no longer provides protection behind which predatory governments can hide abuse of their peoples” (Cillers and Sturman 2002, 1). Some other scholars also try to provide justification for intervention. These scholars contest the issue of what conditions make foreign intervention justifiable. Trachtenberg (1993) argues that the danger of nuclear proliferation can justify intervention. Farer (1993) argues that severe humanitarian crises and accusations of genocide are justifiable causes of external involvement. Bull (1984) argues that when it takes the form of counter-intervention, or intervention that assists a state in repelling an intervention that some other party has already begun, intervention can be justified. In sum, even for humanitarian purposes, unilateral intervention will be very hard to justify under the current international norm. Only multilateral interventions that are authorized by the Security Council are justifiable.

Literature Review

37

3.3 Previous Studies on Intervention in Civil Wars There have been various studies (Brown 1996; Bryman et al. 2001; Carment and James 1995, 1997, and 2003; Carment, James, and Rowlands 1998; Carment, James, and Taydas 2006; Chazan 1991; Dunér 1985; Kaufmann 1996; Lake and Rothchild 1998; Lemke and Regan 2004; Leurdijk 1986; Mitchell 1970; Pearson 1974; Regan 1996, 2000, and 2002; Ryan 1995; Zartman 1992) on the likelihood of external intervention in civil wars. These studies can be grouped broadly into three categories: issues of identity, issues of domestic politics, and issues of geopolitics. First, studies focusing on issues of identity as a key variable in explaining external intervention see that close ties or a close relationship between a potential intervener and a target state constitute one important explanatory variable. These studies hypothesize that third-party states will be more likely to intervene in civil wars when there exists an affective tie resulting from the potential interveners and target states’ having either ethnic, religious, or ideological ties. Some scholars argue that an ideological affinity with either a government or a rebel group will increase the likelihood of external intervention (Leurdijk 1986; Mitchell 1970; Pearson 1974; Zartman 1992). More specifically, scholars have hypothesized that external intervention is more likely when there exists ethnic or religious ties between a third-party state and the rebel group (Brown 1996; Carment, James, and Rowlands 1997; Chazan 1991; Lake and Rothchild 1998; Leurdijk 1986; Mitchell 1970; Pearson 1974; Zartman 1992). Ideological closeness is also believed to have an impact on decisions regarding intervention (Zartman 1992). Second, issues of domestic politics are believed to have an impact on the decision to intervene in civil wars. Scholars argue that the decision to intervene in another country’s civil war is substantially influenced by the domestic political environment. Since most decision makers are also politicians, they are not free from considering the political consequences of their chosen actions. Thus, when deciding whether or not to intervene, the leader of a third-party state should consider the consequences of his or her action (or inaction) on the security of the regime and its political position in domestic politics. Carment and James (1997) and Leurdijk (1986) argue that institutional constraints on the leader of a third-party state are significantly related to the probability of intervention. Related to these arguments, Lemke and Regan (2004) also argue that potential interveners with higher level of democracy are less likely to intervene in another country’s civil war.

38

Chapter Three

Third, in a move very much related to realist perspectives, scholars have identified geopolitical variables as important determining factors in the decision to intervene in foreign civil wars. Among the many geopolitical variables considered to have an impact on calculation of costs and benefits related to intervention, the variable that has the most prominent impact on the decision to intervene is the potential intervener’s geographical proximity to the civil war state (Dunér 1983; Lake and Rothchild 1998; Lemke and Regan 2004; Leurdijk 1986; Luard 1972; Mitchell 1970). Another important variable in this regard is the major power status of the potential intervener (Lemke and Regan 2004). Related to this, the balance of military forces between the potential intervening state and the civil war state is also considered important (Elbadawi 2000; Leurdijk 1986; Pearson 1974). Whether a civil war occurred during the Cold War period or not is also believed to have had an impact on the decision making of third-party states with regard to intervention (Regan 2000, Lemke and Regan 2004). The strategic value of the civil war state, in terms of either military considerations or economic considerations, is also important (Brown 1996; Carment, James, and Rowlands 1997; Chazan 1991; Pearson 1974; Ryan 1995). Related to this, military or political ties (or regime similarities) between the potential intervening state and the target state are also considered to have an impact on intervention decisions (Brown 1996; Bryman et al. 2001; Carment, James, and Rowlands 1998; Lemke and Regan 2004; Leurdijk 1986; Mitchell 1970; Pearson 1974; Ryan 1995; Zartman 1992). Regan (1998, 2000) derives a hypothesis that a third-party state should be less likely to intervene in a more intense civil war. On the contrary, Pearson (1974) and Elbadawi (2000) argue that because of humanitarian considerations, the intensity of a civil war lead to more external intervention. Even though they share the same view that the intensity of the civil war has an impact on the propensity toward external intervention, their argument is different with regard to whether that relationship is positive or negative. As discussed above, scholars have proposed many theories (or hypotheses) with regard to the impact of the variables discussed above on the decision to intervene. However, not many studies have analyzed the whole civil wars. Many studies have focused on ethnic civil wars (Brecher and Wilkenfeld 1997; Carment and James 2003; Carment, James, and Taydas 2006; Carment and Rowlands 1998; Cetinyan 2002; Fearon and Laitin 1996; Heraclides 2001). They exclude identity civil wars from their analysis. Moreover, as pointed out in the introduction of this book, there has been very little quantitative empirical analysis of this specific topic. This

Literature Review

39

lack of rigorous empirical analysis is the largest deficiency in the literature on intervention in civil wars. I will discuss several important studies that are related to the subject of this book, even though they are neither exactly quantitative empirical analyses nor about intervention in civil wars. Then, I will also discuss and critique previous quantitative empirical analyses of intervention in civil wars. There are several ways that scholars have analyzed intervention in civil wars. Kaufmann (1996) focuses on the different characteristics of ethnic and ideological civil war and its implications for external intervention. Some scholars (Feste 1992, 2003; Pearson, Baumann, and Pickering 1994; Tillema 1994) focus on the realist perspective on intervention. Feste (1992) discusses the impact of the Cold War rivalry between the United States and the Soviet Union on intervention. Tillema (1994) tries to explain the impact of alliances during the Cold War period on overt military intervention. Pearson, Baumann, and Pickering (1994) also discuss the importance of the realist perspective in understanding military intervention. Some scholars investigate the impact of ethnic ties on international conflict (Davis and Moore 1997; Moore and Davis 1998). In a move related to those studies, some scholars (Carment and James 2003; Carment, James, and Taydas 2006) focus on ethnic composition and domestic political configuration as they relate to the likelihood of external intervention. These studies will be reviewed briefly. Kaufmann (1996) compares ethnic and non-ethnic civil wars and, then, explains what can be expected if intervention actually occurs in each conflict in terms of costs and the probability of achieving the goal of the intervention. Even though his reasoning and hypotheses are not empirically tested, his comparison of ethnic and non-ethnic civil war with regard to the decision calculus of potential interveners sheds light on the studies of this subject. Actually, his study is more about policy prescription, arguing as it does that intervention in ethnic conflicts is much easier and less costly for the American government. The important aspects of intervention in civil wars are the cost of intervention and the political consequences of intervention failure. His argument starts from the view that Americans fear intervention in foreign civil wars because of America’s failure in Vietnam. Opponents argue that “any military intervention in a foreign civil war must inevitable become an unwinnable quagmire whose escalating cost would ultimately cause the do-gooders to retreat, badly damaging both the international credibility and the domestic cohesion of the intervening powers, while leaving the local victims at least

40

Chapter Three

as badly off as before” (Kaufmann 1996, 62). Because of this fear, they do not want to fully commit U.S. ground troops. However, Kaufmann (1996) argues that intervention in ethnic civil wars is completely different from intervention in ideological civil wars. He denies the conventional wisdom that outside intervention in civil wars can have only limited leverage on their outcomes. He points out that the major difference between ethnic civil war and ideological civil war is “the importance of loyalty competition” (63). The core of ideological civil wars is the competitions between the government and the insurgents for the loyalties of the people, whereas territorial control is relatively unimportant. On the contrary, in ethnic wars, neither side can recruit members of the other group because ethnic identities are fixed at birth. Thus, political competition for individual loyalties is not important in ethnic wars. Control of territory is more important, and conflict outcomes are determined mainly by the balance of military power between the warring factions. Thus, according to Kaufmann (1996, 64), “Foreign aid or foreign troops can make a tremendous difference to the local balance of forces and, unlike in ideological counterinsurgencies, outsiders can reliably tell friend from foe.” Kaufmann (1996) concludes that in ethnic conflicts, outside aid can be decisive. In particular, “military aid, which has limited potential in ideological conflicts, is far more valuable in ethnic wars. Arms transfers and training can greatly increase the recipient community’s military strength. Foreign military purchases and advisers dramatically reversed the military balance…” (88). Feste (1992) argues that superpower intervention can be understood from two distinct perspectives, “one devoted to U.S. and Soviet behavior in their spheres of influence, and another to spheres of competition, specifically those regions that have vied for the attention and interest of the United States and the Soviet Union” (6). She, of course, considers other factors such as capabilities and viable strategies open to each side as it seeks to assert its influence in different domains, as well as various domestic and international constraints on each side’s definition of its interests in operational terms. However, according to Feste (1992), the most important mechanism of intervention by two superpowers during the Cold War period is related to the maintenance of a strategic balance during the Cold War. Borrowing Payne’s (1970, 64) argument about a conception of “ours” and “theirs” versus “not ours” and “not theirs” that both sides can recognize, reference to the status quo is very important in understanding intervention by superpowers. If either side employs its forces to attack the other (this does

Literature Review

41

not mean that the United States directly attacks the Soviet Union or vice versa), the status quo is then violated, which means aggression is occurring. When the status quo is violated, the superpowers on each side are entitled to prevent the other from effecting such changes. Consequently, if any changes in governments or regimes that may occur do not affect the Cold War status quo, intervention will not occur. However, “if the internal violence appeared to lead to a change in the Cold War status quo, then superpower involvement would follow if the problem could not be handled locally and if the affected superpower believed that essential action was necessary to maintain a credible deterrent threat at the global level” (Feste 1992, 27). The implications of this conclusion will be discussed in more detail later in Chapter 4 and 5. Pearson, Baumann, and Pickering (1994) also discuss the motives of intervention by focusing on the importance of a realist perspective in understanding military intervention. Their study is not exactly about intervention in civil wars. However, their study merits reviewing here because meaningful hypotheses can be drawn from their study. They base their argument on traditional realpolitik by arguing that strategic interest is given precedence over others in terms of its importance. In their analysis of international military intervention, they examine two sets of hypotheses concerning the demonstration of power and system stability. The first hypotheses are concerned with the probability and frequency of military intervention among states with differing amounts of power. They expect major powers to use force more frequently than the smallest powers because of their relative capabilities and interests. In general, they hypothesize that if realpolitik applies, intervention patterns should be unvarying over time and intervention will generally be executed by the strong and directed toward the weak, both globally and regionally, with a predominance of hostile intervention (208). They analyze a data set on international military intervention for the Cold War period of 1946–1988. These data include over 600 cases of intervention. They find that “military intervention and therefore the practice of realpolitik continue as prominent features of the international system” (209). Over the entire period covered in this study, major powers such as the United States, Britain, and France were among the most frequent interveners. Moore and Davis (1998) explore the extent to which shared ethnic ties between groups that are in control of country A and groups that are disadvantaged in neighboring country B lead to international tension between the two countries. They explore whether ethnicity influences the intervention of neighboring countries.

42

Chapter Three

They emphasize shared ethnicity in determining intervention by third party states because ethnic ties among peoples across state borders in the international system function as unstated alliances among those people. Transnational ethnic ties may represent an opportunity for elites in many societies to mobilize political support by using ethnic appeals to vilify a rival state for its treatment of ethnic brethren. Also, the elites of the ethnic minority in the second state have an interest in forging a relationship with the first state when an irredentism option is feasible. In their earlier study, Davis and Moore (1997) found that there is a connection between dyadic conflict between two neighboring states and the presence of a disadvantaged ethnic minority in one state when members of that same minority group are in power in the other. To explain the linkage between a disadvantaged minority in one state and that minority group in power in the other, they contend that it is useful to conceptualize ethnic linkages among people across state borders as functionally equivalent to alliances between two states. They are not simply arguing that some kind of ethnic affinity leads to an alliance-like relationship between ethnically homogeneous elites. Rather, they are contending that they expect the two groups to share policy preferences and, therefore, geopolitical preferences. In other words, conflict between a state and an ethnic group will escalate to the international level when other elite members of that same ethnic group play a role in policy-making in another state and that state finds the first state to be politically relevant, given its resource constraints. Their model considers not only conflicts but also all foreign policies because they believe that ethnic ties should matter across all foreign policy areas. They argue that ethnic alliances lead to greater hostility, especially when the disadvantaged group is politically mobilized. However, they do not explain how to measure whether the disadvantaged group is politically mobilized. As discussed by Lobell and Mauceri (2004), many previous studies treated the state as a unitary actor that pursues its national interests and considers international relations as the interaction of sovereign states; they ignored the domestic political environment that characterizes most divided states. Carment, James, and Taydas (2006) and Carment and James (2003) point out the intellectual gap regarding intervention in ethnic wars and emphasize a comprehensive explanatory framework combining both international and comparative politics. Thus, they switch the focus to institutional constraints on decisions. Far less attention has been paid to the characteristics of intervening states. They argue that it is necessary to look at what factors might constitute constraints within intervening states.

Literature Review

43

They criticize previous studies for not being able to explain why some states are more likely to intervene than other states. Thus, they argue that it is useful to consider the possibility that affective ethnic ties and elite incentives are filtered through domestic institutions to yield differential foreign policies with respect to ethnic conflicts in other states. Carment, James, and Taydas (2006) and Carment and James (2003) develop their propositions based on the same model. In both studies, determinants of ethnic interventionism include: (1) ethnic domination versus diversity—the extent to which institutions of the state are in the hands of a single ethnic group; and (2) high versus low institutional constraints—the extent to which leaders enjoy discretionary power over state policies. Based on this model, they predict different types of intervention policies: (1) ethnically dominant, low-constraint states are predisposed to bellicose policies of ethnic intervention; (2) ethnically dominant, high-constraint states are prone to policies of sporadic ethnic intervention; (3) ethnically diverse, low-constraints states are likely to pursue pacific intervention through diplomacy; and (4) ethnically diverse, high-constraint states are likely to engage in realpolitik policies toward interventionism. By studying five cases—the Indo-Sri Lankan crisis and conflicts in Somalia, Malaysia, Yugoslavia, and Cyprus—Carment, James, and Taydas (2006) suggest that “when combined, internal ethnic diversity and institutional constraints are associated with lower levels of interstate ethnic conflict” (212). Woodwell (2004) studies the effect of the presence of trans-border ethnic groups on intergovernmental relations between pairs of states. In his study, trans-border groups are divided into three analytical categories: groups that exist as a majority of the population of two contiguous states; groups that are shared as a minority in both states; and groups that form a minority of the population of one state but are a majority in another. He expects that each demographic categorization of trans-border ethnic groups will have a different effect on international relations. The central assumption in his study is that “ethnic majorities, in particular, are likely the dominant influence on state foreign policy in those states in which they exist” (199). He mainly focuses on two types of dyads. The first dyad of interest to him is “majority-majority.” Majoritymajority dyads are referred to as two states with similar ethnic majorities comprising the populations of both states. The second type of dyad is “minority-majority.” This can be viewed as the classic diaspora situation, where an ethnic majority comprises one state, while ethnic co-nationals comprise a minority of the population in another state. He hypothesizes

44

Chapter Three

that both “majority-majority” and “minority-majority” dyads tend to have higher levels of dyadic hostility than other dyads. His findings generally confirm his hypotheses. Now, I will turn to the multivariate empirical studies on external intervention in civil wars that stimulate this book. Regan’s (1998 and 2000) is the first study to employ a multivariate econometric model for external intervention in civil wars. He focuses on three generic forms of intervention (military, economic, and mixed) and two targets of the interventions (government and opposition). Thus, the definition and the operationalizations of intervention in his study are different from those of this book. Military intervention includes the supply or transfer of troops, hardware, intelligence, or air or naval support, as well as logistical support to the parties in conflict or the cutoff of any such aid previously in place. Economic intervention includes various types of economic aid, as well as economic sanctions or embargoes. Mixed strategies are those that incorporate some combination of military and economic instruments. The critical point in each instance is that the military or economic behavior conforms to “the convention-breaking and authority-targeted criteria” (25). Using the threshold of 200 casualties to define a civil conflict, 140 intrastate conflicts were identified, of which 90 included at least one thirdparty intervention. The unit of analysis in his study is the conflict. As he admits, this poses the most serious problem in his empirical analysis. He admits that “there are conceptual problems in using the conflict as the unit of analysis because the emphasis of the empirical model shifts from the perspective of the individual decision makers to the aggregate case, asking in essence whether certain structural and contextual conditions increase the probability of an intervention” (52). As the conflict is used as a unit of analysis, it is almost impossible to identify a population of potential interveners. Some of the potential interveners may eventually intervene, but some of them may not. Regan (1998, 2000) cannot capture the decision calculus of these potential interveners. His specification of the logit model that will predict the likelihood that a third party will intervene in an ongoing civil conflict is as follows: Intervention = ߚଵ ൈ casualties + ߚଶ ൈ intensity + ߚଷ ൈ humanitarian issues + ߚସ ൈ cold war + ߚହ ൈ number of borders + ߝ. Even though Regan’s work (1998, 2000) should be praised as the first attempt to use multivariate econometric methods to uncover more systematically the factors that lead to third-party intervention, his model is

Literature Review

45

too simple to provide any meaningful explanations about the complex nature of the decision calculus of third-party intervention. Thus, even though his study is the seminal work on the quantitative empirical analysis of civil war intervention, Regan (1998, 2000) should be read with caution. His results show that an increase in the intensity of the conflict and the number of shared borders decreases the likelihood of intervention, while the Cold War and humanitarian crises increase the propensity toward intervention. He argues that the Cold War is a major factor in the decision to intervene. In his next study with Lemke (Lemke and Regan 2004), the most flawed element of his previous studies, civil conflict as a unit of analysis, is corrected. Unlike Regan (1998, 2000), Lemke and Regan (2004) use the potential intervener as the unit of analysis. For that purpose, they treat each civil war as an intervention opportunity for every member of the international system in existence while the civil war occurred. As a result, the 138 civil wars expand to 19,533 cases of intervention opportunity, in which only 197 actual interventions occurred. In Regan (1998, 2000), the number of observations used in the analysis is only 130: just the number of civil conflicts in his data set. They tested the impact of the potential intervener state and civil war state’s being neighbors, allies, or former colonial powers on the likelihood of intervention. Consistent with their expectations about the impact of those variables, they find that the coefficients for neighbors, allies, and colonial history are all statistically significant in the expected direction. Also, as in Regan (1998, 2000), civil wars during the Cold War were more likely to lead to external intervention. Another contribution they made in their study is that they distinguish between the different motives behind backing the government of the civil war state or backing the opposition groups. By relying on Singer’s (1963) inter-nation influence model, they explain the motives and consequences of choosing sides as follows. When foreign intervention supports the government, it is plausible to assume that the intervening state prefers the government’s current policies. Thus, by intervening on behalf of the government, the intervener seeks to reinforce the government already in power and to reinforce the existing policies of that incumbent government. On the contrary, a foreign state wants to intervene on behalf of the opposition and probably prefers the opposition’s alternate policies and/or mode of governing. Such intervention is thus geared toward modifying the policies and/or mode of governance that exists in the civil war state. According to Singer (1963), altering or restoring the pre-conflict status quo requires the modification or reinforcement of state behavior;

46

Chapter Three

furthermore, to fulfill these preferences, rewarding or punishing strategies should be employed. Since punishment is an influence strategy that can be employed in situations in which the influencer wants to modify (rather than reinforce) the policies in question, intervention as punishment is hypothesized to occur when the intervention is made on behalf of the opposition. Thus, they hypothesize that “interventions on behalf of the opposition will coincide with more violent civil wars than will be the case when interventions occur on behalf of the government” (Lemke and Regan 2004, 152). By calculating the frequency of intervention and the number of casualties, they conclude that their hypothesis is supported. The problem is that they just count the frequency of intervention on each side and the number of casualties without any control variables. However advanced in comparison with Regan (1998, 2000), Lemke and Regan (2004) still present problems in terms of their empirical analysis. They still do not model any sequences of multiple interveners. As Dunér (1985, 7) points out, “It is all the more important not to neglect inter-intervener relations when a wide range of interventionist behavior is being focused on.” Balch-Lindsay and Enterline (2000, 617) also point out that “the impact of third-party interventions on the evolution of intrastate conflicts is often a function of the involvement of other third parties.” As reviewed above, while there have been various attempts to analyze the factors that lead to external intervention in civil wars empirically, those studies all have limitations. Reflecting on those limitations, I will develop a conceptual model of intervention in civil wars in the next chapter.

CHAPTER FOUR THEORY

4.1 Introduction To understand the phenomenon of military interventions by third-party states in civil wars, it is important to understand the objectives of such intervention. As Morgenthau (1967) contends, “Intervention is as ancient and well-established an instrument of foreign policy as are diplomatic pressure, negotiations and war” (425). Although third-party intervention in civil wars has been very important to these interveners’ decision making as well as to the evolution of those civil wars, we suffer from a dearth of literature on the subject of what factors lead to third-party intervention. As Carment and James (2004) point out, whereas the instances of ethnic intervention accumulate, “Unfortunately, students of conflict management are only beginning to construct theories about third-party interventionism with a general range of application, most notably with respect to ethnic conflict” (11). Moreover, most current studies have focused on intervention in identity civil wars, and mostly on ethnic civil wars. Thus, studies such as Carment and James (2004) only focus on the ethnic aspects of civil wars. Carment and James (2004) and Carment, James, and Taydas (2006) argue that a decision to intervene is primarily determined by “a combination of opportunities represented by ethnic divisions in neighboring states and ethnic alliances and constraints that correspond to a state’s institutional configuration and ethnic composition” (3). There is no explicit discussion of the purpose of intervention in these studies. They argue that it is a domestic political situation that is considered a policy choice environment constrained by ethnic composition and institutional configuration. No other variables are involved in the decision making process according to them. It would not be an exaggeration to say that certain types of ethnic composition and institutional configuration will always lead to third-party intervention according to these studies. Their studies have not explicitly discussed the purpose of intervention. As Carment, James, and Taydas (2006) argue, “Decisions to intervene are rational calculations even when

48

Chapter Four

based on ethnic attachment” (20). Rational calculations can be made only if one’s action springs from a goal. Without a clear goal in mind, one cannot calculate anything rationally. According to these researchers, an ethnic composition and institutional configuration that are favorable for decision makers can provide opportunities for those decision makers. These studies argue that there exists a certain combination of ethnic composition and institutional configuration that leads third-party states to intervene. The existence of the opportunity does not mean that there will automatically be the willingness on the part of the potential intervener. The existence of transnational ethnic ties may increase the cost of nonintervention. However, as a rational decision-maker, one still needs to compare the cost component with other components in one’s decision making function. That means that an opportunity itself does not automatically lead to intervention. Willingness to be engaged in a war is also required. Whether to take advantage of that opportunity or not depends on the potential intervener’s calculation. Previous studies fail to include one important aspect in determining what factors lead to external intervention. As discussed above, the probability of success and the expected costs associated with the war should be considered. Even if some opportunities were presented, a state would not go to war if its officials believed that the probability of success in that war was relatively low. As a rational decision-maker, one would be constrained not only by domestic constraints such as ethnic composition and institutional configuration but also by the more important power relations between one’s state and the potential enemy in that war. Thus, the potential intervener has many other elements to consider besides ethnic composition and domestic political configuration. We need to consider the factors that will affect the balance of power between the government and the opposition. Moreover, transnational ethnic ties are not the only opportunity. Even in identity wars, we can find opportunities for intervention. If there is a successful revolution in a country, it will lead to a change in that country’s foreign policy orientation. It will, in turn, affect the interests of other states that have previously had a relationship with those countries that have experienced a successful revolution. Then, the countries that can expect benefits or losses from the revolution will feel the temptation to intervene on whatever side advocates for their interests, either the government or the rebels. As Morgenthau (1967) notes, this can particularly be the case when the revolution is oriented toward a communist shift or anti-communist shift.

Theory

49

Another important quantitative one on external intervention was done by Regan (1996, 1998, and 2000). He argues that “states intervene to stop the fighting between groups in conflict” (Regan 2000, 10). However, narrowing down the purpose of intervention to a desire to stop the fighting will lead us to miss other aspects of intervention. The case of Afghanistan can illustrate this. When there was unrest in Afghanistan caused by several Islamic resistance groups, the Soviets intervened with a heavy use of military force to uphold the government of Babrak Karmal (Wafadar 1981). Even though the Soviets’ move instigated more resistance from the Afghanis, the Soviets’ decision to intervene can somehow be interpreted as a desire to stop the fighting and stabilize Afghanistan. However, at the same time, the Soviets’ decision can also be interpreted as fueled by a desire to favor and save the PDP regime “as a prelude to protecting the Soviet Union’s long-held interests and investment in Afghanistan” (Wafadar 1981, 175). As Regan (2000) notes, the Soviets did not want Islamic fundamentalism close to their border. Then, it is not clear whether we can interpret the intention of intervention by the Soviets as representing a desire to stop the fighting. It is even difficult to consider the purpose of intervention as representing a desire to stop the fighting when we see the perspectives of the United States on Afghanistan after the intervention of the Soviets. In reaction to the Soviets’ intervention in Afghanistan, on December 26, 1979, U.S. National Security Adviser Zbigniew Brzezinski sent a memo to President Jimmy Carter regarding the Soviet invasion. In that memo, Brzezinski argues that “the Soviet intervention in Afghanistan poses for us an extremely grave challenge, both internationally and domestically. While it could become a Soviet Vietnam, the initial effects of the intervention are likely to be adverse for us” (Brzezinski 1979). Then, he analyzes the situation and proposes the action plan (Brzezinski 1979): WHAT IS TO BE DONE? What follows are some preliminary thoughts, which need to be discussed more fully: A. It is essential that Afghanistani resistance continues. This means more money as well as arms shipments to the rebels, and some technical advice; B. To make the above possible we must both reassure Pakistan and encourage it to help the rebels. This will require a review of our policy toward Pakistan, more guarantees to it, more arms aid. C. We should encourage the Chinese to help the rebels also. D. We should concert with Islamic countries both a propaganda campaign and in a covert action campaign to help the rebels;

50

Chapter Four E. We should inform the Soviets that their actions are placing SALT in jeopardy and that will also influence the substance of the Brown visit to China, since the Chinese are doubtless going to be most concerned about implications for themselves of such Soviet assertiveness so close to their border. Unless we tell the Soviets will not take our "expressions of concern" very seriously, with the effect that our relations will suffer, without the Soviets ever having been confronted with the need to ask the question whether such local adventurism is worth the long-term damage to the U.S.-Soviet relationship; F. Finally, we should consider taking Soviet actions in Afghanistan to the UN as a threat to peace.

Based on this documental evidence, I would argue that it would be more accurate to conclude that the purpose of intervention was not to stop the fighting. Nowhere in Brzezinski’s memo to President Carter did he mention that the United States should intervene for this purpose. The main purpose of intervention was to make sure that the Soviets did not occupy Afghanistan. For that purpose, Brzezinski proposed to help the Afghani resistance. The United States wanted to intervene in Afghanistan not because they wanted to help Afghanistan and end the fighting but because they wanted to counter the Soviets’ move.

4.2 Model of Intervention Based on the lessons from the previous studies, I propose an answer to the question of why third-party states consider intervening and want to intervene in foreign civil wars. The reason for the intervention is that they want to change the balance of power between the government and the opposition. In other words, the primary reason for intervention into foreign civil wars is to increase the probability of a win on the side that shares interests with the intervener. In supporting the side that is in accordance with it, the intervening state hopes that ultimately it will be able to benefit from the outcome of the conflict. Clarifying the purpose of intervention gives us the sense that potential interveners will make decisions based on their calculations regarding costs and benefits when they actually intervene. A potential intervener state wants to intervene in a civil war because it wants to attempt to ensure a certain outcome by siding with a friend (government or rebel) that stands in accordance with its purposes. In other words, a state will intervene for its own benefit, not others’ benefits. Thus, in their calculation of whether or not to intervene, potential intervener states will compare the expected utility of intervention with that of non-intervention. If the expected utility

Theory

51

of intervention is higher for them than that of non-intervention, potential interveners will decide to intervene. However, if the expected utility of intervention is lower than that of non-intervention, they will step aside and will not intervene. The basic logic of intervention can be expressed as below. First, the expected utility of intervening in a foreign civil war can be expressed as: EUi = pf|i(Uf) + (1 - pf|i)(U~f) - ȈCi where EUi is the expected utility of intervention in the civil war. The subjective probability of a win by the friend when the third-party state intervenes to support that friend is pf|i. Uf is the utility that can be acquired from the outcome of the civil war when the friend wins over the opponent. Similarly, (1 - pf|i) is the subjective probability of a win by the opponent even after the intervention supporting the friend. The utility of the outcome of the civil war when the opponent wins over the friend is represented as U~f. In other words, the utility of the opponent’s winning, U~f, can mean a loss. ȈCi is the sum of the costs associated with intervention. These include military, economic, and audience costs. Using logic similar to that of the above expression, the expected utility of not intervening in a foreign civil war can be expressed as: EU~i = pf|~i(Uf) + (1 - pf|~i)(U~f) - ȈC~i where EU~i is the expected utility of non-intervention in the civil war. pf|~i is the subjective probability of a win by the friendly side without intervention. Uf is again the utility of a win by the friendly side. (1 - pf|~i) is, in turn, the subjective probability of a win by the opposing side without intervention. U~f is also the utility of a win by the opponent in the civil war. ȈC~i is the sum of all the costs that will be incurred in the case of nonintervention. Regarding those costs, it should be noted there will probably be no military or economic costs associated with non-intervention. In most cases, the costs here will primarily be audience costs (Fearon 1994). The model above is modified from Regan (2000). This model is also represented in a very simple game tree in Figure 4-1. Two things should be noted. First, as in Regan (2000), the decision by the third-party state is assumed to be made after the conflict occurs, with no strategic consideration made. Thus, to incorporate the strategic aspects, we need to modify the model yet again. This consideration will be discussed later. Second, the model in this book is different from Regan’s (2000). Regan (2000) distinguishes the utility of a win by the friend when intervention

Chapter Four

52

occurs from that of a win when intervention does not occur. Thus, his logic can be expressed as Uf|i and Uf|~i, respectively. Accordingly, he also distinguishes the utility of a win by the opponent in the case of intervention from that of a win in the case of non-intervention. These are also represented as U~f|i and U~f|~i, respectively. However, I assume that this utility is based on the outcome of the civil war. That means that the utility of the outcome of the civil war when the friend wins over the opponent is assumed to be the same regardless of whether intervention occurs. The subjective probability and the cost element kick in to differentiate the expected utility of intervention from that of nonintervention. Figure 4-1. Model of Third-Party Intervention

Potential Intervener

Intervention EUi = pf|i(Uf) + (1 - pf|i)(U~f) - ȈCi

Non-Intervention EU~i = pf|~i(Uf) + (1 - pf|~i)(U~f) - ȈCi

From the expressions above, the conditions under which a third-party state will choose to intervene in a foreign civil war can be implied. By

Theory

53

combining the expected utility of intervention and non-intervention, we can determine the conditions of third-party intervention. If we are comparing the expected utility of intervention and non-intervention, it is clear that intervention can be expected to occur when the expected utility of intervention is greater than that of non-intervention, which can be expressed as EUi > EU~i. To put it differently, when EUi - EU~i > 0, thirdparty intervention is expected. We can clearly see the conditions that will increase the likelihood of intervention by substituting for EUi and EU~i. Substituting for EUi and EU~i, we have: [pf|i(Uf) + (1 - pf|i)(U~f) - ȈCi] - [pf|~i(Uf) + (1 - pf|~i)(U~f) - ȈC~i] > 0. We can rearrange and simplify the above inequality. This will produce the following inequality: (pf|i - pf|~i)(Uf - U~f) - ȈCi + ȈC~i > 0. From this rearranged form, we can clearly see the effect of each model component on the expected utility outcomes. Any changes in the values of the terms on the left side of the inequality that produce a positive effect on the aggregate calculation will lead to an increase in the likelihood of intervention. In other words, as (pf|i - pf|~i), (Uf - U~f ), and ȈC~i increase, and as ȈCi decreases, the likelihood of intervention will increase. This implies three things. First, with respect to Uf - U~f, it means that there should be a certain outcome that potential interveners want to obtain. The utility of the outcome of the civil war should be greater than that of the outcome where the friend who has the same interests as the potential intervener loses. Second, the cost of the intervention should be minimal. However, the higher the cost of non-intervention is, the higher the likelihood of intervention will be. Third, as a rational decision-maker, the potential intervener will choose to intervene when such an intervention will make a difference. That means, with respect to pf|i - pf|~i, that the probability of a win by the friend with support should be greater than without support. If intervention does not make any difference in terms of the probability of the friend’s winning, there is no need for the outside party to intervene. However, as I mentioned above, this basic model does not incorporate any strategic aspects of intervention. The world would be too simple if no one cared about others’ moves. The calculations regarding intervention will be different if there is a high probability that the intervener’s intended opponent will receive significant support from other states.

54

Chapter Four

4.3 Extended Model of Intervention Like all foreign policy decision making, decision making regarding intervention should also be considered as strategic. In other words, we should incorporate into our study the fact that the decision to intervene is made in anticipation of the actions of other actors. Krasner (1999) argues that inter-power strategic interaction is an important factor in the intervention decisions of states. He contends that powers will hesitate to or will not pursue military intervention if they know that there will be a challenge by another major power. It is also easily anticipated that a leader of a group that is in a disadvantageous position because of third-party intervention would seek for help from an external ally, so that the power asymmetries would disappear. As observed in history, if one external party decides to intervene, that will trigger the intervention of the other external party. In the Armenia-Azerbaijan conflict, Russia and Turkey, which are considered traditional rivals in the Caucasus area, supported opposing sides in the conflict. Croissant (1998) depicts the situation that developed both in Turkey and in Russia with regard to the Armenia-Azerbaijan conflict in 1992 (81–82): On 18 May Demirel pledged unspecified aid to Nakhichevan, and the Turkish Foreign Ministry issues a statement warning Armenia that “faits accomplish created through the use of force cannot be accepted.” Although there was little indication that Turkey was preparing for military intervention in Nakhichevan, Turkish sabre-rattling prompted a provocative response from Russia. In a clear reference to Turkey, Marshall Shaposhinikov of the CIS Joint Armed Forces warned on 20 May that “third party intervention in the dispute [between Armenia and Azerbaijan] could trigger a Third World War.” The presumable basis for the comment was the Treaty on Collective Security concluded in Tashkent on 15 May among Russia, Armenia, and four of the former Soviet Central Asian Republics. According to the Article four of the treaty: If one of the participating states is subjected to aggression by any state or group of states, this will be perceived as aggression against all participating states to this treaty. In the event of an act of aggression being committed against any of the participating states, all the other participating states will give it the necessary assistance, including military assistance, and will also give support with the means at their disposal by way of exercising the right to collective defense in accordance with Article 51 of the UN Charter.

Theory

55

As a consequence of the above provisions, a Turkish attack on Armenia would have been treated as an attack on Russia, and since Turkey is bound to the United States and Western Europe by similar terms under the North Atlantic Treaty, the state would have been set theoretically for a conflict between NATO and Russia.

Thus, it is imperative to incorporate strategic considerations made by the potential intervening states. Here, I will discuss the strategic consideration of third parties with regard to intervention. Let us suppose that there are two potential intervening states in an ongoing civil war. To simplify the discussion, we will stipulate that there are only two potential intervening states, State A and State B, and that the civil war has already broken out between two groups, Group C and Group D, as in Figure 4-2. In modeling two warring internal groups and two potential intervening states, one makes implicit assumptions. This simplification might be slightly unrealistic because in a civil war, various military organizations or factions may be pitted against each other. However, I do not think it is theoretically necessary to incorporate all of them as separate entities. Practically, it is hard to make a meaningful model reflect all warring factions. For analytical purposes, those complex patterns are reduced to a bipolar format. Therefore, civil war in this analysis is an internal struggle between two internal parties, which are usually understood to be the major contending groups. Either side may be backed militarily from abroad by third-party interveners. Of course, there can be a case when the interveners are neutral, as when they participate in international peacekeeping operations. However, for analytical purposes, only interventions supportive of one or the other warring side will be considered. State A is now considering an intervention into a civil war between Group C and Group D. If Group C is a government, then Group D is a rebel group, and vice versa. An imaginary civil war has broken out between Group C and Group D. At this stage, I assume that Group D intends to attack Group C and Group C intends to resist. It is assumed at this stage that both State A and State B perceive that the probability of winning favors Group D, the aggressor. It is also assumed that before the civil war broke out, State A preferred that Group C win and State B preferred that Group D win. There can be multiple reasons for these relationships. They may have shared interests, or State A may just have interests that are opposed to those of State B. Thus, State A will perceive any benefits to State B as representing a loss to them.

56

Chapter Four

Figure 4-2. Expansion of Civil War

State B

Group D Attack Group C

State A

One example is the strong rivalry between the United States and the Soviet Union during the Cold War period. According to Payne (1970), as long as the Cold War status quo was maintained, no intervention would occur. Here, violations of the status quo can be defined as an attack on a communist (or free-world) country by a free-world (or communist) country. If either side employs force to effect such changes, the status quo is being violated. Consequently, the superpowers on each side are entitled to prevent the others from effecting such changes. In examining superpower intervention in the internal political struggle of other countries, this

Theory

57

competitive relationship seeking international alignments must be considered. If the civil war is perceived to lead to a change in the Cold War status quo, then superpower involvement must follow if the problem cannot be handled locally. Another example can be drawn from Davis and Moore (1997) and Moore and Davis (1998). They contend that different members of the same minority group will assume that their kin share similar policy preferences. This case can be directly used here. Let us suppose that State A and Group C have shared interests. That means that State A does not want Group C to lose the civil war. It is in State A’s interest to defend Group C. According to the simple model of calculation, the calculation of whether State A should intervene or not depends on the probability of a win by Group C without intervention by State A versus the probability of a win by a coalition of Group C and State A if State A intervenes—and the costs in each case. However, now the calculation should be changed if the intervention by State B is anticipated once State A intervenes to help Group C. State A wants to intervene because it is expected that Group C will lose the civil war without external help. The probability of a win is believed to be favorable to Group D without any external intervention. However, as State A intervenes to make a coalition with Group C against Group D, the balance of power is changed and the probability of winning now favors Group C (with help from State A). This may not hold if Group D is so strong that the formation of the coalition does not affect the probability of winning. Thus, it is implicitly assumed here that the coalition of Group C and State A is stronger than is Group D alone. As State A decides to intervene in the civil war between Group C and Group D, State B also needs to consider intervention. Before the intervention by State A, Group D is more likely to win in the civil war. That probable outcome is in accordance with State B’s interests. State B does not have to intervene in this case because the probability of winning is favorable to Group D. However, the situation changes when State A intervenes. The probability of winning is now unfavorable to Group D if Group D is going to fight against the coalition of Group C and State A. There can be various reasons why State B would want to help Group D. State B might want to protect its interests that are with Group D, or State B might not want State A to benefit from the outcome of the civil war. In either case, State B prefers that Group D win. Thus, State B will seriously consider an intervention to help Group D when Group D finds itself in a disadvantageous position.

58

Chapter Four

If State B’s intervention is anticipated once State A intervenes, State A has to reevaluate its situation. Before State B’s intervention is incorporated into its calculation of costs and benefits, State A’s calculation is rather simple and its cost of intervention is not that high. However, if State A believes that intervention on the part of State B is very likely, it has to reconsider its decision. Basically, what State A has to do is consider whether Group C is worth defending at the expense of all the costs that it may encounter if State B actually intervenes. Thus, it is in State A’s interest to intervene in the civil war between Group C and Group D when State B is unlikely to intervene. Consequently, before State A intervenes to help Group C, several points should be considered. First, if State B’s intervention is anticipated, State A must now consider the increased cost of intervention if it still wants to intervene to help Group C. Even without intervention by State B, State A already has to pay some cost for the intervention to improve the probability of a win by Group C. However, the cost of intervention is now multiplied. To make sure that Group C is going to win in the civil war given the external interventions, State A must invest more resources in fighting the coalition of Group D and State B. Second, as related to the first point, State A has to consider the worth of Group C. The reason why State A initially considers the intervention is that Group C is worth protecting at the acceptable cost. As in the simple model of intervention calculation, State A will intervene if the expected utility of the intervention exceeds 0. In calculating the expected utility, the most important pieces of information required are the probability of winning, the utility of a win by Group C, and the cost of intervention. This simple calculation should be adjusted now that State B’s intervention is very likely. Intervention by State B will change the probability of a win by a coalition of Group C and State A and the cost of intervention by State A. For my discussion of this strategic consideration that must be undertaken by the potential intervening states, I use the conceptual framework developed from extended deterrence theory (Huth 1988; Fearon 1994). More precisely, what is pertinent here is the concept of extended-immediate deterrence. Huth (1988, 16) defines extendedimmediate deterrence as “a policy in which (a) a potential attacker is actively considering the use of military force against a protégé of the defender; (b) policymakers in the defender state are aware of this threat; and (c) recognizing that an attack is possible, policymakers of the defender state, either explicitly or by the movement of military forces, threaten the use of retaliatory force in an attempt to prevent the use of military force by the potential attacker.” This conceptualization of extended-immediate

Theory

59

deterrence comes from Morgan (1977), who distinguishes immediate deterrence, which concerns “the relationship between opposing states where at least one side is seriously considering an attack while the other is mounting a threat of retaliation in order to prevent it” (28), from general deterrence, which relates to “opponents who maintain armed forces to regulate their relationship even though neither is anywhere near mounting an attack” (28). The logic here is straightforward and is applicable to the discussion of strategic consideration of third parties with regard to intervention in foreign civil wars. The logic of extended deterrence refers to three actors: a potential attacker, a potential target, and a potential third-party defender. In extended-immediate deterrence, the attacker is about to decide whether to attack the target. The target state will have to decide whether it is going to resist. It is the third-party’s role to consider whether or not to intervene to help the target once the attacker’s decision to attack the target is made. If this is the case, the attacker has to consider the retaliatory capacity and credibility of the defender. If the potential defender does not have the capacity to defend its protégé when it actually intervenes, the attacker does not have to think much about the defender’s decision because the defender’s involvement will not change the power balance enough to change the probability of a win. However, if the potential defender does have the capacity to cause a change in the intervention calculation, then the problem becomes the credibility of the defender. Thus, in extended deterrence theory, the attacker’s decision will be dependent upon the credibility of a defender’s threat to intervene. If the defender’s threat to intervene is credible enough, then the attacker will be deterred. Thus, the attacker will not attack the target. On the contrary, if the defender is not that credible, the attacker will attack the target because the attacker now believes that the defender is not going to intervene for the sake of its protégé. As I am discussing the applicability of extended immediate deterrence theory to third-party intervention into civil wars, one more study is also truly applicable to this study. Gartner and Siverson (1996) try to find out why the scope of most wars does not expand beyond the involvement of the initial two participants. Even though their study is designed to find out the empirical regularity of international wars, their conceptual framework can be applied to this study. They contend that potential attackers, as rational decision makers, will initiate a conflict where they see a high probability of success. It is obvious that the probability of success will be lower if a third-party will come to help a target state. Calculating this, potential attackers seek to avoid a target that is backed by a credible

60

Chapter Four

defender, according to Gartner and Siverson (1996). Their conclusion implies that if the initiators correctly estimate the probability of success, they should generally win the war. However, if the initiators are overly confident in their calculation of the probability of success because they underestimate the credibility of third-party intervention, they will lose. Accordingly, Gartner and Siverson (1996) expect that initiators are likely to lose in multilateral wars. Their empirical investigation also confirms that initiators are likely to win if the war remains bilateral. However, initiators are generally going to lose if the war becomes multilateral with a third-party intervening to help the target state. As Huth (1988) notes, it is very difficult “to estimate confidently the intentions and capabilities of opponents” (2). In many situations, uncertainty is a result of scant information about the attitudes and beliefs of key decision makers within an opposing state and the nature of policy debates at the highest levels of that government. Uncertainty about the intentions and capabilities of adversaries can have an important impact on policy choices and the effectiveness of foreign policy. In the context of extended deterrence, uncertainty is likely to undermine the credibility of a defender’s threat. The credibility of a deterrent threat depends upon the defender’s being perceived as possessing (1) military capabilities sufficient to inflict substantial costs on a potential attacker and (2) the will or intention to use those capabilities if necessary (4). This extended model of intervention can be more clearly understood if one uses a game theoretic model, as in Figure 4-3. To construct the basic structure of this model, I am going to use the model developed from the extended-deterrence theory. However, the model I use here is a modification of extended-deterrence theory in many aspects. As in the above scenario, this model has four players: State A, State B, Group C, and Group D. Group C and Group D represent the warring factions within a state. They are fighting a civil war. Since this book focuses on intervention by third-party states in ongoing civil wars, Group C and Group D are assumed to have decided to fight against each other. Here in this model, Group D has decided to attack Group C and Group C has decided to retaliate. Their decisions are not contingent upon the decisions made by the third parties. Since the decisions by Group C and Group D will not change, they are not an integral part of the strategic calculations to be made by State A and State B. There is also another modification of the simple model of intervention present here. In the discussion of the simple model of intervention, the expected utility functions of intervention and non-intervention are EUi = pf|i(Uf) + (1 - pf|i)(U~f) - ȈCi and EU~i = pf|~i(Uf) + (1 - pf|~i)(U~f) - ȈC~i,

Theory

61

respectively. These utility functions will be simplified in the extended model of intervention. Figure 4-3. Extended Model of Third-Party Intervention

State A A

Intervention

Non-Intervention

State B Non-Intervention

PC·X (PC·X; PC·(-X))

PC·(-X)

PA·X - CA PA·(-X)

Intervention

PA|B·X - CA|B PA|B·(-X) - CB|A

As seen in Figure 4-3, there are three probable outcomes of the model. The first probable outcome is that both State A and State B do not intervene. If State A decides not to intervene, State B will make the same choice. This outcome indicates that the civil war will not be internationalized.

62

Chapter Four

The war will instead remain within the borders of the state where the civil war has broken out. Even though this outcome looks simple, there can be many different interpretations of this first probable outcome. First, it can be inferred that State A initially does not have any interest in defending Group C. However, I discard this interpretation because I already assume that Group C and State A have shared interests. The second possible interpretation is that State A intends to intervene but that the cost of intervention is too much for that state to bear. Third, State A might be deterred by State B. With regard to this third probable outcome, previous studies on deterrence (Fearon 1994) can be applicable here. Fearon (1994) argues that instances of apparent peace do not only mean that there exists no intention of aggression. Instances of apparent peace also include the cases where the potential aggressor is deterred by the other state. Related to this, a discussion by Smith (1996) is also worth noting. Discussing the reliability of alliances, Smith (1996) concludes that “the prospect of intervention affects the behavior of nations prior to war” (33). The second probable outcome is that State A will choose to intervene, while State B will decide not to intervene. In this case, State B will simply sit on the sidelines and watch what happens. Here, State A might believe that Group C is worth protecting from the attack by Group D. Or, on the contrary, State B might believe that Group D is not worth all of the costs that fighting against the coalition of Group C and State A would entail. State A might correctly see that State B does not care much about Group D. Thus, State A could conclude that State B will not intervene on Group D’s behalf once State A intervenes in the civil war. The final probable outcome is that State B will intervene to counter the intervention by State A. This final probable outcome can be interpreted in two ways. First, State A might be confident that its coalition with Group C will win the civil war. In this case, State A can correctly estimate that State B will intervene once State A intervenes. It might be State B’s miscalculation in its decision making regarding intervention. Second, State A may be overconfident that State B is not credible. This time, the miscalculation would be State A’s. With these probable outcomes in mind, I discuss the payoff schedule for State A and State B. First, the status quo payoffs for both State A and State B are: (PC·X; PC·(-X)). The first component represents State A’s payoff, and the second represents State B’s payoff. PC denotes the probability of a win by Group

Theory

63

C without any external intervention. X denotes the magnitude of change after the civil war. X plays a very important role in calculating the costbenefit analysis for both sides. X basically reflects the stakes of the war. Thus, it is important to note that State A’s utility increases as X increases, whereas State B’s utility decreases as X increases. These are simplified versions of the basic model of intervention calculation. Here, the costs of non-intervention, such as an audience cost, are not considered. The utility of the opponent’s winning in the case of non-intervention, (1 - pf|~i)(U~f), is assumed to be zero and is therefore not included in this payoff schedule. Second, the payoffs for State A and State B in the case of intervention by State A and non-intervention by State B are: (PA·X - CA; PA·(-X)), where PA denotes the probability of a win by the coalition of State A and Group C when State B does not intervene, X again denotes the stakes of the war, and CA denotes State A’s cost of intervention in fighting against Group D without its being backed by State B. Third, if State B decides to intervene, the payoffs for State A and State B are: (PA|B·X - CA|B; PA|B·(-X) - CB|A), where PA|B denotes the probability of a win by the coalition of State A and Group C in the presence of State B’s counter-intervention, X again denotes the stakes of the war, CA|B denotes State A’s cost of intervention in fighting against the coalition of State B and Group D, and CB|A denotes State B’s cost of counter-intervention. Here, several restrictions are imposed in order to motivate the game. First, there should exist an upper bound for the possible stakes over which the civil war participants and potential interveners are going to fight. The stakes of the war are limited. Even though this depends on what kinds of the states we are talking about, the stakes of the war should not exceed the total resources available within the target country. Even if the stakes of the war is the strategic importance of the target government or the survival of the ethnic brethren, there should be an imaginary upper bound for this. Without this upper bound for the possible stakes of the war, infinite utility could be derived from it. That means that the pursuit of this infinite utility could make the participants go and fight regardless of the cost. Second, with respect to the probability of winning, I assume that the probability of a win by the coalition of State A and Group C against Group

64

Chapter Four

D alone is greater than the probability of a win by the coalition of State A and Group C against the coalition of State B and Group D—in other words, PA > PA|B. This assumption is intuitive rather than theoretical. Also, the probability of a win by the coalition of State A and Group C against Group D alone is assumed to be greater than the probability of a win when Group C alone fights against Group D. Thus, it is represented as PA > PC. Third, it is also assumed that the cost of intervention by State A increases as State B counter-intervenes. That means that State A’s cost of intervention in fighting against Group D alone is smaller than its cost of intervention with counter-intervention by State B. Thus, CA|B > CA. At the same time, State B also has to pay the cost of the intervention, CB|A > 0. In sum, the counter-intervention by State B will reduce the probability of a win by the coalition of State A and Group C and will also increase the cost of intervention by State A. Since PA > PA|B and CA|B > CA, State A is going to prefer intervention without counter-intervention by State B. Without State B’s counterintervention, State A can increase the probability of its winning if it moves from non-intervention to intervention. State A also prefers the policy of intervention without counter-intervention to that of intervention with counter-intervention. With the counter-intervention of State B, State A has to endure more costs. However, the degree to which intervention without counter-intervention by State B is preferable to non-intervention is indeterminate. The stakes of the war, X, are the same for each case. The probability of a in by the coalition of State A and Group C against Group D alone is assumed to be greater than the probability of a win when Group C alone fights against Group D: PA > PC. This depends on the cost of intervention. That is, the decision to intervene is dependent upon how much State A is willing to pay. It is assumed that there is virtually no cost entailed by nonintervention. If the cost of intervention is great enough, then State A’s expected utility will be reduced enough for it to try to avoid that cost of intervention. With respect to State B, the order of preferences is a bit different. Since no cost is involved for State B, there is no cost for State B associated with the first two probable outcomes. However, as discussed, considering the probability of a win by the coalition of State A and Group C and that of a win by Group C alone, State B will prefer that there be no external intervention: PC·(-X) > PA·(-X)—since PA > PC, PA·(-X) is clearly less desirable than PC·(-X) for State B. Between the policy of counter-intervention and that of non-intervention, that is, between PA·(-X) and PA|B·(-X) - CB|A, State B’s order of preference

Theory

65

is indeterminate. On the one hand, through counter-intervention, State B can decrease the probability of its obtaining a less favorable outcome as a result of State A’s successful intervention. As (-X) denotes the magnitude of an undesirable civil war outcome, it would be better for State B if PA were greater than PA|B. On the other hand, in shifting from nonintervention to intervention, State B must pay the cost of intervention, CB|A. CB|A is already assumed to be greater than zero: CB|A > 0. Thus, State B’s decision to intervene is critically dependent upon how much State B can contribute to the probability of a win through intervention—that is, the difference between PA|B and PA, the cost of intervention, CB|A, and the magnitude of the stakes of the civil war, X. By shifting its policy from non-intervention to counter-intervention, State B can reduce the likelihood of State A’s success through intervention. Thus, it is very important to see how much State B can increase the probability of a win. As State B’s involvement in a civil war increases the probability of a win through its coalition with Group D, the value of its intervention will also increase. Another important part of the calculation is the value of the stakes of the war, X. Even in a case where State B’s involvement will increase the probability of success marginally, if the stakes of the war, X, are great enough, State B can endure the cost of intervention. If the stakes of the war, X, are insignificant, then even if State B can contribute significantly to the increase of the probability of a win, this may not be enough to be able to compensate for the cost of intervention. From the discussion above, it is notable that the state of the civil war will critically affect the potential interveners’ decisions. In the next section, I will discuss how these elements affect the process of finding an equilibrium.

4.4 Equilibrium with Complete Information In the simplest world, State A and State B know each other’s preferences. Accordingly, finding an equilibrium in this game based on such complete information is easier. Sadly, our real world is not as simple as this. However, the solution to this game based on access to complete information will provide us with a direction to look in order to see the complications involved. I am going to begin with State B. Observing the intervention by State A in the civil war between Group C and Group D, State B will consider and compare the elements of the payoff schedule. After this comparison has been made, State B will intervene if the expected utility of the

66

Chapter Four

intervention outcome, PA|B·(-X) - CB|A, exceeds the expected utility of the outcome associated with non-intervention, PA·(-X). This condition for intervention is represented by PA|B·(-X) - CB|A > PA·(-X). This inequality can be rearranged. Then we can obtain a condition that will determine State B’s intervention. State B will intervene if X > CB|A/(PA - PA|B). Then we can obtain a critical value of X that will make State B indifferent to the distinction between intervention and non-intervention. The critical value of X is represented as Xcrit, where: Xcrit = CB|A/(PA - PA|B). Now, we can compare the stakes of the war, X and Xcrit. If the stakes of the war are high enough to make its importance greater than the critical value of X, such that X > Xcrit, then State B will be willing to intervene in order to lower the probability of State A’s success. On the contrary, if the stakes of the war are pretty low, such that X < Xcrit, then it is not worthwhile for State B to intervene given the cost of intervention. In this game based on complete information, State A can anticipate the above calculation by State B. As assumed, State A’s utility increases with X. That means that State A wants to intervene in a civil war where the higher value of X is at stake. However, it is not desirable for the state to have to fight against State B because the probability of a win when there is counter-intervention by State B, PA|B, is smaller than the probability of a win when State A is the sole intervener, PA. According to the order of preferences of State A, State A is going to prefer intervention without counter-intervention by State B, since PA > PA|B and CA|B > CA. In other words, since the same amount of rewards is involved, State A will be better off when it has to fight against Group D alone than when it has to fight against the coalition of State B and Group D. One important point to note here is that State A and State B are assumed not to be able to manipulate the size of X. Potential interveners just have to decide whether to intervene or not by observing the stakes of the war and by considering all of the associated costs and the probability of success. Thus, to avoid counter-intervention on the part of State B, State A will have to choose intervention where the value of the stakes, X*, is the same or less than the critical value of X for State B. There is another restriction placed on this calculation. If X* ” Xcrit, State A knows that State B will not intervene on Group D’s behalf. In this case, State A has to compare the payoff of intervention without counter-intervention with that of non-intervention. State A will intervene if PA· X*-CA > PC· X*. This condition can be rearranged as X* > CA/(PA - PC). If X is smaller than

Theory

67

CA/(PA - PC), such that X* < CA/(PA - PC), State A will not intervene. Thus, State A will intervene if X* falls between CA/(PA - PC) and Xcrit, such that CA/(PA - PC) < X* ” Xcrit = CB|A/(PA -PA|B). If the stakes of the war are greater than the critical value of the stakes for State B, such that X* > Xcrit, then State A can anticipate that State B will always counter-intervene once State A intervenes in the civil war. Thus, State A has to choose between intervention and non-intervention by comparing the payoffs. State A will intervene when the payoff from intervention given the counter-intervention of State B is greater than that of non-intervention, such that PA|B·X* - CA|B > PC·X*. This condition can be rearranged as X* > CA|B/(PA|B - PC). The solution of this game can be stated in terms of State A’s decision regarding the stakes of the civil war and State B’s decision whether or not to intervene. State A: Intervention if CA/(PA - PC) < X* ” Xcrit = CB|A/(PA - PA|B) when X* ” Xcrit or if X* > CA|B/(PA|B - PC) as long as X* > 0: Nonintervention otherwise. State B: Intervention if X > CB|A/(PA - PA|B); Non-intervention otherwise. The propositions based on this equilibrium will be discussed in section 4.6.

4.5 Equilibrium with Incomplete Information In the previous game with complete information, State A could maximize its utility by carefully choosing the target state, which itself could be achieved by considering the stakes of the war, the probability of winning, and the cost of intervention for both State A and State B. Given the premises of the complete information game, it would be possible for State A to select the target because State A would know the cost schedule for State B. However, in the game with incomplete information, State A will no longer know the cost that intervening would present to State B. Only State B will know the information about the costs of its intervention, CB|A. In the previous game with complete information, the information regarding CB|A is critical for State A to use in calculating the range of X so that it can avoid counter-intervention by State B. As similarly described in Powell (1999), even though State A is uncertain about the exact value of the cost, CB|A, it may hold beliefs about State B’s costs. Following Powell (1999), State A holds the belief that State B’s cost cannot be below a certain level, CB|AL, and cannot be above an upper limit of CB|AU. In other words, State A

68

Chapter Four

believes that State B’s cost of intervention will fall between CB|AL and CB|AU. Furthermore, State B’s cost is assumed to follow a uniform probability distribution over the interval [CB|AL, CB|AU], where CB|AL > 0. It should be immediately understood that the cost of intervention cannot be less than zero. This cost of intervention does not only represent the monetary or physical costs of intervention. We know that if the cost of intervention is low, then the payoff from the intervention will be higher. Thus, the cost of fighting can be interpreted “more broadly as a measure of …state’s willingness to use force or of its resolve” (Powell 1999, 98). Borrowing Powell’s argument, we can understand that State A’s uncertainty about State B’s cost of intervention, CB|A, also can be used to indicate that State A is unsure about State B’s willingness to intervene in the civil war at hand. Thus, this game becomes a game between two states that are not sure of each other’s willingness to intervene. For State B, there exists a critical value that makes State B feel indifferent to the difference between intervention and non-intervention. That critical value makes the expected value of intervention and the expected value of non-intervention the same. That is, State B, in facing State A’s intervention into the civil conflict between Group A and Group B, will compare the expected value of non-intervention, PA·(-X), with its expected value from intervention, PA|B·(-X) - CB|A. We can get to the critical point for State B by making these two payoffs equal to each other. Since State A cannot know the exact value of this critical value, it can only have expectations about the cost schedule. I will define C(X) as C(X) = X·(PA - PA|B), so that State B with cost C(X) is indifferent about intervention when facing State A’s intervention in a civil conflict of which the value is X. Thus, if State B’s actual cost of intervention is less than C(X), then the state will intervene. On the other hand, if State B’s actual cost of intervention is bigger than C(X), then it will not intervene. It is important to see where C(X) is located in order to know whether State B is going to intervene or not. That critical value of C(X) suggests that it is a function of (1) what the stakes of the war are, X, and (2) how much State B can affect State A’s probability of winning, (PA - PA|B). State A wants to intervene in a civil war if the stakes of the war, X, are significant. State A might have to intervene if the stakes of the civil war are too big for them to avoid. However, if X becomes greater, it will lead to counter-intervention by State B, which will not be welcomed by State A. As discussed above, State A cannot just choose a target conflict based on the magnitude of the stakes of the war, X. State A must incorporate the

Theory

69

willingness of intervention by State B—in other words, State B’s cost of intervention. The problem is that State A does not know State B’s cost of intervention. It can only assume that State B’s willingness to intervene is a function of the cost. Thus, if the stakes of the war are defined as X, the probability that State B will intervene is F[C(X)]. By the same token, the probability that State B will not intervene is (1 - F[C(X)]). Turning to State A’s perspective, the expected utility of intervention for State A can be represented as UA(X) = F[C(X)]·(PA|B·(X) - CA|B) + (1 - F[C(X)])·(PA·(X) - CA). This equation expresses the problem that State A faces in its consideration of intervention. As seen in the game with complete information, State A faces the same kind of problem. As the magnitude of the stakes of the civil war, X, increases, State A’s utility from intervention will increase if its intervention ends in success. At the same time, the greater value of X means that it will be very difficult for State A to avoid intervening to help Group C. However, the problem posed with reference to State A’s decision is that it is more likely that State B will counterintervene as X increases. To find a solution, the equation above is differentiated with respect to X and is then rearranged. Then, Uiƍ(X) is represented as Uiƍ(X) = {F[C(X)]·PA|B + (1 - F[C(X)]·PA)} - {F[C(X)]·Cƍ(X)·((PA - PA|B)·X + (CA|B - CA))}. This equation is more intuitively understandable. The first part of the right-hand side of the equation represents the expected gain from intervention for State A given a marginal increase in X. This signifies how much benefit gain State A can achieve via intervention as X increases. The second part of the right-hand side of the equation represents the expected costs of the intervention for State A given a marginal increase in X. Thus, to find a solution where the marginal benefits from the intervention in a civil war with a varying value of X equal the marginal costs, we can set {F[C(X)]·PA|B + (1 - F[C(X)]·PA)} = {F[C(X)]·Cƍ(X)·((PA - PA|B)·X + (CA|B - CA))}. From the equation above, we can find Ui(Xopt), where

70

Chapter Four

{F[C(Xopt)]·PA|B + (1 - F[C(Xopt)]·PA)} = {F[C(Xopt)]·Cƍ(Xopt)·((PA PA|B)·Xopt + (CA|B - CA))}. Thus, Xopt is the optimal magnitude of the stakes of a civil war where State A can maximize the benefit of intervention. The final decision by State A with regard to intervention can be determined by comparing the expected value of intervention with the expected value of non-intervention. If the expected value of intervention exceeds the expected value of nonintervention, then State A will intervene. On the other h and, if the expected utility of intervention falls short of the expected value of nonintervention, then State A will maintain the status quo. In the next section, I will discuss the propositions suggested by both games.

4.6 Propositions From the extended model of intervention, I developed two games: one including complete information and the other including incomplete information. From the game with complete information, the following equilibrium is suggested. State A: Intervention if CA/(PA - PC) < X* ” Xcrit = CB|A/(PA - PA|B) when X* ” Xcrit or if X* > CA|B/(PA|B - PC) as long as X* > 0: Nonintervention otherwise. State B: Intervention if X > CB|A/(PA - PA|B); Non-intervention otherwise. From the equilibrium above, we can imagine three possible scenarios. First, there will be no intervention in a civil war between Group C and Group D. State A decides not to intervene after the cost-benefit intervention calculation. One probable reason for non-intervention can be found in State A’s cost of intervention. If it is too costly for the state to intervene, it will not intervene. Another reason can be found in the interaction between State A and State B. This is the situation where the potential counter-intervention by State B imposes additional costs in State A. Therefore, the expected spoils from intervention will not exceed the expected costs. Second, only State A intervenes. The spoils from the intervention in the civil war are large enough for State A to intervene, but they are not large enough for State B to intervene. This situation can also be explained by the idea that the cost of intervention is too high for State B to justify its intervention. This is the situation where CA/(PA - PC) < X* ” Xcrit = CB|A/(PA - PA|B).

Theory

71

Third, both State A and State B intervene. This is where the stakes of the war are greater than the critical value of the stakes for State B, such that X* > Xcrit. If this is the case, then State A can anticipate that State B will always counter-intervene once State A intervenes in the civil war. Thus, State A has to choose between intervention and non-intervention by comparing the payoffs. State A will intervene when the payoff from intervention given counter-intervention by State B is greater than that of non-intervention, such that PA|B·X*- CA|B > PC·X*. This condition can be rearranged as X* > CA|B/(PA|B - PC). It is important to note that what kind of outcome we are going to see is determined by the possible range of X*. A broader range for X* means that State A can intervene in a civil war without fearing counter-intervention by State B. On the contrary, a narrower range means either that the civil war will remain an internal conflicts or that it will escalate to the level of multilateral international warfare. Four propositions are suggested with regard to intervention in civil wars. All four propositions are related to the costs and probabilities of winning based on intervention by State A and State B.

X*.

Proposition 1 An increase in State B’s costs of intervention will widen the range of

According to the solution above, State B’s cost of intervention, CB|A, is one of the elements that determine the upper end of the range of X*. The higher cost of intervention for State B means that the range of X* will become broader. This proposition means that counter-intervention by State B becomes less likely, as the counter-intervention is expected to be costly for State B. From the perspectives of State B, it would be very difficult to justify its counter-intervention when the cost of counter-intervention is this high. Due in part to this difficulty experienced by State B, State A is given more freedom in choosing whether or not to intervene. Proposition 2 A decrease in State A’s costs of intervention will widen the range of X*. State A’s cost schedule is also related to the range of X*. The lower end of the range is determined in part by the magnitude of State A’s cost of intervention without State B’s counter-intervention, CA. The lower State A’s costs of intervention, the wider the range of X*. In other words, State A faces a lower barrier to deal with when it considers intervention in a civil war.

72

Chapter Four

Proposition 3 An increase in the probability of winning in the case of State A’s intervention, as compared with the status quo, (PA - PC), will widen the range of X*. It is one of the assumptions of the game with complete information here that the probability of a win by the coalition of State A and Group C against Group D alone is greater than the probability of a win when Group C alone fights against Group D. Thus, it is always true that PA > PC in this study. Since this (PA - PC) is the denominator of the lower end, the lower end will become smaller as the denominator, (PA - PC) increases. This proposition can be interpreted as meaning that State A is more likely to intervene when its contribution to a potential victory is more apparent. That means that when its contribution to the win is deemed to be trivial, State A can hardly justify intervention. Proposition 4 An increase in the marginal impact of State B on the probability of a win by the coalition of B and D, which makes the (PA - PA|B) bigger, will decrease the range of X*. It is also assumed that the probability of a win by the coalition of State A and Group C against Group D alone is greater than the probability of a win by the coalition of State A and Group C against the coalition of State B and Group D—in other words, PA > PA|B. This assumption is intuitive rather than theoretical.

4.7 Summary By incorporating the potential counter-interveners on the opposite side, I showed that the potential intervener’s decision calculus is completely different from that indicated by the simple model of intervention. Previous studies, including those of Regan (1998, 2000) and Lemke and Regan (2004), have not incorporated this element into their conceptual modeling. To make our model of third-party intervention more realistic, it is necessary to include various factors. According to the discussion above, to understand the mechanism of third-party intervention, we have to know at least the factors listed below. (1) X: The magnitude of change after the civil war. (2) PC: The probability of a win by Group C without any external intervention. (3) PA: The probability of a win by the coalition of State A and Group C when State B does not intervene.

Theory

73

(4) PA|B: The probability of a win by the coalition of State A and Group C in the presence of State B’s counter-intervention. (5) CA: State A’s cost of intervention in fighting against Group D without its being backed by State B. (6) CA|B: State A’s cost of intervention in a fight against the coalition of State B and Group D. (7) CB|A: State B’s cost of counter-intervention.

In sum, to understand how and why first interventions and second or counter-interventions are made, it is important to note these three factors: X, the magnitude of change after the civil war; P, the probability of winning for potential participants; and C, the cost of intervention for the participants. Since these three factors are derived from an abstract formal model, it is necessary to consider what factors in the real world affect these three factors. In the next chapter, operational terms for these factors will be discussed along with the framework of analysis. Another important concept to incorporate into the model is the sequence of the intervention by third-party states. As Balch-Lindsay and Enterline (2000) argue, civil war is a multi-stage process that evolves over time. Thus, the decision to intervene in a civil war should be treated as such. However, the model by Lemke and Regan (2004) is built on a crosssectional set up. Accordingly, their data are collected as if they were a snapshot. Usually, civil wars last more than a year in most cases, even though there are a number of civil wars that have ended within a year. Lemke and Regan (2004) treat all civil wars as if they started and ended within a year. They model both a civil war and the decision to intervene as instantaneous events. Consequently, their model does not distinguish the first intervention from counter-intervention. If a state that does not have any intention of intervening in the first place intervenes in a civil war with the explicit purpose of counterintervention, then their model completely misses the meaning of that counter-intervention and cannot explain the mechanism that leads to the counter-intervention. According to the model and empirical testing by Regan (2000) and Lemke and Regan (2004), the rationale behind the intervention by the Soviet Union and Cuba in the Angolan civil war in 1975 would be the same. Their model also cannot distinguish the decision calculus of the United States from that of the Republic of Korea when the two states decided to intervene in Vietnam. With my model, as presented here, the mechanism behind the decision to intervene can be more clearly explained.

CHAPTER FIVE FRAMEWORKS OF ANALYSIS

In this book, I propose that four dimensions should be investigated if we want to fully understand the conditions under which foreign countries will intervene in civil wars. Four dimensions have been depicted in Figure 5-1. Combined with the propositions put forth in Chapter 4, this framework will help us to develop a multivariate econometric model to test my hypotheses. As discussed in Chapter 4, I consider the factors that can cause variations in those three factors: X, the magnitude of change after the civil war; P, the probability of winning for potential participants; and C, the cost of intervention for the participants. These variations can be explained along with the four dimensions in this section.

5.1 First Dimension: Characteristics of Civil War and Civil War State The first dimension is the civil war itself. In other words, the characteristics of civil war and the country where the civil war occur are to be considered via this first dimension. There are many factors related to this dimension that can lead to variations in X, the magnitude of change after the civil war, P, the probability of winning for potential participants, and C, the cost of intervention for the participants. With regard to this first dimension, several factors can influence the propensity toward foreign intervention. First, as Sambanis (2001) notes, “Not all civil wars are the same” (259). He explores the differences that may exist between the causes of identity (ethnic/religious) and nonidentity civil wars. With the analysis of 161 countries over a 40-year period, he argues that there exists a significant difference between the determinants of identity civil wars and those of nonidentity civil wars.

Chapter Five

76

Figure 5-1. Four Dimensions of Intervention in Civil War

State A Dimension 2

Dimension 4

State B Dimension 2

Dimension 3

Group D

Group C

Dimension 1

Frameworks of Analysis

77

Since the dataset used in this book is heavily borrowed from Sambanis, it is worth mentioning how he coded different war types. He himself coded war types relying on the state failure projects by Esty et al. (1997), which define ethnic wars as: Episodes of violent conflict between governments and national, ethnic, religious, or other communal minorities (ethnic challengers) in which the challengers seek major changes in their status…Rioting and warfare between rival communal groups is not coded as ethnic warfare unless it involves conflict over political power or government policy.

There have been many theories with regard to the causes of ethnic violence. However, in this book, those theories about the causes of ethnic violence are not that relevant. Thus, I will focus more on the effect of different war types on external interventions. With regard to the war type, there can be two opposite arguments. The first one is related to X, the magnitude of change after the civil war, whereas the second one is related to P, the probability of winning for potential participants, and C, the cost of intervention for the participants. Regarding X, Payne (1970) and Feste (1992) provide a theory of external intervention. Payne (1970) defines threat as “the creation of expectations about one’s likely punitive action in response to the future action (or inaction) of another” (4). Responding to threats, superpowers have tried to block or inhibit the aggressive tendencies of their opponent. According to him, both the U.S. bloc and the Soviet bloc have an implicit agreement about what belongs to who based on the application of the same measures of jurisdiction. This is the “status quo,” as refers to the recognition of possession and ownership. Thus, the concept of changes in the status quo is very important. Following Payne’s (1970) logic, violations of the status quo can be understood as an attack on a communist (or free-world) country by a freeworld (or communist) country. If either side tries to make changes, it means that the status quo is violated. If a country is currently perceived as a U.S. bloc country (or a Soviet bloc country), a challenge to that allegiance would be perceived as a serious infringement to the status quo between the two superpowers. Consequently, the superpowers on each side are entitled to prevent the other from effecting such changes. Payne (1970) also argues that intervention in civil wars by superpowers can be understood when we incorporate the relationship of this competition for international alignments. According to him, when expected changes in governments are not believed to affect the Cold War status quo, then intervention by superpowers will not occur. On the

78

Chapter Five

contrary, if the outcome of the civil war appeared to lead to a change in the Cold War status quo, then superpower involvement would follow if the problem could not be handled locally. Feste (1992) also argues that this ideological struggle between the U.S. bloc and the Soviet bloc is related to their foreign policy instrument, which is intervention in other countries. Feste (1992) defines spheres of influence as “respective locales of interest asymmetry in which one superpower’s interests are clearly and substantially more important than those of other” (6). She categorizes and ranks spheres of competition according to level of interests in the following way (6): 1. High-interest symmetry: locales in which both the United States and the Soviet Union have strong, if not vital, interests. 2. Disputed-interest symmetry: locales in which the United States and the Soviet Union do not agree on the relative balance of their interests. 3. Low-interest symmetry: locales in which both the United States and the Soviet Union have modest interests. 4. Uncertain-interest symmetry: locales of an ambiguous or fluid nature in which one or both superpowers are not certain of their own or the other’s interests and find it difficult to assess how and to what extent their interests will become engaged in a developing, unstable situation.

She also argues for the importance of other factors such as capabilities and viable strategies open to each side in asserting their influence in different domains, as well as various domestic and international constraints on their definition of their interests in operational terms. However, she argues that the most important element that defines the superpowers’ behavior is the concept of national interests. Thus, losing a country in a high-interest geopolitical area cannot be accepted by either side. According to Payne (1970) and Feste (1992), we can expect that superpowers have strong motives for intervening in civil wars whose outcome is expected to transfer the control of the government from one side to the other. More specifically, during the Cold War period, it is assumed that each superpower cannot accept the changes of the government. On the contrary, if a civil war is not aimed at changing the government from one side to the other ideologically, it does not concern either superpower. This assumption is totally in accordance with Payne (190). If ethnic warfare were not related to changes in governments as will affect the Cold War status quo, superpowers would have a very minor interest in the outcome of that civil war. Thus, superpowers are less likely to intervene in those ethnic civil wars. On the contrary, if the civil war

Frameworks of Analysis

79

occurs based on ideological differences, the outcome of the civil war will lead to a change in the Cold War status quo. Then, it provides a motive for superpowers to involve themselves in those civil wars. Kaufmann (1996) gives us an opposite argument regarding intervention in ethnic and non-ethnic civil wars. He explains the differences that the potential interveners can face in ethnic civil wars and non-ethnic civil wars in terms of the cost and the probability of success. Thus, if his ideas are applied to my model as presented in Chapter 4, we can see that Kaufmann (1996) focuses more on P, the probability of winning for potential participants, and C, the cost of intervention for the participants. Kaufmann (1996) argues that for any state considering intervention in a foreign civil war, the most important question is whether the intervention can be a determinant of the outcome. According to him, the outcomes of ideological civil wars are determined “primarily by the relative political competence of the local factions, which outsiders can rarely do much to change” (Kaufmann 1996, 81). On the contrary, the outcomes of ethnic civil wars are mainly determined by the balance of force, which outside forces can change dramatically. Thus, there is a good deal of room for foreign powers to contribute to ethnic civil wars. Another important aspect of ethnic civil wars is that there is no loyalty competition. It is agreed that while not everyone in the ethnic group will participate in the fight against the other ethnic group, they certainly will not support the opposing ethnic group. He argues that this is the most important factor that distinguishes ethnic conflict from non-ethnic conflicts. The implications of this lack of loyalty competition in ethnic civil wars can be explained in two ways. First, for outsiders, it is often impossible to gather good intelligence on individual loyalties. This is very critical for foreign interveners. As observed in Vietnam, if the war is based on ideological differences, foreign interveners cannot distinguish one group from the opposing group. This is because in ideological civil wars, the entire population is perceived as “the shared mobilization base of both sides” (Kaufmann 1996, 66). However, in ethnic civil wars, outside forces are able to distinguish one group from the other because each side can mobilize only members of its own group and usually those groups are easily located in friendly-controlled territory. Second, because of the reason above, in ethnic wars, territorial control is much more critical than in the ideological civil wars. In ideological civil wars, the military control of a territory does not secure for foreign interveners the loyalties of the residents. In ideological civil wars, the outcome is determined by the “hearts and minds” battle, not by territorial control by military forces.

80

Chapter Five

Outside forces cannot contribute much to this “hearts and minds” battle in ideological civil wars. One thing should be noted. As Kaufmann (1996) also acknowledges, there have been many ethnic conflicts where one side or the other has taken on “an ideological identification as a tactical maneuver to gain outside support” (69). This was not uncommon during the Cold War, where the U.S. and the Soviets were ready to provide aid in hopes of preserving the balance of power. Pereira (1994) describes the external intervention in the Angolan civil war in this regard. MPLA (Movimento Popular de Libertação de Angola, The Popular Movement for the Liberation of Angola) was mainly composed of the Mbundu tribe and urban mestizos. They aligned with the Soviet bloc, so they received aid from the Soviet Union and Cuba. FNLA (Frente Nacional de Libertação de Angola, The National Front for the Liberation of Angola) was composed of the Bakongo and received aid from the United States and Zaire. UNITA (União Nacional para a Independência Total de Angola, The National Union for the Total Independence of Angola), which mainly opposed the MPLA, was composed of the Ovimbundu, Chokwe, and Ngangela. They received aid first from China and later from the United States and South Africa. This is why the Angolan civil war is considered the most prominent Cold War proxy war (Simpson 2002). In the case of this proxy warfare, it is difficult to gauge the propensity toward external intervention because this type of war includes both aspects. Second, as Fearon (2007) notes, the aim of the civil war can be “to take power at the center or in a region, or to change government policies.” Civil wars are usually fought between the government and the rebel forces. Certainly, rebel forces can have various aims when they launch an attack on the government. They might aim to replace the government with their own. This is usually the case when revolutionary war (identity war) takes place. However, ethnic groups have also fought for power, because it is hard to maintain power-sharing agreements between opposing ethnic groups. Security becomes a problem for all ethnic groups. Thus, they aim either to take over the government or to achieve total autonomy in their region. These different aims of civil wars are directly related to X, the magnitude of change after the civil war. The magnitude of change after the civil war is a very important determinant of external intervention in my model. X, the magnitude of change after the civil war, is much bigger when the aim of the civil war is to take over the government and to replace it. The change in the government as the outcome of the civil war poses a

Frameworks of Analysis

81

much bigger problem for external forces, encouraging interest in intervention. If the issues of the civil war are merely local ones, this means that X is small. Accordingly, foreign forces have less interest in these civil wars. Another point is that local issues will be considered more emphatically to be the domestic affairs of the countries in question. As the United Nations supports the principle of sovereignty, this means that foreign forces will hesitate to intervene in foreign civil wars when the issues at stake are merely local. Third, a country’s reliance on natural or primary commodity exports is also believed to be important. Since Collier and Hoeffler (2000), it has been generally agreed that countries with a higher percentage of national income from primary commodity exports have been more prone to civil war. They argue that rent-seeking behavior can lead to a struggle over the control of these natural primary commodities. Their works on this issue have been highly influential in both academic and policy communities. However, Fearon and Laitin (2003) and Elbadawi and Sambanis (2002) challenge the Collier and Hoeffler findings. They both argue that Collier and Hoeffler’s empirical test is very fragile and that “there is no significant relationship between civil war and natural resource dependence” (Elbadawi and Sambanis 2002, 325). Reynal-Querol (2002) separately studies the impact of these primary commodity exports on ethnic and non-ethnic civil wars. She finds that primary commodity exports have a statistically significant impact only at the onset of non-ethnic civil wars. The debates between the Collier and Hoeffler side and the Fearon side are still going on. Thus, it is too early draw conclusions about whether there exists significant a correlation between primary commodity exports and the onset of civil wars. There are two ways that this heavy dependence on primary commodity exports is related to external intervention. First, when a government faces a strong rebel group and needs external assistance, that government can promise a foreign government that it will share the profits from the primary commodity exports or give them rights to them, since revenues from these primary commodity exports will be impossible unless the government receives the help from outside. This was observed during recent history. It is believed that during the civil war in the Democratic Republic of Congo, the Kabila government offered timber concessions to Zimbabwe in exchange for military assistance (Global Witness 2002).

82

Chapter Five

Second, a government or rebel group can use these primary commodity exports as their funding sources or as incentives for foreign intervention. Ross (2005) calls this type of behavior “booty futures.” It means that rebels sell future resource exploitation rights to foreign companies or states. He argues that the sale of booty futures was a key source of start-up funds for rebel groups in at least five of Africa’s resource-rich states between 1989 and 2004, and that it was almost impossible for the rebels in Sierra Leone, Democratic Republic of Congo, Congo-Brazzaville, and Equatorial Guinea to begin their insurgencies. Thus, it is logically acceptable to extend this “booty futures” argument in this context. A third-party state can provide military aid to rebels in exchange for future resource exploitation rights. Finally, the ethno-political configuration of the target state is an important element to consider in this dimension. There have been debates regarding whether civil war is more likely in more ethnically diverse countries (Fearon and Laitin 2003) and whether civil war is also more likely when the state is controlled by an ethnic minority (Cederman and Girardin 2007; Fearon et al. 2007). With regard to the latter argument, it has been believed by many scholars of nationalism that this might be the case (Fearon et al. 2007). As Gellner (1983) puts it, if plurality groups are excluded from power, it triggers the nationalist sentiment so strong and it “constitutes a quite outstandingly intolerable breach of political propriety” (1). However, for both debates on civil war onset, Fearon et al. (2007) finds that there is no evidence that more ethnically diverse countries are at a greater risk of civil wars, and also that a government’s being controlled by minority ethnic groups is not related to the onset of civil war. However, the statistical significance of those studies is not that important here because in this study I assume that civil war has already begun, so the exact subjects of those studies are irrelevant. More relevant is a situation in which the civil war has already begun. If a civil war occurs in a state where a government is controlled by an ethnic minority, it will be easier for external forces to intervene in supporting the rebels. As discussed above, in ethnic civil wars, the military power balance is the most important determinant of civil war outcomes. Moreover, those ethnic civil wars usually take place in economically lessdeveloped countries. That means that the warring factions in ethnic civil wars cannot be equipped with modern high-tech weapons. Thus, military power typically means the quantity of armed forces fighting. In ethnic civil wars, because there is no loyalty competition, one can only recruit from one’s ethnic group. One cannot recruit armed forces from the opposing

Frameworks of Analysis

83

ethnic groups. If this is the case, then the probability of a win for a rebel group is higher and, accordingly, the cost of intervention on behalf of the rebel group is lower. This situation will make it easier for third-party states to intervene.

5.2 Second Dimension: Characteristics of the Potential Intervener The second dimension is represented by the potential interveners. Almost every potential intervener has different political and power resources. These different resource bases will give potential interveners different opportunities and differing degrees of willingness to intervene. First, the domestic political configuration of the potential intervener should be considered. As Carment and James (2004) and Carment, James, and Taydas (2006) argue, the leader of the potential intervener state will take into account the various constraints imposed on his or her choices. They argue that among many constraints, ethnic composition and institutional constraints are the two most important dimensions. In both studies, those two important determinants of ethnic interventionism are the extent to which the institutions of the state are in the hands of a single ethnic group (ethnic domination versus diversity) and the extent to which leaders enjoy discretionary power over state policies (high versus low institutional constraints). Based on this model, they predict different types of intervention policies: (1) ethnically dominant, low-constraint states are predisposed to bellicose policies of ethnic intervention; (2) ethnically dominant, high-constraint states are prone to policies of sporadic ethnic intervention; (3) ethnically diverse, lowconstraint states are likely to pursue pacific intervention through diplomacy; and (4) ethnically diverse, high-constraint states are likely to engage in realpolitik policies toward interventionism. Another important factor that is believed to affect the propensity of third-party intervention is the capability of those potential interveners. First, if the potential intervener is too weak to contribute to the outcome of the civil war, there is no reason for that state to even consider intervention as its foreign policy instrument. This is directly related to Proposition 3 in Chapter 4. It is one of the assumptions of the game with complete information here that the probability of a win by the coalition of State A and Group C against Group D alone is assumed to be greater than that of a win when Group C alone fights against Group D. Thus, it is always true that PA > PC in this study. Since this (PA - PC) is the

84

Chapter Five

denominator of the lower end, the lower end will become smaller as the denominator, (PA - PC), increases. This proposition can be interpreted to mean that State A is more likely to intervene when its contribution to the victory is more apparent. That means that when its contribution to the win is deemed to be trivial, State A can hardly justify its intervention. Second, even when third-party states can contribute to the outcome of the civil war, they might not be able to intervene because their resources to protect themselves are limited. This means that unless third-party states are very strong, it is difficult for them to launch an intervention. Thus, it seems to make sense when Bull (1984) argues that “a basic condition of any policy that can be called interventionary … is that the intervener should be superior in power to the object of intervention” (1).

5.3 Third Dimension: Relationship between Interveners and the Host Country The third dimension is the relationship between interveners and the host country. With regard to this relationship between interveners and the host country, there are several elements to be considered: geographical proximity, formal alliances, ethnic ties, and the power capability ratio. First, agreement has been established in the field of international relations that geographical proximity matters. Regan (2000) and Lemke and Regan (2004) both find that sharing a border is very strongly and significantly related to the probability of external intervention. This importance of geographical proximity in international relations has been discussed in various ways (Boulding 1962; Clark and Regan 2003; Most and Starr 1989; Pearson 1974; Richardson 1960; Russett 1975; Siverson and Starr 1990; Small and Singer 1976, 1982; Starr and Most 1976; Vasquez 1995). Conceptually, Most and Starr (1989), Starr and Most (1976), and Siverson and Starr (1990) argue that borders and proximity are the principal indicators of interaction opportunity. Agreeing in part with Richardson (1960), Starr and Most (1978) note that “if borders do not cause wars, they at least create structure of risks and opportunities in which conflictual behavior is apparently more likely to occur” (444). This proximity-as-opportunity explanation, thus, argues that proximity does not cause war, but that proximity does provide the opportunity for war. The reason behind this is that “only a few states have the capability or the opportunity to fight wars with non-neighbors” (Vasquez 1995, 278). This reasoning has already been discussed by Boulding (1962). By using

Frameworks of Analysis

85

the term “loss-of-strength gradient,” Boulding (1962) argues that the strength of a state decreases when it is further away from its homeland. It follows that a state is more able to use force if the civil war is closer to home. Pearson (1974a) argues that it is more difficult for third-party states to undertake hostile intervention than it is for them to pursue friendly intervention far from home, “since there is a high probability of resistance to hostile intervention, which may increase intervener’s costs” (435). Thus, according to this proximity-as-opportunity explanation, we can see that the geographical proximity of the potential interveners to the civil war state, or lack thereof, is very closely related to C, the cost of intervention for the participants. Second, another important element to consider as part of this third dimension is the existence of any alliances between the potential intervener and the host government. If one’s country has an alliance with the government experiencing the internal conflicts, it is very unlikely that one is going to support the opposition. Two reasons can be provided for this argument. First, the allied partners are believed to have an interest in keeping the status quo. It is logically natural that countries form alliances with other countries for military, diplomatic, or economic benefits. It would be better for neighboring states to have a friendly government across the border rather than a government that is hostile to them. Thus, it is also beneficial for the potential interveners if the current government in the host country remains in power as long as their alliance lasts. In sum, if the outcome of the civil war is expected to be a change of the government, and if it is believed that this change in the government will affect the interests of the potential interveners, third-party states will intervene to protect those interests. Second, it is difficult for third-party states to ignore their alliances. There will be a reputation cost for the third-party state that does not respect an alliance. However, one thing to note in this regard is that a current alliance may have been created by a previous government with different diplomatic or political purposes. If the current government does not have any intention of honoring the formed by the former government, the alliance only exists on paper. Thus, at the aggregate data level, it is impossible for us to distinguish whether current alliances are real or nominal. In relation to the issue above, it is also very important to note that the fact that a third party does not support the opposition does not necessarily indicate that it supports the allied government. Studies (Siverson and King 1980; Kim 1991) show that alliances are not reliable. According to Kim (1991), if a nation is at war, only one out of four times will the allies of that nation give their assistance. Considering this, we can hardly expect

86

Chapter Five

that nations will become involved every time their allies are experiencing domestic conflicts. Theoretically, we can assume that being in an alliance does not guarantee external intervention on the side of the government in a civil war. Therefore, we need to consider why those allies do not fully honor their commitment to their alliances. The reason is very simple in this case: the cost of honoring the commitment to an alliance exceeds the benefit of it. Given this, the balance of power between the government and the opposition should be considered. The relative power balance between the government and the opposition is assumed to be important because it will affect the decision of the third-party states. When the expected probability of winning is low because of the relative power of the opposition, the potential intervener has to hesitate to intervene. The relative strength of the opposition means that more resources from the potential intervener are required if that state wants to change the balance of power between them (the intervener and the government) and the opposition. It would cost more for the potential interveners to ignore their alliances and start to support the opposition. Third, some studies have shown that shared ethnicity between the intervener and the host country is the important determinant of third-party intervention. Davis and Moore (1997) argue that ethnic ties across the border, which they call transnational ethnic alliances, indeed have an impact on foreign policy behavior. In their earlier study, Davis, Jaggers, and Moore (1997), they found support for the argument that there is a connection between the dyadic conflict level between two neighboring states and the presence of a disadvantaged ethnic minority in one state when members of that same minority group are in power in the other. However, as in the discussion on the relationship between alliances and intervention, ethnic ties will not automatically lead to third-party intervention. Even if the majority of third-party states share the ethnicity of the government or the opposition in a host state, it would be reasonable to assume that the third-party state might hesitate to intervene if the side that it was supposed to join were much weaker. I anticipate that the level of intensity of the third-party intervention will be different depending on whether the third-party assists the government or the opposition. Even though Carment and James (2003) and Carment, James, and Taydas (2006) argue that shared ethnicity and ethnic composition constitute the main variable in explaining the decision making regarding intervention, as I already mentioned above, opportunity does not automatically lead to willingness.

Frameworks of Analysis

87

The assumption of Mason, Weingarten, Jr., and Fett (1999) is that the government is stronger than the opposition when a civil war commences. In calculating the probability of success, in case a state is actually going to intervene in the civil war, the consideration of whether the state would support the government in the host state or the opposition can be an important factor. The other thing to consider is that the support from the external forces will change the calculations of the original participants in the war and will, then, affect the decision to intervene. If the third-party states join the government with which they share an ethnicity, the expectation might be that the conflict will end sooner. That means that the cost of intervention will be lower and thus would make third-party intervention easier. However, if the third-party states share the ethnicity of the opposition, which is assumed to be weaker, the expectation is that third-party intervention on the side of the opposition will establish a rough balance of power between the government and the opposition. This means that the duration of the war will be longer due to the intervention and that the cost of war will be higher. Thus, not only domestic political and ethnic constraints but also realistic constraints play an important role in deciding the behavior of potential intervener states. A related point is the power of the target country and of the potential interveners, and especially the power ratio between the two. Bull (1984) argues that a state with a great amount of power is no longer a great power if it becomes the object of foreign intervention. This fact is observed from the dataset. In the data collected by Sambanis (2004), there were 10 civil wars coded that occurred within major power states. There was one in the United Kingdom, there were seven in Russia (including the Soviet Union), and there were three in China that were coded. None of these civil wars included foreign intervention. First of all, no small state will intervene on behalf of the opposition when the civil war occurs within the borders of a major power. Let us imagine a case where there is a rebel insurgency against the government in a major power state. First, the probability of a win for the rebels is very low. More importantly, an increase in the probability of a win in the case of State A’s intervention, as compared with the status quo, (PA - PC), will be extremely minimal because the major power’s government has strong military forces already. Second, as related to the first argument, the intervening state should be almost equal to the major power in terms of military might in order to make any meaningful changes in the balance of power in favor of the rebel forces. However, in the case where a major power state intervenes in a civil war in another

88

Chapter Five

major power state, it is more likely that there will be a major inter-state war between these two major powers. No superpower would assume that risk of escalation.

5.4 Fourth Dimension: Relationship between Interveners Finally, we need to take note of the interaction between the potential interveners. As discussed before in this book, previous studies such as those by Regan (2000) and Lemke and Regan (2004) do not incorporate this strategic aspect of the relationship between the interveners. Actually, Regan (2000) notes that “I will make the assumption that the decision to intervene is not strategic, by which I mean that the decision is not a function of the moves or countermoves of the target country but rather a result of internal process in the intervening country” (42). He argues that his assumption is valid because his work is based on “the premise that a conflict is under way and a third party—for a host of reasons—is deciding whether or not to intervene in this ongoing conflict” (42). His argument holds true only if there is only one intervening state. The situation, and the calculation of the potential intervener, would be different if that state knew that there would be external support available to the enemy. We have observed from history that among the 151 civil wars recorded by Sambanis (2004) from 1944 to 2001, 63 civil wars have included at least one intervention. The important thing to note here is that out of those 63 intervention cases, 32 cases have also included counter-interventions. More than 50 percent of the instances of intervention include both intervention and counter-intervention. Thus, it is hard to justify a theory that ignores the strategic moves of the countries that are already participating or have the potential to become involved. If one state actually intervenes in a civil war, then that state becomes a factor to be considered by other potential interveners. For example, two scenarios are easily imaginable. First, State A considered intervention because of its interest in the outcome of a certain civil war. However, it decided not to intervene either because the cost of intervention is too high or because the probability of winning is too low. As explained in the model introduced in Chapter 4, those two factors keep a potential intervener from actually intervening. While State A is hesitating because of the cost and the probability-ofwinning question, State B intervenes on behalf of the same side on whose behalf State A would intervene if other constraints were removed. Now, the intervention calculation has changed. As State B is already intervening on the same side on which State A would intervene, the cost of

Frameworks of Analysis

89

intervention for State A becomes very low. At the same time, as there exists another external force assisting the same side, the probability of winning for the coalition of State A and State B is increased. These changes in the intervention calculation of State A can now make intervention a more feasible strategy for State A. Second, State A has a vested interest in a civil war. The outcome of the civil war, thus, will affect State A. As long as State A believes that its vested interest in the civil war state is not affected, State A will not intervene. More specifically, if a group in a civil war that State A would support is winning or is believed to be capable of winning without support from State A, State A does not have to intervene. State A’s interest would not be affected in this case, regardless of the intervention by State A. That means that State A would not intervene as long as the outcome of the civil war is expected to be in accordance with its interest. While State A is enjoying the civil war from outside, State B intervenes in support of the opposite side in the civil war. State B’s involvement now changes the balance of power between the participants in the civil war. The probable outcome of the civil war is now not as was expected before the intervention of State B. As State A wants to keep this change in the expected outcome from happening, State A has a very strong motive for intervening. In the same vein, another counter-intervention scenario is possible. In this case, State A does not have a strong interest in the outcome of the civil war. As long as the outcome of the civil war is limited to the civil war state, State A is indifferent to the outcome. That means that State A will not intervene at the current stage. However, for whatever reason, State B intervenes into the civil war. Given the expected outcome, the government-to-be group in the civil war will align with State B. The consequent expected outcome of the civil war bothers State A. Since what State A cares more about is the strategic/power balance between State A and State B, the alignment of the civil war state with State B is not an acceptable outcome for State A. To prevent this alignment from occurring, State A will intervene on behalf of the opposite side, which will fight against State B’s coalition powers. Thus, it is expected that the relationships between the actual and/or potential interveners may result in one of two scenarios: bandwagoning intervention and counter-intervention.

CHAPTER SIX DESCRIPTIVE STATISTICS

Based on the concepts and definitions provided in Chapter 2, 151 civil wars were identified according to Sambanis (2004). Those 151 civil wars were broken down according to their starting dates (see Table 6-1). From the 151 civil wars between 1944 and 2001, ambiguous cases are dropped from this study. These were cases that were not definitely recorded as civil wars and ones that Sambanis also notes are ambiguous. They include Chad 1994–1997, Korea 1948–1949, Namibia 1973–1989, and Syria 1979–1982. Also, all of the civil wars in major power states (China, France, Russia [including the period of the Soviet Union], the United States, and the United Kingdom) were also dropped. The dataset generated by EUGene (Bennett and Stam 2000) codes Germany and Japan as major powers since 1990. However, in this study, major power status is only assigned to those states that have been traditionally considered major powers. As theory dictates and history shows, there cannot be external interventions in civil wars in states that are considered major powers. Of these 151 civil wars, 63 civil wars included external interventions (see Table 6-2). Those 63 civil wars with external interventions were connoted with a total of 153 individual interventions. More than 30 percent of the total individual interventions, 48 individual interventions out of 153 total individual interventions, were carried out by major powers, and the remainder was attributed to minor powers. The United States intervened 22 times, accounting for the most instances of interventions by far. Russia (the Soviet Union) follows with 14 interventions. France intervened 5 times. China and the United Kingdom intervened 3 times and 4 times, respectively.

Chapter Six

92

Table 6-1. Number of Conflicts Initiations per Year .

1944

5

1951

0

1961

1

1971

5

1981

2

1991

11

1945

0

1952

1

1962

2

1972

3

1982

0

1992

8

1946

2

1953

1

1963

4

1973

2

1983

3

1993

1

1947

2

1954

0

1964

0

1974

4

1984

2

1994

7

1948

6

1955

1

1965

3

1975

6

1985

1

1995

1

1949

0

1956

2

1966

3

1976

4

1986

1

1996

4

1950

3

Totals

18

1957

0

1967

3

1977

1

1987

2

1997

2

1958

2

1968

0

1978

7

1988

3

1998

4

1959

1

1969

0

1979

3

1989

4

1999

3

1960

4

1970

2

1980

3

1990

5

2000

1

12

18

38

23

42

Among the minor powers, Uganda intervened 6 times and Cuba intervened 5 times. Angola, Belgium, Iran, Libya, Rwanda, South Africa, Syria, and Turkey all carried out 4 interventions each. These individual interventions can be further broken down. Among those 153 individual interventions, 90 interventions were either bandwagoning or counter-interventions. The motives of those 63 first interventions and 90 second interventions should be different. Of its 22 interventions, 14 of the ones executed by the United States were first interventions. Thus, nearly 64 percent of the interventions by the United States were first intervention. Out of 14 Russian interventions, 6 interventions (about 43 percent) were first interventions. All of the interventions carried out by Rwanda and Syria were first interventions.

Start

1978

1975

1991

1972

1970

1975

Country

Afghanistan

Angola

Azerbaijan

Burundi

Cambodia

Cambodia

1991

1975

1972

1994

1991

1992

End

Gov Opp

Gov Opp

Opp

Gov

Gov Opp

Opp

Gov

Gov Opp

Side

D.R.V. (78.12)

U.S.A. (70.3)

D.R. Congo (72.5) None

Armenia (91.12)

Russia (75.11)

Russia (79.5)

Interveners

Russia (79.1)

R.V.N. (70.3)

Turkey (92.3)

Cuba (75.12)

U.S.A. (80.1)

Table 6-2. Cases of Civil Wars with Intervention by Third-Party States

Descriptive Statistics

China (79.1)

D.R.V. (70.6)

U.S.A. (75.12)

Iran (80.1)

Laos (79.3)

S. Africa (76.1)

Pakistan (80.1)

U.S.A. (79.12)

Nigeria (78.2)

93

1965

1980

1948

1978

1993

1998

1960

Chad

Chad

Colombia

Colombia

Congo

Congo

Zaire

94

1965

1999

1997

Cont.

1966

1994

1979

Gov

Opp

Uganda, Rwanda (98.8)

Rwanda (96.10)

Opp

Gov

None

None

Opp Gov

U.S.A. (89.8)

U.S.A. (62.5) None

Libya (80.11)

France (68.12)

Gov

Gov Opp

Opp

Gov

Gov Opp

U.S.A. (60.7)

Uganda (96.12)

France (83.6)

Libya (70.10)

Chapter Six

Russia (61.12)

Zimbabwe (98.8)

Angola (97.3)

U.S.A. (83.7)

Angola (98.8)

D.R.C. (83.7)

Sudan, Chad (98.9)

1967

1977

1996

1998

1948

Zaire

Zaire

Zaire

Zaire

Costa Rica

1948

2001

1997

1978

1967

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

U.S.A. (67.7)

Guatemala (48.3)

Uganda, Rwanda (98) Nicaragua (48.4)

Uganda, Rwanda, Angola (96.10) None

France, Belgium (78) None

Belgium (67.7)

Belgium (60.7)

Descriptive Statistics

Angola, Zimbabwe (99)

Ethiopia (67.7)

Egypt (64.11)

Algeria, Uganda, Ghana (64.12)

95

1963

1974

1991

1965

1979

1974

Cyprus

Cyprus

Djibouti

Dominican Republic

El Salvador

Ethiopia

96

1991

1992

1965

1994

1974

1967

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Cuba (76.12) None

U.S.A. (81.3) None

U.S.A. (65.4) None

France (92.2) None

Turkey (74.7)

Opp

Gov

None

Turkey (63.12)

Gov

Opp

Gov

Honduras (82.6)

Greece (63.12)

Chapter Six

1976

1991

1992

1944

1966

1978

Ethiopia

Georgia

Georgia

Greece

Guatemala

Guatemala

1994

1972

1949

1994

1992

1988

Gov

Opp

Gov

Opp

U.S.A. (78)

U.S.A. (66.8) None

U.K. (44.12)

Russia (92.10)

Opp

Gov

None

Russia (92.6)

Opp

Gov

None

Somalia (77.6)

Gov

Opp

Gov

U.S.A. (47.3)

Russia, Cuba (77.9)

Descriptive Statistics

Bulgaria, Albania, Yugoslavia (47.9)

Yemen PR (78.3)

97

1998

1978

1961

1974

1985

Guinea-Bissau

Iran

Iraq

Iraq

Iraq

98

1996

1975

1970

1979

1999

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Iran (86.2)

Iran (74)

Syria (63.5) None

Iraq (79.6) None

Guinea, Senegal (98.6) None

None

Turkey (86.8)

Russia (74.6)

Chapter Six

1970

1960

1958

1978

1989

1992

Jordan

Laos

Lebanon

Lebanon

Liberia

Liberia

1997

1990

1991

1958

1973

1971

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Israel (76.12)

U.S.A., U.K. (58.6)

Russia (60.12)

Senegal, Ghana, Nigeria (92) None

Senegal, Ghana, Nigeria (90.8) None

Syria (76.1)

Syria (58.5)

U.S.A. (60.9)

Syria (70.7)

Opp

Gov

None

Gov

Descriptive Statistics

Vietnam (61.1)

China (62.1)

Thailand (69.9)

99

1991

1975

1976

1978

1981

Moldova

Morocco

Mozambique

Nicaragua

Nicaragua

100

1990

1979

1992

1991

1992

Opp

U.S.A. (82)

U.S.A. (78)

Opp

Gov

None

Gov

Opp

S. Africa (80.10)

Algeria (75.12)

Opp

Gov

None

Russia (92.5) None

Gov

Opp

Gov

Cuba (83.7)

Russia (80.11)

Chapter Six

Honduras (85.4)

Tanzania (82.2)

Russia (88.6)

Zimbabwe (84) Malawi (86.8)

1971

1971

1973

1980

1990

1988

Oman

Pakistan

Pakistan

Peru

Rwanda

Somalia

1991

1993

1996

1977

1971

1975

Gov

Opp

Gov

Opp

Yemen PR (71.12)

U.S.A. (89.6)

Belgium, France, D.R. Congo (90.10) None

U.S.A. (85.8) None

Afghanistan (75.7)

Opp

Gov

None

India (71.7)

Opp

Gov

None

U.K. (71.1)

Gov

Opp

Gov

Descriptive Statistics Saudi Arabia, Iran (72.2)

Jordan (72.7)

101

1983

1963

1983

1992

1966

Sri Lanka

Sudan

Sudan

Tajikistan

Thailand

102

1982

1997

2002

1972

2002

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

U.S.A. (66.11)

Afghanistan (93.2)

Libya (84.3)

Russia (71.5) None

India (84.4)

Ethopia (88.4)

Uganda (97.4)

China (72.11)

Uzbekistan, Kazakhstan (93.3)

Malaysia (70.5)

Iran (91.12)

U.K. (85.2)

Chapter Six

Russia (93.3)

1978

1981

1960

1962

1972

1983

Uganda

Uganda

Vietnam

Yemen

Zimbabwe

Zimbabwe

1987

1979

1970

1975

1987

1979

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

Opp

Gov

S. Africa (83.1) None

S. Africa (74.2)

Saudi Arabia (62.10)

U.S.A. (61.5) None

U.S.A. (84.8)

Tanzania (79.1)

Mozambique (76.4)

Jordan (62.10)

S. Korea (65.1)

N. Korea (85.11)

Libya (79.2)

Descriptive Statistics

Zambia (77.6)

Egypt (62.11)

Cuba (78.2)

103

104

Chapter Six

Among those 63 first interventions, as seen in the Table 6-3, 33 interventions were carried out to support the government in the civil war in question, while 30 interventions were executed to help the rebel groups. One notable thing in this table is that there are distinct differences between the second interventions. When the first intervention was carried out to support the government in the civil war, 14 out of 33 cases (about 42 percent) did not include second interventions. However, if the first intervention was made to support the opposition in the civil war, the chance of another intervention following it increased. As seen in Table 6-3, 22 out of 30 cases (more than 73 percent) included second interventions. The types of the second interventions were also very different. In cases where the first intervention was made to support the government in the civil war, 16 of the second interventions were bandwagoning interventions, while 11 of them were counter-interventions. Among those 19 second interventions, both types of second interventions occurred in 8 cases. Table 6-3. Number of Interventions for Government and Oppositions Government

Opposition

No Second Intervention Band-Wagoning Intervention Counter-Intervention

14

8

8

1

3

15

Both

8

6

Total

33

30

In cases where the first interventions were carried out to help the opposition in the civil war, the types of second interventions were very different. Among the 22 of these second interventions, only 7 precipitated bandwagoning second interventions, while 21 counter-interventions were made against the original interventions on behalf of the opposition. It is difficult to draw any meaningful conclusions from these results without controlling for the effects of other variables. However, the numbers show here that more counter-interventions were made when the aim of the first interventions was to help the opposition. When the first interventions were made to help the government in the civil war, 22 out of 33 cases did not include counter-interventions. That means that only 11 cases (about 33 percent) included counter-interventions. On the contrary, when the first interventions were made to support the

Descriptive Statistics

105

opposition in the civil war, 21 out of 30 cases (70 percent) included counter-interventions. We can also see some distinctive patterns in the frequency of intervention between identity (ethnic/religious) civil wars and identity civil wars. Among 151 civil wars recorded by Sambanis (2004), 97 civil wars are identified as identity civil wars, whereas 54 civil wars are coded as identity civil wars. Within those 97 identity civil wars, 39 civil wars (about 40 percent) included external interventions, whereas there were foreign interventions in 24 out of 54 identity civil wars (about 44 percent). There is not much of a difference between identity civil wars and identity civil wars in terms of the ratio of the number of interventions to the number of civil wars. As seen in Table 6-4, in identity civil wars, the number of interventions for the government and for the opposition is roughly equal. On the other hand, in identity civil wars, intervention in support of the government is twice as common as intervention for the opposition. Another notable thing found in Table 6-4 is the pattern of the counterinterventions. When external intervention was aimed at helping the opposition in the identity civil war, 15 out of 22 cases (about 68 percent) included counter-interventions by other external forces. When first interventions were carried out to assist the government in question, only 5 out of 17 cases (about 29 percent) experienced counterinterventions. A similar pattern exists in interventions into identity civil wars. When the first interventions were made to help the government, 6 out of 16 cases (about 38 percent) included counter-interventions. On the contrary, when first interventions were made to support the opposition in identity civil wars, 6 out of 8 cases (75 percent) included counter-interventions.

Chapter Six

106

Table 6-4. Number of Interventions by War Types Identity Civil Wars

Identity Civil Wars

Government

Opposition

Government

Opposition

No Second Intervention

8

7

6

1

BandWagoning Intervention

4

0

4

1

CounterIntervention

2

12

1

2

Both Types of Second Intervention

3

3

5

4

Totals

17

22

16

8

A final comparison can be made between the number of interventions during and after the Cold War. There were 109 civil wars during the Cold War period, whereas 42 civil wars have been fought since the Cold War. As seen in Table 6-5, among those 109 civil wars fought during the Cold War period, there were 48 interventions (44 percent). 15 external interventions (about 36 percent) were observed in connection with the 42 civil wars after the Cold War period. Similar patterns were noticed in this table, as well. During the Cold War period, when first interventions were made to help the opposition in a civil war, 15 out of 21 cases (about 71 percent) spurred counterinterventions. On the contrary, when first interventions were made to aid the side of the government, the number of cases with counter-intervention decreased to 9 out of 27 cases (about 33.3 percent).

Descriptive Statistics

107

Table 6-5. Number of Intervention During and After the Cold War During the Cold War

After the Cold War

Government

Opposition

Government

Opposition

No Second Intervention

11

6

3

2

BandWagoning Intervention

7

0

3

1

CounterIntervention

3

9

0

4

Both Types of Second Intervention

6

6

0

2

Totals

27

21

6

9

This pattern of second intervention continues after the Cold War period. After the Cold War, when first interventions were made to help the government side in the civil war, no counter-intervention was made. However, when first interventions were made to help the opposition, 6 out of 9 cases (about 66.7 percent) included counter-interventions.

CHAPTER SEVEN ANALYSIS I: FIRST INTERVENTION

In this chapter and in Chapter 8, specific hypotheses will be explored based on the preceding discussions. In this chapter, I will try to discover under which conditions we are likely to observe intervention in civil wars. Here, I am only interested in cases of first intervention. As discussed in the preceding chapters, there exist differences between first and the second interventions in terms of their intervention calculations. The conditions that lead to a second intervention, in the form of either bandwagoning or counter-intervention, will be investigated in the next chapter. In this chapter, two categories of hypotheses will be developed. The first category includes hypotheses about the first external intervention without specifying the target. This category focuses more on general conditions under which external forces consider intervention, regardless of the side they will choose to help. The second category includes hypotheses about the first external intervention, including specification of the target. Which conditions would lead external forces to intervene to help the government or the opposition in the civil war will be hypothesized.

7.1 Hypotheses without Specifying the Target The first set of hypotheses is made without specifying which side will receive the external help. General characteristics of the civil war and civil war states will be used to form this first set of hypotheses. The first set of hypotheses does not distinguish whether a certain condition will lead a foreign state to intervene to help the government or the rebels. First, the probability of foreign intervention is hypothesized with respect to the capability of the civil war state and of the potential intervener. Hypothesis 1: The greater the capability of the potential intervener, the more likely it will be to intervene in the civil war, other things being equal.

110

Chapter Seven

Most scholars agree that intervention is the foreign policy instrument typically used by superpowers. This is because potential interveners cannot use military force unless they have sufficient capability to do so. Fordham (2004) shows that greater military capabilities actually lead to the frequent use of military force. Common sense also tells us that one cannot help others unless one is capable. Theoretically, as modeled in Chapter 4, one of the conditions to be satisfied for external intervention is that intervention by the third-party should increase the probability of winning by the coalition of the target and the intervening forces. If a potential intervener does not have enough military strength, it is hard to expect that its involvement will increase the probability of winning. Thus, it can be hypothesized that stronger potential interveners are more likely to intervene in civil wars. Related to hypothesis 1, two more hypotheses can be developed. Hypothesis 1-1: The weaker the capability of the civil war state, the more likely it will be that there will be outside intervention, all other things being equal. If other conditions are the same, the potential intervener will enjoy a better chance of winning when the participants in the war are weak. I assume that when the capability of a state is low then the opposition power should be also weak. The opposition might be stronger than the government forces. However, when it comes to a foreign intervention calculation, the capability of the opposition cannot surpass the capability of the whole state. Thus, when the capability of the civil war state is weak, there will be more chances for foreign intervention in the civil war. Combining hypothesis 1 and hypothesis 1-1, we can develop another hypothesis with regard to the capabilities of the target and the potential interveners. Hypothesis 1-2: The more favorable the capability ratio is to the potential intervener, the more likely it is that the potential intervener will intervene in the civil war, all other things being equal. This hypothesis is more direct in addressing the dyadic relationship between the civil war state and the potential intervening states. Second, hypotheses are developed that address geographical proximity and the probability of foreign intervention.

Analysis I: First Intervention

111

Hypothesis 2: The closer the distance between the civil war state and the potential intervener, the more likely it is that there will be external interventions in the civil war, all other things being equal. Close proximity to a civil war country can increase the expected utility of the potential intervention. First, fearing the contagion effect or regional instability, potential interveners may have strong interests in terminating the civil wars in a neighboring country early. Also, close proximity may be an indication that there exist ethnic ties between the civil war state and the neighboring potential interveners. Ethnic affinities in cross-border countries are thought be generally higher than in those cases where vast distances separate ethnic groups. This is related to X, the magnitude of change after the civil war. If a civil conflict occurs in a distant place, the outcome of the civil war or the regional instability caused by it will not present much of a problem. Second, proximity can affect the cost of intervention, C. Not many countries can project their military powers to a distant place. To do so, the country in question must be equipped with modern technologies. Moreover, sending troops to a distant place also causes significant budget issues in domestic politics. The operational costs of sending troops to a distant place will be enormous, so not many countries can sustain such an effort. Conversely, small powers can send troops without losing too much power or without incurring too much when the target country is nearby. Even for major powers, sending troops to intervene in far-off civil wars is not an easy decision. Third, with relation to X, the magnitude of change after the civil war, the aim of the rebel group in the civil war is hypothesized to be closely related to the likelihood of foreign intervention. Hypothesis 3: If the aim of the rebel group in the civil war is to replace the government with its own, foreign states are more likely to intervene into the civil war, all other things being equal. X, the magnitude of change after the civil war, is greater when the aim of the rebel group in the civil war is to replace the government with its own, as compared to when the aim of the rebel group is limited to local issues. As described in Chapter 4, a greater value of X will increase the likelihood of foreign intervention regardless of its potential recipients. Fourth, economic benefits and the probability of foreign intervention are assumed to have a positive relationship.

112

Chapter Seven

Hypothesis 4: The more the dependence of the civil war state on primary commodity exports, the more likely it is that there will be external intervention in the civil war, all other things being equal. As discussed before, when there exist significant economic benefits such as export rights for primary commodities or natural resources, one can expect tangible economic benefits by aiding one side in the civil war. As indicated by Collier and Hoeffler (2005), the so-called “resource curse” provides low opportunity costs for rebellion, so it makes civil war more likely. The same reasoning can be applied to third-party intervention. Rebellions or even governments, in the context of a civil war, can utilize abundant natural resources to gain external aid. Fourth, civil wars during the Cold War period are assumed to attract more external intervention. Hypothesis 5: External interventions in civil wars were more likely during the Cold War period than in the post-Cold War period, all other things being equal. As discussed by Feste (1992), Regan (2000), and Lemke and Regan (2004), civil wars during the Cold War were more likely to lead to external intervention than were the civil wars after that period. Civil wars during the Cold War could easily escalate based on the ideological contests between the East and the West. Even participants in ethnic civil wars aligned themselves with ideological partners to get external aid (Kaufmann 1996). During the Cold War, external forces used intervention and counter-intervention as important policy instruments to advance their strategic interests. Since the Cold War, when these ideological issues became less salient, the magnitude of change after the civil war, X, from my model in the previous chapter, has not been great enough to outweigh the potential costs of intervention. As the Cold War ended, it became very difficult for the leaders to justify intervention and the costs of intervention in the civil wars of other states. Fifth, the domestic ethnopolitical condition of potential interveners should also be considered. Hypothesis 6: If a government leader is from an ethnic minority, it is more difficult for that government leader to pursue intervention policies. Thus, intervention in foreign civil wars is less likely when a potential government is not ethnically dominant.

Analysis I: First Intervention

113

As discussed by Carment and James (2003) and Carmet, James, and Taydas (2006), the ethnopolitical condition of potential interveners is an important determinant of foreign intervention in civil wars. Even if their discussions are limited to interventions in the ethnic civil wars, their framework can be used here. They argue that “ethnically diverse states are less likely to initiate crisis with violence” (Carment, James, and Taydas 2006, 39). This means, conversely, that forceful intervention is more likely when there is less political or ethnopolitical resistance from the constituents.

7.2 Hypotheses with Specifying the Target A second set of hypotheses can be made that specifies on behalf of which side of the civil war foreign states are likely to intervene. In contrast to the first set of hypotheses, these hypotheses specify which side, that of the government or that of the rebels, will be targeted by potential interveners. More specifically, given those general conditions of civil war and civil war states that make foreign intervention easier, the second set of hypotheses makes specific predictions about what kinds of conditions will determine whether foreign countries intervene to help the government side or the rebel side. These hypotheses and their empirical testing are thought to contribute more to the current literature on external intervention in civil wars. Previous studies do not make hypotheses with regard to the recipients of foreign intervention. In most cases, they only care about whether or not there was intervention in a certain civil war (Regan 2000) or whether there has been intervention or not in certain dyads between the civil war state and the potential interveners (Lemke and Regan 2004). These studies do not investigate whether certain conditions of civil wars lead foreign states to intervene for the government or the rebels in a civil war. Certain sets of variables are only meaningful when the hypotheses made specify the beneficiaries of foreign intervention. First, a formal alliance is assumed to have an impact on decisions regarding foreign intervention. Hypothesis 7: A potential intervener is more likely to intervene on the side of the government if it has made a formal alliance with the civil war government, all other things being equal. I expect that a potential intervener is more likely to support the government side when the two parties share convergent interests. One way

114

Chapter Seven

that inter-state shared interests are shown is through the formation of an alliance. Even though a formal alliance may not always be honored in an inter-state conflict, it is expected that an allied foreign government is more likely to help its alliance partner. The interpretation of Hypothesis 7 is, however, not as simple as it appears. This is a very complicated situation in reality. When a potential intervener with a formal alliance decides not to intervene on the side of the government, it can mean two things. First, it may mean that a potential intervener believes that the potential benefits of intervention do not outweigh the potential costs of intervention, even given the reputational cost of not honoring the formal alliance. Second, it may mean that in abstaining from intervention, the intention of a potential intervener is actually to support the opposition. The potential intervener may expect that if it does not provide military support to the government the opposition will win the civil war. This is a plausible scenario if a potential intervener does not care about the possibility of regime change in the civil war state. However, I do not have any way to distinguish one from the other in this non-intervention case. Related to Hypothesis 7, another hypothesis can be made with respect to formal alliances. Hypothesis 7-1: At the same time, it is less likely for a potential intervener to intervene on the side of the opponent if it has a formal alliance with the civil war government, all other things being equal. As discussed above, there are two ways to interpret a potential intervener’s decision not to intervene. The state can implicitly help the opposition by abstaining from intervention on the side of the government with whom there is a formal alliance. By not helping the government explicitly, the state implicitly and in actuality helps the opponent. However, it is important to note that explicit military intervention on the side of the opponent at the expense of a formal alliance is a completely different story. A potential intervener would experience more of a reputational burden in explicitly helping the opponent when its ally, the government in the civil war, is in danger. Thus, it is hypothesized here that the existence of a formal alliance relationship will make it less likely for a potential intervener to intervene on the side of the opponent in the civil war. Second, the proportion of the population representing the governmental leader’s ethnic group in the civil war country is assumed to have an impact on decisions regarding intervention.

Analysis I: First Intervention

115

Hypothesis 8: As the proportion of the population representing the governmental leader’s ethnic group in the civil war state increases, the potential intervener is more likely to intervene on the side of the government, all other things being equal. In most civil wars, the number of armed personnel is very important, since warfare is fought on the ground with conventional military forces. It is not about which side has better-developed, more high-tech weaponry. It is rather about how many people actually control what portion of the land in the country. Especially in ethnic civil wars, the outcome of the war can be determined according to how many armed soldiers each side has and how many they can recruit from their ethnic base. This might explain why we have often observed in many ethnic civil wars that many young boys have had limbs amputated. Those gross incidences indicate that number of available armed soldiers is an important determinant in an ethnic civil war. Given the discussion above, it can be assumed that the proportion of a population represented by a particular ethnic group roughly determines its power base. Thus, if the proportion of the population representing the government leader’s ethnic group is large, it is believed that the probability of a win by the government in the civil war will be higher. If this is the case, then it will be easier for a foreign state to intervene on the side of the government because the intervention cost will be less. Related to Hypothesis 8, another hypothesis can be drawn. Hypothesis 8-1: As the proportion of the population representing the governmental leader’s ethnic group in the civil war state decreases, a potential intervener is more likely to intervene on the side of the opponent, all other things being equal. Based on the same logic that is applied in Hypothesis 8, if the proportion of the population representing the government leader’s ethnic group is small, it is believed that the probability of a win by the opponent in the civil war will be higher. If this is the case, then it will be easier for the foreign state to intervene on the side of the opponent because there will be fewer intervention costs for the opponent. Related to Hypothesis 8, Hypothesis 8-1, and also Hypothesis 6, transnational ethnic ties were also considered to be included in the hypotheses and the analysis in the next chapter. Actually, the variable for transnational ethnic ties turns out to be statistically significant and the direction of the effect as expected. Transnational ethnic ties are operationalized to be existent if the governmental leaders of a dyad are

116

Chapter Seven

from the same ethnic group. However, a deeper investigation of the data shows that the aggregate level data for ethnic groups is similar to that referring to race. Meaningful transnational ethnic ties in the aggregate level data exist between Rwanda and Burundi in 1994, between Burundi and Rwanda in 1994 and 1997–2000, and between Cyprus and Greece. The governmental leaders of both Rwanda and Burundi were from the Hutu-Twa ethnic group in 1994 and from the Tutsi ethnic group during the 1997–2000 period. In other cases, they were mostly White-White, ArabArab, and Mestizo-Mestizo. These pairs do not have any theoretical meaning and cannot be interpreted as transnational ethnic ties. Thus, in this book, the variable representing transnational ethnic ties is excluded from the study. Third, it is assumed that democracies are less prone to intervene in a civil war to overthrow the incumbent government. Hypothesis 9: The more democratic a potential intervener is, the less likely it is to intervene on the side of the opponent in the civil war, all other things being equal. Even though we have observed some cases where a democratically developed country intervenes or wages a war against another country to replace a regime, it is not the norm. As discussed in democratic peace theory texts, democracies have many constraints on using violence against other countries. Sending military forces to help the opponent in a civil war can mean that an intervening state is actually aiming to replace the regime in the civil war state with a new one, that is, with the rebels. Democratic countries have diplomatic and reputational reasons not to intervene on behalf of the opponent in the civil war.

7.3 Research Design-Selection of Cases The hypotheses outlined in the last section will be tested by using the data regarding the civil wars in the post-World War II era described in Chapter 2. The most difficult question to answer in the empirical test might be what will be the unit of analysis in the study. One obvious problem in Regan (2000) is that he uses the conflict as the unit of analysis. As he admits, “There are conceptual problems in using the conflicts as the unit of analysis because the emphasis of the empirical model shifts from the perspective of the individual decision maker to the aggregate case, asking in essence whether certain structural and contextual conditions increase the probability of an intervention” (Regan 2000, 52). Using the conflict as

Analysis I: First Intervention

117

the unit of analysis results from the difficulty of identifying a population of potential interveners. To correct this problem, Lemke and Regan (2004) “treat each civil war as an intervention opportunity for every member of the international system (as defined by COW) in existence while the civil war occurred” (155). They use the dyads as the unit of analysis. However, their solution creates a few problems as well, both theoretically and methodologically. They treat a civil war as an intervention opportunity for every country in the world regardless of the duration of the civil war. They use all the variables from the year when the civil war started. This is mainly because their model does not consider the sequential moves by the potential interveners. It is more of a snapshot analysis. Their model does not distinguish between the first and second (either counter- or bandwagoning) interventions. All dyads in their analysis are recorded on a cross-sectional basis. Thus, their model does not cause much of a problem even after they expand the cases to include every country in the world system. However, as in my model, if you want to incorporate the sequential moves of the potential interveners, their solution cannot be used. To incorporate the sequential moves, annual dyads should be used in the empirical test. Then a difficult question becomes what kind of annual dyads should be used in the test. If all possible annual dyads are to be used, then there will be an obvious bias toward non-intervention. The civil war in Myanmar from 1948 to 1988 alone can generate 40 years multiplied by the number of countries in the world system in each year, which includes roughly more than 4,000 observations. It would not be practically manageable or theoretically meaningful to include all yearly dyads in the test. This is not only a problem of analyzing such a huge amount of data but also an issue of collecting accurate information for so many variables believed to be potentially important. We have a practical reason to reduce the number of observations. One way to reduce the number of observations is to select a random sample from the population. However, as Lemke and Reed (2001) note, using a random sample is “impractical for analysis of international conflict because the behaviors of interest, wars, and militarized disputes are so rare that few if any would be included in a random sample of dyadic annual observations” (126–27). There is a very high chance that a random sample would select out most of the conflicts. There is also a theoretical concern. Using dyads as a unit of analysis is a way to treat each civil war as an intervention opportunity for all countries. The question here is whether all civil wars are intervention opportunities for all countries. Not all countries will see a civil war as an

118

Chapter Seven

opportunity for intervention. Some countries would never consider intervention as one of their policy tools. Maoz and Russett (1993) argue that an analysis of all dyads is inappropriate because “the vast majority are nearly irrelevant. The countries comprising then were too far apart and too weak militarily, with few serious interests potentially in conflict, for them plausibly to engage in any militarized diplomatic dispute” (627). Thus, in this book, I consider the method that international conflict researchers have developed to reduce the number of observations. This is the notion of politically relevant dyads (Maoz and Russett 1993)—pairs of states directly or indirectly contiguous and/or in which at least one of the states is a major power. As Lemke and Reed (2001) note, “Relevant dyads comprise a small fraction of the population of annual dyadic observations yet account for the vast majority of interstate wars and disputes” (127). Even though relevant dyads were developed for a practical reason, their use is also theoretically more reasonable when the concept of opportunity is considered. Lemke (1995) also argues that “relevant dyads matter because they comprise the correct referent group, and thus function as a true control group, against which war dyads are compared” (29). As defined, two criteria should be met for a dyad to be considered politically relevant: geographic contiguity and major power status. To meet the contiguity criterion, states must physically touch or be separated by a small distance of open water (up to 400 miles in most classifications). I use the 400-mile rule in this book. Major power status is usually based on a state’s Correlates of War designation as a major power. I also dropped small countries, whose population is less than 500,000, by following the Correlates of War project. Those states are too small or weak to function as potential interveners. Consequently, based on the Sambanis (2004) civil war list (excluding Chad 1994–1997, Korea 1948–1949, Namibia 1973–1989, and Syria 1979–1982 for ambiguity) and the politically relevant dyads method, the data set used in this book comprises a total of 10,742 observations. Using politically relevant dyads as a unit of analysis is, of course, not free from criticism. As already discussed by Maoz and Russett (1993), a substantial number of militarized interstate disputes (MIDs) occur in irrelevant dyads. According to Maoz and Russett (1993), 26 percent of the disputes in their timeframe (1946–1986) occur among irrelevant dyads. In my 10,742 annual dyads from 147 civil wars, military interventions occur in 859 annual dyads. In terms of cases, there are 153 individual interventions in 147 civil wars. Among those 153 individual interventions, 33 interventions were made outside politically relevant dyads. Consequently, those 33 interventions were not included in this analysis.

Analysis I: First Intervention

119

Then, how much of a problem is caused by the use of politically relevant dyads in the analysis? As Lemke (2001) notes, using relevant dyads can cause both measurement error and selection bias. However, he also finds that there is “little or no evidence that such error or bias leads to erroneous estimation. The error and bias appear to be relatively small and substantively unimportant” (Lemke 2001, 140–41).

7.4 Variables and Method Dependent Variable First Intervention The dependent variable in this first analysis is whether an intervention occurs on the side of the government or the opposition in the annual dyads when no other states intervene in the same civil war. This variable indicates whether the first intervention in the civil war is made on behalf of the government or the opposition in the civil war. As discussed in chapter 2, only overt military intervention is recorded as an intervention. Regan (2002) actually uses a six-point nominal scale to identify the existence and type of military intervention, if any. According to his coding, sending troops or naval forces, providing equipment or aid, providing intelligence or military advisors, providing air support, or imposing military sanctions were recorded as different types of military intervention. Among those various types of military intervention, I only use troops and naval and air support as types of military intervention according to the definition of military intervention in Chapter 2. Thus, if an intervention on behalf of the government in a dyad exists, it is coded 1. If an intervention is made for the opposition, it is coded 2. If no intervention is recorded, then it is 0. Independent Variables Alliance This variable is generated by EUGene v 3.2. This dyadic alliance relationship in EUGene uses the COW v3.0 alliance data released in 2002. The codes for various types of alliances are 1 = defense pact, 2 = neutrality, 3 = enténte, and 4 = no agreement.

120

Chapter Seven

Cap_1 This variable is the National Capability Index of the civil war state. Cap_2 This variable is the National Capability Index of the potential intervening state. Cap Ratio This variable is generated by the ratio of Cap_1 to Cap_2. Thus, this is simply Cap_1 divided by Cap_2. Majorpower_2 This variable indicates the major power status of the potential intervening state. If it is a major power (one of the Security Council member countries), it is coded 1, otherwise 0. EUGene v3.2 actually gives major power status to Germany and Japan after 1990. However, I excluded those two states from the politically relevant dyads if they were included for their major power status. In particular, the Japanese Constitution’s Article 9 states that “aspiring sincerely to an international peace based on justice and order, the Japanese people forever renounce war as a sovereign right of the nation and the threat or use of force as means of settling international disputes. (2) In order to accomplish the aim of the preceding paragraph, land, sea, and air forces, as well as other war potential, will never be maintained. The right of belligerency of the state will not be recognized” (National Diet Library). Thus, Japan cannot be included. Distance According to EUGene v3.2, the distance between two states will be calculated as the distance between their two capitals. The distance between two contiguous states is coded 0. Islands If a civil war state is located on the island and, thus, separated from other countries by open sea, it is coded 1, otherwise 0.

Analysis I: First Intervention

121

War type Following Sambanis (2004), identity civil wars (ethnic and/or religious civil wars) are coded 1 and identity civil wars are coded 0. Aim Following Fearon (2004), the different aims of the rebels in the civil war are coded. If the rebels aim to overthrow the government, the aim is coded 1. If the rebels aim for secession or local autonomy, the aim is coded 3. If the aim of the rebel is mixed or ambiguous, it is coded 2. Sxp This variable is of the greatest interest to Collier and Hoeffler (2000). This is a World Bank measure of primary commodity exports as a proportion of GDP. The data is from Collier and Hoeffler (2005). Since the data only extend back to 1960, I use the 1960 value as a proxy for the pre-1960 period. Cold War If the annual dyads exist in the Cold War period (until 1990), they are coded 1, otherwise 0. Leader’s Ethnic_1 This is the proportion of the population representing the ethnic group of the governmental leader of the civil war state. Leader’s Ethnic_2 This is the proportion of the population representing the ethnic group of the governmental leader of the potential intervening state. Both Leader’s Ethnic_1 and Leader’s Ethnic_2 data were borrowed from Professor Kimuli Kasara at Columbia University through personal communication. ELF_1 This is the index of ethno-linguistic fractionalization (ELF) of the civil war state. ELF_2 This is the index of ethno-linguistic fractionalization (ELF) of the potential intervening state. The index of ethno-linguistic fractionalization data is from Fearon (2004).

Chapter Seven

122

Dem_1 This is the polity score of the civil war state from Polity IV. Dem_2 This is the polity score of the potential intervening state from Polity IV. Method Since the dependent variable in this analysis is nominal, the categories are unordered. Thus, a multinomial logit model is used. In this analysis, the multinomial logit model can be written as Ž Ÿ௠ȁ௕ ሺܺሻ ൌ  Ž

ܲ௥ ሺ‫ ݕ‬ൌ ݉ȁܺሻ ൌ ܺߚ௠ȁ௕ ݂‫ ݉ݎ݋‬ൌ ͳ‫ܬ݋ݐ‬ ܲ௥ ሺ‫ ݕ‬ൌ ܾȁܺሻ

where b is the base category, which is referred to as the comparison group. In this analysis, the base category is 0, that is, when no intervention occurs. This represents the log odds of an outcome as compared with the base category. To compute the predicted probabilities, the equation can be solved as ܲ௥ ሺ‫ ݕ‬ൌ ݉ȁܺሻ ൌ 

‡š’ሺܺߚ௠ȁ௕ ሻ ௃ σ௝ୀଵ ‡š’ሺܺߚ௝ȁ௕ ሻ

. ෢ With this equation, we can obtain estimates ߚ෢ ଵȁ଴ , and ߚଶȁ଴ , where ߚ଴ȁ଴ ൌ Ͳ.

7.5 Analysis Results are reported in Table 7-1. Overall, the model used to explain the pattern of the first intervention shows a decent fitness according to the Likelihood Ratio (LR) Chi-Square test, the probability of getting an LR test statistic, and a Pseudo R2 of 0.2699. Interpreting the multinomial logit result is not easy. Coefficients indicate the change in log odds of the dependent variable per 1 unit change in the continuous independent variable or the difference in the log odds of the dependent variable for one value of the categorical variable versus the reference group. Interpreting the log odds does not make any sense if we try to interpret them directly.

Analysis I: First Intervention

123

The easiest way to interpret the coefficients of the results is to use the predicted probabilities by putting the coefficients and the values of the variables into the equation below. ܲ௥ ሺ‫ ݕ‬ൌ ݉ȁܺሻ ൌ 

‡š’ሺܺߚ௠ȁ௕ ሻ ௃ σ௝ୀଵ ‡š’ሺܺߚ௝ȁ௕ ሻ

Let us take the case of the intervention by France for the government in the Djibouti Civil War of 1991–1994 as an example. The alliance level of the two countries is 4. The capability index of Djibouti is 0.000122 and that of France is 0.021242. The capability ratio of the two countries is 0.0057433. The major power status of France is 1. The proportion of the population representing the governmental leader’s ethnic group is 0.5 for Djibouti and 0.85 for France. ELF is 0.6933391 for Djibouti and 0.2610468 for France. The adjusted Polity Score is -7 for Djibouti and 9 for France. The distance between the two countries is 3477. Since this is an identity civil war, the war type is 0. The aim of this civil war is coded 2. Djibouti is not located on an island. Thus, the island variable is 0. Since this is in 1992, the Cold War variable is also coded 0. Sxp is recorded as 0.26. Putting these values into the equation above, I can get a predicted probability of 0.977302. My model predicts that France will intervene in the civil war in Djibouti, with a 97.7 percent probability specifically. France actually intervened in Djibouti in 1992. Table 7-1. Multinomial Logit for the First Intervention Cases Number of Obs = LR Chi2 (34) = Prob > Chi2 = Pseudo R2 = GOVERNMENT Alliance Cap_1 Cap_2

Coefficient (Standard Error) - 0.874392** (0.0717618) - 135.79** (37.01034) 10.08067** (1.357767)

10490 1077.3 0.000 0.2699 Relative Risk Ratio (Standard Error) 0.4171155** (0.0299329) 1.06e-59** (3.94e-58) 23876.91** (0.833929)

124

Cap Ratio Major Power_2 Distance Islands War Type Aim Sxp Cold War Leader’s Ethnic_1 Leader’s Ethnic_2 ELF_1 ELF_2 Dem_1 Dem_2 Constant OPPOSITION Alliance Cap_1 Cap_2 Cap Ratio Major Power_2 Distance Islands

Chapter Seven - 0.7489721** (0.1763614) 0.3394827 (0.3073786) - 0.0001485** (0.0000448) -2.996345** (0.5007554) -0.1756731 (0.1580961) -0.7510835** (0.132982) -0.0089878 (0.0303239) 0.2487227 (0.1868898) 2.071839** (0.2969091) 1.630836** (0.5421213) 1.225996** (0.3464581) 1.867577** (0.498961) 0.0000249 (0.0125763) 0.0694539** (0.0114884) -3.163578** (0.535182) Coefficient (Standard Error) -0.4619416** (0.0671928) -111.5364** (34.1831) 22.43883** (4.119013) -0.2820453** (0.0657394) -4.179937** (0.7842108) -0.0002199* (0.0001228) 1.002474** (0.3011977)

0.4728524** (0.0833929) 1.404221 (0.4316275) 0.9998515** (0.0000448) 0.0499694** (0.0250224) 0.8388922 (0.1326255) 0.471855** (0.0627482) 0.9910525 (0.0300526) 1.282386 (0.239665) 7.939411** (2.357284) 5.108142** (2.769232) 3.407559** (1.180576) 6.472596** (3.229573) 1.000025 (0.0125766) 1.071923** (0.0123147) N/A Relative Risk Ratio (Standard Error) 0.6300591** (0.0423354) 3.63e-49** (1.24e-47) 5.56e+09** (2.29e+10) 0.7542395** (0.0495833) 0.0152995** (0.011998) 0.9997801* (0.0001228) 2.725014** (0.8207679)

Analysis I: First Intervention 1.10417** (0.2194075) 0.0039825 Aim (0.122079) -1.705807* Sxp (0.9018071) 0.2615337 Cold War (0.1961854) -1.408256** Leader’s Ethnic_1 (0.3514302) 0.7054931* Leader’s Ethnic_2 (0.3621348) -0.6514516* ELF_1 (0.0120657) 0.4320684 ELF_2 (0.3954942) -0.0453904** Dem_1 (0.0120657) -0.0292318** Dem_2 (0.012052) -2.314182** Constant (0.3920375) First Intervention = 0 is the base outcome ** significant at 95 %, * significant at 90 %.

War Type

125 3.016718** (0.6618906) 1.00399 (0.1225662) 0.1816257* (0.1637914) 1.298921 (0.2548292) 0.2445695** (0.0859491) 2.024845* (0.7332668) 0.5212885* (0.1796973) 1.54044 (0.6092353) 0.9556243** (0.0115303) 0.9711913** (0.0117048) N/A

Even though it is easier to interpret the predicted probabilities, this interpretation of the predicted probabilities does not provide a means of interpreting the magnitude of the coefficients directly. Another way to interpret the multinomial logit results is to use relative risk ratio. The relative risk ratio can be obtained by exponentiating the multinomial logit coefficients, ecoef. This procedure can be done using the rrr option in STATA. However, the exponentiated coefficients are commonly interpreted as odds ratios. Consequently, the standard interpretation of the relative risk ratios is for a unit change in the independent variable, the relative risk ratio of outcome m relative to the referent group is expected to change by a factor of the respective parameter estimate if the variables in the model are held constant. I report the relative risk ratio of each variable in Table 7-1. Even though the interpretation of the relative risk ratio is not intuitively clear, we can obtain an idea of the magnitude and the direction of the effect of each variable. With respect to Hypothesis 1, Hypothesis 1-1, and Hypothesis 1-2, I find that these hypotheses are supported. The relationship between

126

Chapter Seven

capability and intervention is statistically significant at a 95 percent confidence level. As the capability index of the civil war state is lower, the capability index of the civil war state is higher and the capability ratio between the civil war state and the potential intervening state is not favorable to the civil war state, external intervention is more likely. Comparing the relative risk ratios of intervention on behalf of the government and intervention for the opposition, we can see that power parity is more obviously required when the intervention is executed on the side of the opposition. This result to some extent confirms the assumption that the government is stronger than the opposition when a civil war commences, as put forth by Mason, Weingarten, Jr., and Fett (1999). Since the government is stronger than the opposition, the potential intervener should be stronger than when it intervenes on the side of the government to expect a favorable outcome. I also find support for Hypothesis 2. Distance between the civil war state and the potential intervening state is statistically significant at a 95 percent level in the case of intervention on the side of the government and at a 90 percent level in the case of the intervention on the side of the opposition. In either case, the long distance makes the potential intervener abstain from intervening in a foreign civil war. With regard to Hypothesis 3, foreign countries are more likely to intervene on the side of the government when the aim of the civil war is to overthrow the government. As the variable Aim shows in Table 7-1, the odds of intervention on the side of the government decrease by a factor of 0.471855 for a unit change in the variable Aim. This result is statistically significant at the 95 percent level. This result conforms to the model developed in Chapter 4. X, the magnitude of change after the civil war, is larger when the war is fought between the government and the opposition for the control of the government than when it is fought for local issues. Thus, when rebels aim to replace the government with their own, it is more likely that foreign countries will intervene to rescue the government in the civil war. However, the effect on intervention on the side of the opposition is not statistically significant. Thus, I can conclude that Hypothesis 3 is partially supported by the data. Hypothesis 4, which concerns the dependence of the civil war state on primary commodity exports, is not supported. It is hypothesized that heavy dependence on primary commodity exports will attract more intervention on both sides. However, this is not statistically significant for intervention on the side of the government, and it is statistically significant at only a 90 percent level in the case of intervention on behalf of the opposition. With

Analysis I: First Intervention

127

regard to the latter case, the direction is opposite to what was expected. Here, for a unit increase in sxp, the odds of intervention on behalf of the opposition are decreased by a factor of about 0.18. More dependence on primary commodity exports decreases the probability of intervention on the side of the opposition. Whether this results from the use of the fiveyear interval or the estimated proxy for the pre-1960 period cannot be determined. As Fearon (2005) points out, the theories about the so-called sxp variable may be flawed. This should be studied further in the future. The Cold War does not seem to have any statistically significant effect on intervention on behalf of either side. It was hypothesized that the polarized structure associated with the Cold War would lead potential interveners to use intervention as a policy instrument. However, Hypothesis 5 is not supported. After controlling for the effects of other variables, the Cold War variable turns out insignificant. This finding is clearly different from those of Regan (2000) and Lemke and Regan (2004). Both argue that the Cold War was clearly a major factor in the decision to intervene. However, at the very least, the Cold War was not an important factor in first interventions. I find that Hypothesis 6, which suggests that intervention in a foreign civil war is less likely when the governmental leader of the potential intervening state is from an ethnic minority group, is supported. As the proportion of the population representing the leader’s ethnic group increases, the likelihood of intervention on the side of the government increases at a 95 percent confidence level and on the side of the opposition at a 90 percent level. To obtain a clearer picture of this dynamic, I also calculated the standardized factor change here. With respect to intervention on the side of the government, for one standard deviation increase in the Leader’s Ethnic_2 variable, the odds of intervention are 1.5882 times greater with all other variables remaining constant. That is, one standard deviation will increase the odds by 58.82 percent. With respect to intervention on the side of the opposition, an increase of one standard deviation for the Leader’s Ethnic_2 variable leads to the odds of intervention being 1.2216 times greater with all other variables remaining constant. That is, one standard deviation of change will increase the odds by 22.16 percent. Turning to the hypotheses regarding the specification of the target, we can first see that Hypothesis 7-1, regarding the relationship between the alliance and the intervention on the side of the opposition, is contrary to my expectations. Hypothesis 7, the relationship between the alliance and the intervention on the side of the government, is statistically significant at a 95 percent level, and the relationship is as expected. As the alliance level

128

Chapter Seven

moves from the strongest (defense pact) to the weakest (no agreement), for a one-unit change, the odds of intervention on the side of the government are decreased by a factor of 0.4171155 with all other variables remaining constant. This result is as expected, since it is believed that an allied partner will help the government to which it is allied in a civil war. However, contrary to expectations, as the alliance level moves from the strongest (defense pact) to the weakest (no agreement), for a one-unit change, the odds of intervention on the side of the opposition are also decreased. This time, they are decreased by a factor of 0.6300591 with all other variables held constant. That is, the magnitude of the effect is bigger than the case of intervention on the side of the government. This suggests that the alliance relationship may not be honored in the face of possible competing interests. However, this surprising result is to some extent similar to the findings of Bueno de Mesquita (1981) and Ray (1990). Bueno de Mesquita (1981) shows empirical evidence that war is more likely between very close allies than between enemies. Ray (1990) concludes that “in light of the fact that it would be surprising to find that allies are even as conflict prone as unallied pairs of states, it is not unreasonable to conclude that allied dyads were disproportionately involved in international conflict with each other in the 1816–1974 time period” (86). Thus, with regard to the relationship between an alliance and intervention on the side of the opposition, more study is required to develop hypotheses in the future. Hypothesis 8 and Hypothesis 8-1 are very well-supported by the data. As the proportion of the population representing the governmental leader’s ethnic group in the civil war increases, intervention is more likely to occur on behalf of the government. This finding is statistically significant at a 95 percent level. Via the standardized factor change method, for a standard deviation (0.28599801) increase in the Leader’s Ethnic_1 variable, the odds of intervention on the side of the government become 1.8086 times greater with all other variables held constant. On the contrary, the odds of intervention on the side of the opposition decrease by a factor of 0.6685 for a standard deviation increase in the Leader’s Ethnic_1 variable. With respect to Hypothesis 9, more democratically developed countries are proved to be less likely to intervene on the side of the opposition. A one-unit increase in the Polity score leads to about a 2.9 percent decrease (the factor change is about 0.9712) in the odds of intervention on the side of the opposition, holding all other variables constant. If we see the intervention on the side of the opposition as being in part a war against the government in the civil war state, this pattern can be easily understood.

Analysis I: First Intervention

129

Related to Hypothesis 9, more democratically developed countries are more likely to intervene on the side of the government. A one-unit increase in the Polity score leads to 1.071923 times greater (about 7.2 percent increase) odds of intervention on the side of the government with all other variables remaining constant. This result is statistically significant at a 95 percent level. There are also other important findings to be culled from this analysis. Some variables are found to be statistically significant, but theoretical interpretations of those variables are not easy to develop. The variables below are found to be statistically significant. However, with the current literature, it is difficult to determine what leads to these statistical outcomes. If these findings hold true even after a robust test with different operationalizations of variables and different statistical methods, we need to find out the reason for the relationships found here. With regard to war types, foreign countries are more likely to intervene on the side of the opposition when the war in question is an identity civil war. The type of war is not a statistically significant factor when it comes to intervention on the side of the government. However, an identity civil war increases the odds of intervention on the side of the opposition by about 10.4 percent with all other variables held constant. The index of ethno-linguistic fractionalization (ELF) also has a statistically significant impact on the likelihood of foreign intervention. For a standard deviation (0.26913268) increase in the ELF_1, the odds of intervention on the side of the government are 1.3909 times greater. That is, an increase of one standard deviation in the ELF_1 will lead to an increase the likelihood of intervention on the side of the government of 39.09 percent with all other variables remaining constant. This is a fairly strong effect. One possible interpretation is that when a society is more ethno-linguistically diverse, it is difficult for a rebel group to have a strong power base. Conversely, government forces in such a society are thought to be stronger than the rebel forces would be. This indicates fewer costs for potential interveners. On the contrary, for a standard deviation increase in the ELF_1, the odds of intervention on the side of the government decrease by 16.08 percent (standardized factor change is 0.8392) with all other variables held constant. That is, an ethnically less fragmented society will attract more external intervention on the side of the opposition. It is imaginable that when a society is less fragmented in terms of ethnic groups, it is easier to organize a fairly strong ethnic power base. The probability of mobilizing a stronger opposition is much higher. Thus, in this case, the capability of the opposition in the civil war is thought to be higher. This situation provides

130

Chapter Seven

an opportunity for intervention on the side of the opposition associated with fewer costs. When it comes to the ELF of the potential intervening states, to construct an interpretation is more difficult. The ELF of a potential intervening state is positively correlated with the likelihood of intervention on the side of the government, whereas it is statistically insignificant for intervention on the side of the opposition. For a standard deviation (0.25669837) increase in the ELF_2, the odds of intervention on the side of the government are 1.6151 times greater. That is, an increase of one standard deviation in the ELF_2 will lead to an increase the likelihood of intervention on the side of the government of 61.51 percent with all other variables remaining constant. This is a fairly strong effect. With regard to the Islands variable, it is easily assumed that foreign countries are less likely to intervene in the civil wars of states located on the islands. One has to have extra military equipment besides ground forces to launch an intervention in a country located on an island. Thus, it is easy to understand the finding that foreign countries are less likely to intervene on the side of the government if the civil war state is located on an island. Being on an island decreases the odds of intervention on the side of the government by a factor of 0.471855 with all other factors constant. The likelihood of intervention decreases as expected. However, it turns out that being on an island increases the odds of intervention on the side of the government by a factor of 3.016718 with all other factors remaining constant. I think that the theory can hardly explain this result. Overall, the results of the analyses support the hypotheses developed using the framework of the model, suggesting that the decision to intervene results from a mixture of constraints along four dimensions. It can be inferred from these results that strategic considerations, including capability, distance, and the aim of the civil war, as well as domestic political conditions, including ethnic composition and domestic democratic level, are all important.

CHAPTER EIGHT ANALYSIS II: SECOND INTERVENTION

In this chapter, I turn my focus to the intervention that occurs after the first intervention by another country. The analysis of second interventions in this chapter is more interesting to see because all interventions, regardless of whether they are first or second, are treated equally in previous studies. In this chapter, two separate analyses will be made. One will concern second intervention when the original intervention occurs on the side of the government. This analysis investigates the reaction by other potential intervening states when a country intervenes in the civil war on the side of the government. The other regards second interventions executed when the first intervention is made on the side of the opposition. Then, I am going to compare those two analyses. Preceding the discussion of the hypotheses, the concept of second intervention should be defined. In reality, not all second interventions, either counter-interventions or bandwagoning interventions, are the same. There is also the matter of sequential order. For example, in Laos’s civil war during the period of 1960–1973, the United States made the first intervention on the side of the government in September 1960. Then, the Soviets counter-intervened for the opposition in December 1960. After that, North Vietnam (DRV) intervened, following the Soviets, on the side of the opposition in January 1961; China did the same in January 1962. Later on, in September 1969, Thailand intervened on the side of the government. Thus, the Soviet counter-intervention in December 1960 was basically the only pure counter-intervention. The intervention by Vietnam was bandwagoning counter-intervention. The intervention by Thailand displayed characteristics of both counter-intervention and bandwagoning. Conceptually, those interventions cannot be treated the same way because these intervention decisions are also reactions to earlier third-party intervention into the same conflict. Consequently, the decision making calculus is completely different. For a more rigorous conceptualization, these sequential dynamics should also be incorporated.

132

Chapter Eight

However, the selection process will leave us with a very small number of observations in each case, which makes it very difficult to have enough freedom for statistical analysis. Thus, I choose to view all second interventions as reactions to the first intervention. In the example above, Vietnam’s decision calculus is a reaction to the intervention by the United States on the side of the government. In reality, it should be a reaction to the intervention by the United States on the side of the government and the intervention by the Soviets on the side of the opposition. In the same way, China’s intervention decision is assumed to be based on the intervention by the United States on the side of the government. In a rigorous conceptualization, its decision calculus should include the intervention by the United States on the side of the government, the intervention by the Soviets on the sides of the opposition, and the intervention by Vietnam on the side of the opposition. I admit that this is a compromise. However, this is one-step-forward progress from the previous research in that at least I distinguish the first intervention from ones that follow.

8.1 Hypotheses Two more hypotheses are added to the eight hypotheses developed in the previous chapter. Thus, the nine hypotheses used in the previous chapter are also to be used in this chapter, too. Hypothesis 10: A potential intervener is more likely to intervene on the opposite side if this state has a weak or no formal alliance with the state that intervenes first, all other things being equal. This is the hypothesis regarding counter-intervention. As discussed previously, there can be two plausible explanations regarding the content of this hypothesis. Let us imagine that State A has a vested interest in the civil war. Thus, the outcome of the civil war will affect State A. However, State A believes that its vested interest in the civil war state will not be affected so far. Thus, State A has decided not to intervene. While State A is viewing the civil war from outside, State B intervenes in support of the opposite side in the civil war. State B’s involvement now changes the balance of power between the participants in the civil war. The probable outcome of the civil war is now changed because of the intervention on the part of State B. As State A does not desire this change in the expected outcome, State A now has very strong motives to intervene.

Analysis II: Second Intervention

133

Another counter-intervention scenario is also possible. In this case, State A initially does not have a strong interest in the outcome of the civil war. As long as the effect of the outcome of the civil war is limited to the civil war state, State A is indifferent to the outcome. That means that State A will not intervene. However, the situation changes when, for any reason, State B intervenes into the civil war. With the help of State B, the government-to-be group in the civil war will win, and it will align itself with State B. State B will enjoy a very strong position in the civil war state as a result. This expected outcome of the civil war bothers State A. Since what State A cares more about is the strategic and power balance between State A and State B, the alignment of the civil war state with State B is not an acceptable outcome for State A. To prevent this from happening, State A will intervene on behalf of the opposite side, the one that will fight against State B’s coalition powers. However, Hypothesis 10 does not conversely mean that a potential intervener is more likely to intervene on the opposite side if it has a weaker formal alliance relationship with the state that intervenes first, all other things being equal. This is bandwagoning intervention. Thus, when a state’s strongly allied partner intervenes in a civil war on the side of the government, that allied state is more likely to intervene on the side of the government because it now has fewer burdens. This reasoning will lead to bandwagoning intervention. This scenario does not always play out in real life, however. There also exists the concept of the free-rider. Since other countries already have intervened in the civil war, a free-rider may decide not to intervene but rather to enjoy the outcome, which it believes will be favorable to it because of the first intervention on the side for which the free-rider would have intervened. Since there are conflicting interests between bandwagoning intervention and free-rider non-intervention, we cannot say that a potential intervener is more likely to intervene on the same side if it has a stronger formal alliance relationship with the state that intervenes first, all other things being equal. Hypothesis 11: When the first intervention is made by a major power country, a potential intervener is less likely to intervene on the opposite side, all other things being equal. This variable refers to PA|B, the probability of a win by the coalition of State A and Group C given State B’s counter-intervention. When State A is a major power, PA|B becomes pretty high. Thus, State B’s counterintervention can hardly make any difference unless State B itself is another

134

Chapter Eight

major power. Knowing this, State B will refrain from intervening when the first intervention is made by a major power on the opposite side. As with Hypothesis 10, Hypothesis 11 does not conversely mean that a potential intervener is more likely to intervene on the same side when the first intervention is made by a major power, all other things being equal. Again, this is about bandwagoning intervention. When a major power intervenes in a civil war, another state is more likely to intervene on the same side because of bandwagoning. This scenario is, however, not always what happens on the ground. There is also a free-rider. Since a major power state already intervened into the civil war, there is nothing much another state can contribute more. A free-rider decides not to intervene and rather decides to enjoy the outcome which is believed to be favorable to it because of the first intervention is made by the major power. Thus, we can see that there is conflicting interests between bandwagoning intervention and free-rider non-intervention. Hypothesis 12: If the aim of the rebel in the civil war is to replace the government with its own, foreign states are more likely to execute a counter-intervention into the civil war, all other things being equal. This hypothesis is closely related to Hypothesis 3. This is a more specific version of Hypothesis 3 in that after the first intervention, it is not clear whether bandwagoning intervention or free-rider behavior will occur. Thus, when the first intervention is made on the side of the government, it is impossible to predict whether the second intervention will also be made on the side of the government even if the aim of the rebel force in the civil war is to overthrow the government. However, when it comes to counter-intervention, the expected direction is clear when the aim of the rebel group in the civil war is to replace the government. There is one point to note. As in Analysis I, Analysis II also cannot fully reflect all the components of the model and the framework developed in Chapter 4 and 5, mainly because of data limitation. Thus, the empirical testing models in this book are limited to fully capture the dynamics of foreign intervention.

8.2 Research Design The cases to be used in this analysis are the cases in which the first intervention was already made. Thus, the civil war cases that did not include any external intervention have been dropped from the data set and

Analysis II: Second Intervention

135

excluded from the analysis here. This procedure left 2,377 observations for cases with the first intervention on the side of the government and 1,706 observations for cases with the first intervention on the side of the opposition. Dependent Variable Second Intervention The dependent variable in this analysis is whether an intervention occurs on the side of the government or on the side of the opposition in the annual dyads when the intervention by another state has already been made on the side of the government (Analysis 1) or the opposition (Analysis 2). Thus, if there exists an intervention for the government in a dyad, it is coded 1. If intervention is made for the opposition, it is coded 2. If no intervention is recorded, then it is coded 0. All other characteristics are the same as with the First Intervention variable in the previous chapter. Independent Variable Two independent variables are added to investigate the relationships between interveners according to Hypotheses 10 and 11. First Int_Alliance This is the alliance relationship between the first intervener and the potential intervening states. As with the Alliance variable, this variable is also generated by EUGene v 3.2. This dyadic alliance relationship in EUGene uses the COW v3.0 alliance data released in 2002. The codes for various types of alliances are 1 = defense pact, 2 = neutrality, 3 = enténte, and 4 = no agreement. First Int_Major This variable indicates the major power status of the first intervener. If it is a major power (one of the Security Council member countries), it is coded 1, otherwise 0. Besides these two variables, other variables used in the analysis of the first intervention are also used in the analysis of the second intervention.

136

Chapter Eight

8.3 Analysis 1: With the First Intervention on the Side of the Government Overall, the results show fairly good fitness for the model. Pseudo R2 is 0.3447. Other statistics also show that this model has explanatory power. Table 8-1. Multinomial Logit When the First Intervention for the Government Cases Number of Obs = LR Chi2 (38) = Prob > Chi2 = Pseudo R2 =

GOVERNMENT

First Int_Alliance First Int-Major Alliance Cap_1 Cap_2 Cap Ratio Major Power_2 Distance Islands War Type Aim Sxp

Coefficient (Standard Error) -0.801558** (0.2385937) -0.9846829** (0.4593942) 0.9483133** (0.4178951) - 694.6643** (116.6525) 43.39605** (6.774963) -0.0028133 (0.0019098) -4.918687** (1.108939) - 0.0005389** (0.0001986) -32.6763 (1.25e+07) -5.224099** (1.017542) -0.8041158* (0.4404314) 9.682129** (1.956098)

2377 644.33 0.000 0.3447 Relative Risk Ratio (Standard Error) 0.4486295** (0.1070402) 0.3735577** (0.1716102) 2.581352** (1.078734) 2.0e-302** (2.4e-300) 7.03e+18** (4.76e+19) 0.9971907 (0.0019045) 0.0073087** (0.0081049) 0.9994612** (0.0001985) 6.44e-15 (8.04e-08) 0.0053852** (0.0054797) 0.4474834* (0.1970858) 16028.59** (0.0054797)

Analysis II: Second Intervention

Cold War Leader’s Ethnic_1 Leader’s Ethnic_2 ELF_1 ELF_2 Dem_1 Dem_2 Constant OPPOSITION First Int_Alliance First Int-Major Alliance Cap_1 Cap_2 Cap Ratio Major Power_2 Distance Islands War Type Aim Sxp Cold War

1.14503* (0.6930756) 3.120103** (1.291238) 1.354871* (0.8140687) -2.291112** (0.9477524) -0.3261674 (0.8575446) 0.1071996** (0.0515305) 0.0115964 (0.0236617) -2.909246** (1.041826) Coefficient (Standard Error) 0.5398865** (0.1901596) 0.055726 (0.2753364) 0.3147393 (0.221783) -818.8734** (95.16092) 31.09216** (3.549699) -0.0046969** (0.0011358) -2.486426** (0.5005549) -0.0001467** (0.000576) 0.9615959 (0.628972) -1.741463** (0.2966034) -0.5714391** (0.189165) 2.006599** (0.8131659) 1.122257** (0.4267984)

137 3.142535* (2.178014) 22.64871** (29.24488) 3.876262* (3.155544) 0.1011539** (0.0958689) 0.7216844 (0.6188765) 1.113156** (0.0573615) 1.113156 (0.0573615) N/A Relative Risk Ratio (Standard Error) 1.715812** (0.3262782) 1.057308 (0.2911153) 1.369902 (0.303821) 0** (0) 3.19e+13** (1.13e+14) 0.9953142** (0.011305) 0.0832068** (0.0416496) 0.9998533** (0.0000576) 2.615868 (1.645308) 0.1752638** (0.0519838) 0.5590932** (0.1057608) 7.43798** (6.048312) 3.071778** (1.31103)

138

Chapter Eight

1.270144** (0.47957) 1.350059* Leader’s Ethnic_2 (0.6923378) 2.44517** ELF_1 (0.6767563) -1.606081** ELF_2 (0.7101798) -0.0212054 Dem_1 (0.0230388) 0.0305561** Dem_2 (0.0148411) -6.181709** Constant (1.003938) First Intervention = 0 is the base outcome ** significant at 95 %, * significant at 90 %.

Leader’s Ethnic_1

3.561364** (1.707923) 3.857653* (2.670799) 11.53251** (7.804697) 0.2006725** (0.1425135) 0.9790178 (0.0225554) 1.031028** (0.0153016) N/A

Since the first interventions have already been made on the side of the government, a second intervention on the side of the government is considered bandwagoning intervention and a second intervention on the side of the opposition is considered counter-intervention. As seen in Table 8-1, Hypothesis 10 is well-supported by the empirical analysis. The results show that when the first intervention has been made on the side of the government, a one-unit change in First Int_Alliance (moving toward a weaker alliance relationship) makes the odds of intervention on the side of the opposition (counter-intervention) increase by a factor of 1.715812 with all other variables held constant. That is, for a one-unit change in First Int_Alliance, the odds of intervention on the side of the opposition are 71.58 percent greater with all other variables held constant. This result is statistically significant at a 95 percent confidence level. As hypothesized, the results show that a potential intervener is more likely to intervene on the opposite side if it has a weak or no formal alliance with the state that intervenes first, all other things being equal. The results also show that when the first intervention has been made on the side of the government, a one-unit change in First Int_Alliance (moving toward a weaker alliance relationship) makes the odds of intervention on the side of the government (bandwagoning intervention) decrease by a factor of 0.4486295 with all other variables remaining constant. That is, for a one-unit change in First Int_Alliance, the odds of intervention on the same side decrease about 55.14 percent with all other variables held constant. This result is also statistically significant at a 95 percent confidence level. This result suggests that there also exists a pattern of bandwagoning intervention with respect to alliance relations

Analysis II: Second Intervention

139

between interveners. When a state intervenes in the civil war on the side of the government, its allies tend to intervene on the same side. The concept of the deterring counter-intervention effect of major power intervention is not supported. When a major power has executed the first intervention on the side of the government, this has no effect in terms of deterring counter-intervention. Hypothesis 11 is not supported by this analysis. However, one interesting finding is that intervention by a major power on the side of the government actually causes free-riders. As shown in Table 8-1, when a major power intervenes on the sides of the government, the odds of intervention on the side of the government decrease by about 62.64 percent with all other variables held constant. That is, states are less likely to intervene on the side of the government when a major power has already intervened on the side of the government. With regard to Hypothesis 12, if the aim of war is more focused on local issues, states are less likely to execute a counter-intervention. For a one-unit change toward an emphasis on local issues in terms of the aim of the civil war, the odds of intervention on the side of the opposition (counter-intervention) decrease about 44.1 percent with all other variables held constant. This result is statistically significant at a 95 percent level. In other words, if the aim of the civil war is more to overthrow the government, states show a strong tendency toward counter-intervention. Thus, Hypothesis 12 is strongly supported. For a one-unit change toward a focus on local issues in terms of the aim of the civil war, the odds of intervention on the side of the government decrease by about 55.25 percent with all other variables held constant. In other words, if the aim of the civil war is more to overthrow the government, states show a tendency toward both counter-intervention and bandwagoning intervention. This also confirms Hypothesis 3. In the analysis of first intervention, Hypothesis 3 is only partially supported. The results of the first analysis show that aiming at the center increases the likelihood of intervention on the side of the government, but it does not have a statistically significant effect on the likelihood of intervention on the side of the opposition. The story becomes different with the first intervention on the side of the government. Aiming at the center in a civil war actually increases the likelihood of both counterintervention and bandwagoning intervention. Hypothesis 4 (The more the dependence of the civil war state on primary commodity exports, the more likely there will be external interventions in the civil war, all other things being equal.) is strongly supported. When the dependence of the civil war state on primary

140

Chapter Eight

commodity exports is high, the likelihood of both bandwagoning intervention and counter-intervention becomes very high. Other capability-related variables, including Cap_1, Cap_2, Cap Ratio, and Distance, are generally statistically significant, and the direction of the effect is as expected. With regard to Hypothesis 5, the Cold War variable becomes significantly important in this analysis. The Cold War variable has a positive impact on both intervention on behalf of the government and intervention on behalf of the opposition. The Cold War variable was not statistically significant in the analysis of the first intervention. When the first intervention has been made on the side of the government, a conflict’s occurring during the Cold War period increases the odds of intervention on the side of the government by about 3.14 times (statistically significant at 90 percent) and by about 3.07 times (statistically significant at 95 percent) for intervention on the side of the opposition, with all other variables remaining constant. Hypothesis 6 is also supported. As the proportion of the population representing the leader’s ethnic group increases, the odds of both bandwagoning intervention and counter-intervention increase. The results regarding the War Type variable show that states are more likely to execute bandwagoning and counter-intervention in identity civil wars. For the War Type variable, a change from identity civil wars to identity civil wars in reference to the first intervention on the side of the government decreased the odds by about 0.5 percent; in reference to intervention on the side of the opposition, the decrease was one of about 17.5 percent. The magnitude of the effect can be understood as more strongly deterring counter-intervention in identity civil wars. However, there are results for some variables that are contrary to expectations and/or to the findings of the first analysis. The results for the alliance variable are again contrary to our expectations. Hypothesis 7 (A potential intervener is more likely to intervene on the side of the government if it has a formal alliance relationship with the civil war government, all other things being equal) and Hypothesis 7-1 (It is less likely for a potential intervener to intervene on the side of the opponent if it has a formal alliance with the civil war government, all other things being equal) are again not supported. The results actually show that a weaker or no alliance between the civil war state and the potential interveners increases the likelihood of intervention on the side of the government, which is totally the opposite of what is suggested in Hypothesis 6.

Analysis II: Second Intervention

141

With regard to the proportion of the population representing the governmental leader’s ethnic group in the civil war state, the results show that Hypothesis 8 (As the proportion of the population representing the governmental leader’s ethnic group in the civil war state increases, the potential intervener is more likely to intervene on the side of the government, all other things being equal) is supported. However, results show findings that are the opposite of what one would expect based on Hypothesis 8-1 (As the proportion of the population representing the governmental leader’s ethnic group in the civil war state decreases, a potential intervener is more likely to intervene on the side of the opponent, all other things being equal). Both hypotheses were supported in the analysis of the first intervention. Hypothesis 9 (The more democratic a potential intervener is, the less likely it is to intervene on the side of the opponent in the civil war, all other things being equal) is not supported in this analysis. The findings are the opposite of what is expected. Hypothesis 9 is well-supported in the first analysis. In the analysis of first interventions, more democratically developed countries are proven to be less likely to intervene on the side of the opposition. They are more likely to intervene to support the government. However, in this analysis, democratically developed countries are actually more likely to intervene on the side of the opposition, and this result is statistically significant at a 95 percent level. Even though the magnitude of the effect is not great (about a 3 percent increase in the odds of intervention on the side of the opposition), the pattern is clearly significant. The effect of intervention on the side of the government is not statistically significant. It is difficult to interpret these surprising findings. Before any interventions have been made in a civil war, as in the first analysis, Hypothesis 8-1 and Hypothesis 9 are well supported. In this analysis, the direction of the effect is totally opposite that of those hypotheses. One plausible interpretation of these findings with respect to Hypothesis 8-1 and Hypothesis 9 is that the first intervention indeed altered the intervention decision calculus of the potential interveners. Variables related to the index of ethno-linguistic fractionalization are generally statistically significant. However, the direction of their effects is the opposite of that of the findings of the first analysis. Specifically, with respect to ELF_1 (the index of ethno-linguistic fractionalization of the civil war state), the effects are the total opposite of the findings of the first analysis. In the first analysis, more fractionalized society attracts more interventions on the side of the government (statistically significant at a 95 percent level) and fewer interventions on the side of the opposition

142

Chapter Eight

(statistically significant at a 90 percent level). In the analysis with the first intervention occurring on the side of the government, a more fractionalized society actually attracts fewer interventions on the side of the government and more interventions on the side of the opposition with all other variables constant. Both results are statistically significant at a 90 percent level. To interpret these findings, one plausible scenario can be described. Given that potential interveners are more likely to intervene on the side of the government and are less likely to intervene on the side of the opponent in the first analysis with respect to ELF_1, more opportunities for counterintervention are created in a society with higher ELF. Thus, when there is an intervention on the side of the government in a civil war state with higher ELF, this situation leads to a lower likelihood of bandwagoning intervention and a higher likelihood of counter-intervention. With respect to Hypothesis 9, this analysis shows that more democratically developed countries are more likely to intervene on the side of the opposition. For a one-point increase in the Polity score, the odds of intervention on the side of the opposition are 1.031028 times greater. This is totally opposite the findings from the first analysis. These contradictory findings with regard to Hypothesis 8-1 and Hypothesis 9 indicate that the intervention decision calculus can be different based on the relationships between the interveners.

8.4 Analysis 2: With the First Intervention on the Side of the Opposition Since the first interventions have already been made on the side of the opposition, a second intervention on the side of the government is considered counter-intervention and a second intervention on the side of the opposition is considered bandwagoning intervention. As Table 8-2 shows, the goodness of fit is a little lower than for the model with the first intervention on the side of the government. Pseudo R2 is 0.1941. With regard to the variables of interest in this analysis, the likelihood of counter-intervention is statistically significant at a 90 percent level and the direction is as expected. As seen in Table 8-1, Hypothesis 10 is wellsupported in this analysis. The results show that when the first intervention was made on the side of the opposition, a one-unit change in the variable First Int_Alliance (moving toward a weaker alliance relationship) made the odds of intervention on the side of the government (counterintervention) increase by a factor of 1.232068 with all other variables

Analysis II: Second Intervention

143

constant. That is, for a one-unit change in the variable First Int_Alliance, the odds of counter-intervention are about 23.21 percent greater. As hypothesized, the results confirm that a potential intervener is more likely to make a counter-intervention if it has a weak or no formal alliance with the state that intervenes first, all other things being equal. Table 8-2. Multinomial Logit When the First Intervention for the Opposition Cases Number of Obs = LR Chi2 (38) = Prob > Chi2 = Pseudo R2 = GOVERNMENT First Int_Alliance First Int-Major Alliance Cap_1 Cap_2 Cap Ratio Major Power_2 Distance Islands War Type Aim Sxp Cold War Leader’s Ethnic_1 Leader’s Ethnic_2

Coefficient (Standard Error) 0.2086942* (0.1152812) -1.294877** (0.5665139) -0.2938372** (0.1115774) 36.47406 (64.23676) 21.84032** (3.135071) -0.003329** (0.0012473) -0.8770124** (0.3968861) - 0.0002965** (0.0000765) 5.316162** (0.7892315) -1.249899** (0.3147962) -0.5315402** (0.2241807) -3.29443** (1.418809) 0.210322 (0.2603413) -6.407043** (1.180091) 2.539209**

1706 218.00 0.000 0.1941 Relative Risk Ratio (Standard Error) 1.232068* (0.1420343) 0.2739316** (0.1551861) 0.7453979** (0.0831695) 6.93e+15 (4.45e+17) 3.06e+09** (9.58e+09) 0.9966766 (0.0012432) 0.416024** (0.1651141) 0.9997036** (0.0000765) 203.601 (160.6884) 0.2865338** (0.0901997) 0.5876991* (0.1317508) 0.0370892** (0.0526225) 1.234075 (0.3212808) 0.0016499** (0.001947) 12.66964**

144

ELF_1 ELF_2 Dem_1 Dem_2 Constant OPPOSITION First Int_Alliance First Int-Major Alliance Cap_1 Cap_2 Cap Ratio Major Power_2 Distance Islands War Type Aim Sxp Cold War Leader’s Ethnic_1 Leader’s Ethnic_2

Chapter Eight (0.6328931) -0.3138738 (0.4756026) 1.072433* (0.6508253) 0.005509 (0.0136827) 0.0794179** (0.0156343) -0.5556807** (1.010421) Coefficient (Standard Error) -0.0854717 (1.134693) 6.310938 (9.82591) 0.5946325 (0.800373) -499.8528 (830.5013) -623.3498 (679.6451) 0.0804176 (0.4514594) -111.1525 (1.49e+08) 0.0083172 (28508.9) -7.915454 (4.87e+07) -8.607522 (8.904745) 8.639293 (8.624136) -6.775263 (8.488174) -5.772691 (7.59126) -9.771701 (14.11543) -1.025022 (1.324082)

(8.018529) 0.7306112 (0.3474806) 2.922482* (1.902025) 1.005566 (0.0137589) 1.082657** (0.0169266) N/A Relative Risk Ratio (Standard Error) 0.9180791 (1.041738) 550.561 (5409.763) 1.812365 (1.450568) 8.3e-218 (6.9e-215) 1.9e-271 (1.3e-268) 1.083739 (0.4892644) 5.33e-49 (7.97e-41) 1.008352 (28747) 0.0003651 (17772.3) 0.0001827 (0.0016271) 5649.337 (48720.65) 0.0011417 (0.0096907) 0.0031114 (0.0236192) 0.000057 (0.0008052) 0.3587885 (0.4750652)

Analysis II: Second Intervention 15.61616 (25.8878) 3.695348** ELF_2 (1.863065) -0.0080304 Dem_1 (0.0732288) 0.1179826* Dem_2 (0.064971) -21.87532 Constant (22.34739) First Intervention = 0 is the base outcome ** significant at 95 %, * significant at 90 %. ELF_1

145 6053585 (1.57e+08) 40.2596** (75.00625) 0.9920018 (0.0726431) 1.125225* (0.0726431) N/A

The results show that the effect of First Int_Alliance on bandwagoning intervention is not statistically significant. With regard to Hypothesis 11, the concept of a deterring counterintervention effect of major power intervention is also supported. When a major power has made the first intervention on the sides of the opposition, this actually deters counter-intervention. Intervention by major powers on the side of the opposition decreases the odds of intervention on the side of the government by a factor of 0.2739316. That is, the odds of counterintervention decrease by 72.6 percent with all other variables constant. This result is statistically significant at a 95 percent level. Thus, Hypothesis 11 is well-supported in this analysis. However, intervention by major powers on the side of the opposition does not have a statistically significant effect on the likelihood of a second intervention on the side of the opposition. With regard to Hypothesis 12, if the aim of war is more about local issues, states are less likely to execute a counter-intervention. For a oneunit change toward a focus on local issues as the aim of the civil war, the odds of intervention on the side of the government (counter-intervention) decrease by about 41.23 percent with all other variables constant. This result is statistically significant at a 95 percent level. In other words, if the aim of civil war is more to overthrow the government, the states show a strong tendency toward counter-intervention. Thus, Hypothesis 12 is strongly supported. Hypothesis 5 is not supported. The Cold War variable does not have a statistically significant effect in this analysis. Hypothesis 6 is partially supported. As the proportion of the population representing the leader’s ethnic group increases, the odds of intervention on the side of the government (counter-intervention) increase. This is

146

Chapter Eight

statistically significant at a 95 percent level. However, its effect on the intervention on the side of the opposition is not statistically significant. Variables related to capability, including Cap_2, Cap Ratio, Distance, and Islands, are all in general statistically significant at 95 percent, and the directions of their effects are as expected. However, there are results for some variables that are contrary to expectations and/or to the findings of the first analysis. The results with reference to the alliance variable are contrary to expectations. Hypothesis 7 (A potential intervener is more likely to intervene on the sides of the government if it has formal alliance relationship with the civil war government, if other things being equal) and Hypothesis 7-1 (It is less likely for a potential intervener to intervene on the sides of the opponent if it has formal alliance relationship with the civil war government, if other things being equal) are again not supported. The results actually show that a weaker or no alliance between the civil war state and the potential interveners increases the likelihood of intervention on the side of the government, which is completely the opposite of what is suggested by Hypothesis 7. With regard to the proportion of the population representing the leader’s ethnic group in the civil war state, the results show that Hypothesis 8 (As the population proportion of governmental leader’s ethnic group in the civil war state increases, a potential intervener is more likely to intervene on the sides of the government, all other things being equal) is not supported. However, the results show findings that are the opposite of what is suggested by Hypothesis 8. Hypothesis 9 (As a potential intervener is more democratic, it is less likely to intervene on the sides of the opponent in the civil war, if other things being equal) is not supported in this analysis. The findings are contrary to the expectations put forth. In this analysis, democratically developed countries are actually more likely to intervene on the side of the opposition, and this result is statistically significant at a 90 percent level.

8.5 Summary From the analyses in this chapter, it is clear that the relationships between interveners are important determinants of second intervention. Hypotheses regarding the relationships between interveners are in general strongly supported by the data. I found strong support for Hypothesis 10 (A potential intervener is more likely to intervene on the opposite side if it has a weak or no formal alliance with the state that intervenes first, all other things being equal). In

Analysis II: Second Intervention

147

both Analysis 1 and Analysis 2, there exist strong tendencies toward counter-intervention. With regard to Hypothesis 11 (When the first intervention is made by a major power, a potential intervener is less likely to intervene on the opposite side, all other things being equal), I found a strong deterring counter-intervention effect based on Analysis 2. Analysis 1 does not support Hypothesis 11. Hypothesis 12 (If the aim of the rebel group in the civil war is to replace the government with its own, foreign states are more likely execute a counter-intervention in the civil war, all other things being equal) is strongly supported. The effect of the Aim variable becomes stronger when it comes to counter-intervention. These findings strongly suggest that the relationships between interveners should be included in the study of foreign military intervention in civil wars.

CHAPTER NINE CONCLUSION

The purpose of this book is to identify the conditions under which foreign countries intervene in civil wars. I contend that we should consider four dimensions of civil war intervention. The first dimension is the civil war itself. The characteristics of civil war itself are important determinants of a third party’s decision making regarding intervention. The second dimension is the characteristics of intervening states. Their capabilities and domestic political environments are included in this dimension. The third dimension is the relationship between the host country and the intervening country. These states’ formal alliances and the differences in military capability between the target country and the potential intervener have an impact on the decision making process. The fourth dimension is the relationship between the interveners. This framework of four dimensions proves critical in understanding foreign intervention in civil wars. Based on this framework, the model for the intervention mechanism can reflect reality better. By including the relationships between the interveners, I show that it is important to distinguish between intervention on the side of the government and intervention on behalf of the opposition. Without distinguishing between intervention on the side of the government and intervention on behalf of the opposition, it is impossible to consider the concepts of counterintervention and bandwagoning intervention. Counter-intervention and bandwagoning intervention cannot be conceptualized unless the sequential dynamics of intervention are theorized. Since previous studies have not paid attention to the relationships between interveners, they also have not distinguished the target of intervention and the sequence of intervention. Consequently, their models cannot conceptualize counter-intervention or the bandwagoning intervention. I believe that this book contributes to our understanding of foreign intervention in this sense. Using the framework of four dimensions, more specific hypotheses can be formed. In this book, 12 hypotheses are formed. The effects of the variables are not static. The magnitude and the direction of the effects vary according to the different models in this study. This also reflects the fact

150

Chapter Nine

that the intervention decision calculus changes as a reaction to earlier intervention by other countries in the same civil war. I found that relationships between interveners are a critical aspect of foreign intervention and should be studied further. These findings suggest that the model and the framework used in this book better reflect the reality of foreign intervention in civil wars than did previous studies on this subject. Nevertheless, this study cannot be considered to complete the study of foreign intervention in civil wars. There are several ways that this study can be improved upon. First, the lack of availability of certain data limits my effort to reflect on and test the formal model of intervention. In particular, there are no available data on the capabilities of the warring factions in civil wars. Even more so, no data are available on how those capabilities change over time during civil wars. As conceptualized in my model in Chapter 4, parity in power or the power balance is crucial in determining the likelihood of foreign intervention. Second, including other methods of intervention is believed to be important. There are many ways that foreign countries affect the outcome of the civil war. Overt military intervention is only the tip of the iceberg. Considering the scale of civil war, covert military intervention through channels such as equipment aid, financial aid, and the provision of mercenaries and military training can have a huge impact on the outcome of a civil war. Accordingly, this will also affect other aspects of civil war politics, so including these covert military interventions in future research will significantly expand our understanding of this subject. Again, the problem is that there is very scant, if any, information on those covert military interventions.

REFERENCES

Africa Union Constitutive Act. Accessed December 8, 2008. http://www.africa-union.org/root/au/AboutAu/Constitutive_Act_en.htm. Atack, Iain. 2002. “Ethical Objections to Humanitarian Intervention.” Security Dialogue 33 (3): 279–92. Balch-Lindsay, Dylan, and Andrew J. Enterline. 2000. “Killing Time: The World Politics of Civil War Duration, 1820–1992.” International Studies Quarterly 44 (4): 615–42. Bennett, D. Scott, and Allan Stam. 2000. “EUGene: A Conceptual Manual.” International Interactions 26 (2): 179–204. Boulding, Kenneth E. 1962. Conflict and Defense: A General Theory. New York, NY: Harper Torchbook. Brecher, Michael, and Jonathan Wilkenfeld. 1997. “The Ethnic Dimension of International Crisis.” In Wars in the Midst of Peace: The International Politics of Ethnic Conflict, edited by David Carment and Patrick James. Pittsburgh, PA: University of Pittsburgh Press. Bueno de Mesquita, Bruce. 1981. The War Trap. New Haven, CT: Yale University Press. Bull, Hedley. 1984. Intervention in World Politics. New York, NY: Oxford University Press. Burg, Steven L. 2004. “Intervention in Internal Conflict: The Case of Bosnia.” In Military Intervention: Cases in Context for the Twenty-first Century, edited by William J. Lahneman, 47–66. Lanham, MD: Rowman & Littlefield Publishers. Carment, David, and Dane Rowlands. 1998. “Three’s Company: Evaluating Third-Party Intervention in Intrastate Conflict.” Journal of Conflict Resolution 42 (5): 572–99. Carment, David, and Patrick James. 1995. “Internal Constraints and Interstate Ethnic Conflict: Toward a Crisis-based Assessment of Irredentism.” Journal of Conflict Resolution 39 (1): 82–109. —. 1997. “Secession and Irredenta in World Politics: The Neglected Interstate Dimension.” In Wars in the Midst of Peace: The International Politics of Ethnic Conflict, edited by David Carment and Patrick James. Pittsburgh, PA: University of Pittsburgh Press. —. 2003. “Third-Party States in Ethnic Conflict: Identifying the Domestic Determinants of Intervention.” In Ethnic Conflict and International

152

References

Politics: Explaining Diffusion and Escalation, edited by Steven E. Lobell and Philip Mauceri, 11–34. New York, NY: Palgrave. Carment, David, Patrick James, and Zeynep Taydas. 2006. Who Intervenes? Ethnic Conflict and Interstate Crisis. Columbus, OH: The Ohio State University Press. Cederman, Lars-Erik, and Luc Girardin. 2007. “Beyond Fractionalization: Mapping Ethnicity onto Nationalist Insurgencies.” American Political Science Review 101 (1): 173–85. Cetinyan, Rupen. 2002. “Ethnic Bargaining in the Shadow of Third-Party Intervention.” International Organization 56 (3): 645–77. Chesterman, Simon. 2002. “Legality versus Legitimacy: Humanitarian Intervention, the Security Council, and the Rule of Law.” Security Dialogue 33 (3): 293–307. Cillers, Jakkie, and Kathryin Sturman. 2002. “The Right Intervention: Enforcement Challenges for the African Union.” African Security Review 11 (3). Accessed July 24, 2008. http://www.iss.co.za/pubs/ASR/11No3/Cilliers.html. Clark, David H., and Patrick M. Regan. 2003. “Opportunities to Fight: A Statistical Technique for Modeling Unobservable Phenomena.” Journal of Conflict Resolution 47 (1): 94–115. Collier, Paul, and Anke Hoeffler. 1998. “On Economic Causes of Civil War.” Oxford Economic Paper 50 (4): 563–73. —. 2000. “Greed and Grievance in Civil War.” World Bank Working Paper Series 2000-18. April 26, Washington, DC. —. 2005. “Resource Rents, Governance, and Conflict.” Journal of Conflict Resolution 49 (4): 625–33. Croissant, Michael P. 1998. The Armenia-Azerbaijan Conflict: Causes and Implication. Prager. Cronin, Bruce. 2007. “The Tension between Sovereignty and Intervention in the Prevention of Genocide.” Human Rights Review: 293–305. Dallaire, Romeo. 2003. Shake Hands with the Devil: The Failure of Humanity in Rwanda. New York, NY: Carroll & Graf Publishers. Damrosch, Lori Fisler. 1993. “Changing Conceptions of Intervention in International Law.” In Emerging Norms of Justified Intervention, edited by Laura W. Reed and Carl Kaysen, 91–114. Cambridge: American Academy of Arts and Sciences. —, ed. 1993. Enforcing Restraint: Collective Intervention in Internal Conflicts. New York, NY: Council on Foreign Relations Press. David, Steven R. 1997. “Internal War: Causes and Cures.” World Politics 49 (4): 552–76.

Foreign Intervention in Civil Wars

153

Davis, David R., Keith Jaggers, and Will H. Moore. 1997. “Ethnicity, Minorities, and International Conflict Patterns.” In Wars in the Midst of Peace: The International Politics of Ethnic Conflict, edited by David Carment and Patrick James. Pittsburgh, PA: University of Pittsburgh Press. Davis, David R., and Will H. Moore. 1997. “Ethnicity Matters: Transnational Ethnic Alliances and Foreign Policy Behavior.” International Studies Quarterly 41 (1): 171–84. DeRouen, Karl Jr., and David Sobek. 2004. “The Dynamics of Civil War Duration and Outcome.” Journal of Peace Research 41 (3): 303–20. Dunér, Bertil. 1983. “The Many-Pronged Spear: External Military Intervention in Civil Wars in the 1970s.” Journal of Peace Research 20 (1): 59–72. —. 1985. Military Intervention in Civil Wars: The 1970s. Hampshire, England: Gower Publishing. Elbadawi, Ibrahim. 2000. “External Interventions and the Duration of Civil Wars.” World Bank Policy Research Working Paper 2433. http://ssrn.com/abstract=632504. Elbadawi, Ibrahim, and Nicholas Sambanis. 2002. “How Much War Will We See?: Explaining the Prevalence of Civil War.” Journal of Conflict Resolution 46 (3): 307–34. Esty, Daniel C., Jack Goldstone, Ted Robert Gurr, Barbara Harff, Pamela T. Surko, Alan N. Unger, and Robert Chen. 1997. “The State Failure Project: Early Warning Research for US Foreign Policy Planning.” In Preventive Measure: Building Risk Assessment and Crisis Early Warning Systems, edited by John L. Davies and Ted Robert Gurr. Boulder, CO: Rowman & Littlefield. Farer, Tom. 1993. “A Paradigm of Legitimate Intervention.” In Enforcing Restraint: Collective Intervention in Internal Conflicts, edited by Lori Fisher Damrosch, 316–47. New York, NY: Council on Foreign Relations Press. Fearon, James D. 1994. “Signaling versus the Balance of Power and Interests: An Empirical Test of a Crisis Bargaining Model.” Journal of Conflict Resolution 38 (2): 236–69. —. 2004. “Why Do Some Civil Wars Last So Much Longer Than Others?” Journal of Peace Research 41 (3): 275–302. —. 2005. “Primary Commodity Exports and Civil War.” Journal of Conflict Resolution 49 (4): 483–507. —. 2007. “Iraq’s Civil War.” Foreign Affairs 86 (2). Accessed September 15, 2008. http://www.foreignaffairs.org/20070301faessay86201/jamesd-fearon/iraq-s-civil-war.html.

154

References

Fearon, James D., and David D. Laitin. 1996. “Explaining Interethnic Cooperation.” American Political Science Review 90 (4): 715–35. —. 2003. “Ethnicity, Insurgency, and Civil War.” American Political Science Review 97 (1): 75–90. Fearon, James D., Kimuli Kasara, and David D. Laitin. 2007. “Ethnic Minority Rule and Civil War Onset.” American Political Science Review 101 (1): 187–93. Feste, Karen A. 1992. Expanding the Frontiers: Superpower Intervention in the Cold War. New York, NY: Praeger. —. 2003. Intervention: Shaping the Global Order. New York, NY: Praeger. Finnemore, Martha. 2003. The Purpose of Intervention. Ithaca: Cornell University Press. Fordham, Benjamin O. 2004. “A Very Sharp Sword: The Influence of Military Capabilities on American Decisions to Use Force.” Journal of Conflict Resolution 48 (5): 632–56. Gartner, Scott Sigmund, and Randolph M. Siverson. 1996. “War Expansion and War Outcome.” Journal of Conflict Resolution 40 (1): 4–15. Gleditsch, Kristian Skrede, and Kyle Beardsley. 2004. “Nosy Neighbors: Third-Party Actors in Central American Conflicts.” Journal of Conflict Resolution 48 (3): 379–402. Gleditsch, Nils Peter, Peter Wallenstein, Mikael Eriksson, Margareta Sollenberg, and Harvard Strand. 2002. “Armed Conflict 1946–2001: A New Dataset.” Journal of Peace Research 39 (5): 615–37. Global Witness. 2002. “The Logs of War: The Timber Trade and Armed Conflict.” Programme for International Co-operation and Conflict Resolution, FAFO Report 379. Accessed September 23, 2008. http://www.fafo.no/pub/rapp/379/379.pdf. Gow, James. 1994. “Nervous Bunnies: The International Community and the Yugoslav War of Dissolution, the Politics of Military Intervention in a Time of Change.” In Military Intervention in European Conflicts, edited by Lawrence Freedman, 14–33. Cambridge, MA: Blackwell Publishers. Heraclides, Alexis. 2001. “Secessionist Minorities and External Involvement.” International Organization 44 (3): 341–78. Hoffmann, Stanley. 1995. “The Politics and Ethics of Military Intervention.” Survival 37 (4): 29–51. Huth, Paul K. 1988. “Extended Deterrence and the Outbreak of War.” American Political Science Review 82 (2): 423–43.

Foreign Intervention in Civil Wars

155

Kaufmann, Chaim. 1996. “Intervention in Ethnic and Ideological Civil Wars: Why One Can Be Done and The Other Can’t.” Security Studies, 6 (1): 62–100. Khadiagala, Gilbert M. 2004. “Intervention in Internal Conflict: The Case of Rwanda.” In Military Intervention: Cases in Context for the Twentyfirst Century, edited by William J. Lahneman, 67–88. Lanham, MD: Rowman & Littlefield Publishers. Kim, Chae-Han. 1991. “Third Party Participation in Wars.” Journal of Conflict Resolution 35 (4): 659–77. Konstandaras, Nikos. 2006. “Turkey Must Change First.” Newsweek/ Washington Post (December 4). Accessed August 3, 2008. http://newsweek.washingtonpost.com/postglobal/nikos_konstandaras/2 006/ 12/ turkey_must_change_first.html. Krain, Matthew. 2005. “International Intervention and the Severity of Genocides and Politicides.” International Studies Quarterly 49: 363– 87. Krasner, Stephen D. 1999. Sovereignty: Organized Hypocrisy. Princeton: Princeton University Press. Lahneman, William J. 2004. “Introduction.” In Military Intervention: Cases in Context for the Twenty-first Century, edited by William J. Lahneman, xiii-xvii. Lanham, MD: Rowman & Littlefield. Lemke, Douglas. 1995. “The Tyranny of Distance: Redefining Relevant Dyads.” International Interactions 21 (1): 23–38. Lemke, Douglas, and William Reed. 2001. “The Relevance of Politically Relevant Dyads.” Journal of Conflict Resolution 45 (1): 126–44. Lemke, Douglas, and Patrick M. Regan. 2004. “Interventions as Influence.” In The Scourge of War: New Extensions on an Old Problem, edited by Paul F. Diehl, 145–68. Ann Arbor, MI: The University of Michigan Press. Little, Richard. 1975. Intervention: External Involvement in Civil Wars. Birmingham: Martin Robertson. Lobell, Steven E., and Philip Mauceri. 2004. “Diffusion and Escalation of Ethnic Conflict” In Ethnic Conflict and International Politics: Explaining Diffusion and Escalation, edited by Steven E. Lobell and Philip Mauceri. New York, NY: Palgrave Macmillan. Maoz, Zeev, and Bruce Russett. 1993. “Normative and Structural Causes of Democratic Peace, 1946–1986.” American Political Science Review 87 (3): 624–38. Mason, T. David., Joseph P. Weingarten, Jr., and Patrick J. Fett. 1999. “Win, Lose, or Draw: Predicting the Outcome of Civil Wars.” Political Research Quarterly 52 (2): 239–68.

156

References

Moore, W.H., and D.R. Davis. 1998. “Transnational Ethnic Ties and Foreign Policy.” In The International Spread of Ethnic Conflict, edited by David Lake and Donald Rothchild. Princeton, NJ: Princeton University Press. Morgan, Patrick M. 1977. Deterrence: A Conceptual Analysis. London: Sage Publications. Most, Benjamin A., and Harvey Starr. 1989. Inquiry, Logic and International Politics. Columbia: University of South Carolina Press. National Diet Library. The Constitution of Japan. Accessed November 2008. http://www.ndl.go.jp/constitution/e/etc/c01.html#s2. Otte, Thomas G. 1995. “On Intervention: Some Introductory Remarks.” In Military Intervention: From Gunboat Diplomacy to Humanitarian Intervention, edited by Andrew M. Dorman and Thomas G. Otte, 3–15. Aldershot: Dartmouth. Payne, James L. 1970. The American Threat: The Fear of War as an Instrument of Foreign Policy. Chicago: Markham. Pearson, Frederic S. 1974a. “Geographic Proximity and Foreign Military Intervention.” Journal of Conflict Resolution 18 (3): 432–60. —. 1974b. “Foreign Military Intervention and Domestic Disputes.” International Studies Quarterly 18 (3): 259–90. Pearson, Frederic S., Robert A. Baumann, and Jeffrey Pickering. 1994. “Military Intervention and Realpolitik.” In Reconstructing Realpolitik, edited by Frank W. Wayman and Paul F. Diehl, 205–25. Ann Arbor, MI: University of Michigan Press. Pereira, Anthony. 1994. “Neglected Tragedy: The Return to War in Angola, 1992–1993.” Journal of Modern Africa Studies 32 (1): 1–28. Powell, Robert. 1999. In the Shadow of Power. Princeton, NJ: Princeton University Press. Ramsbotham, Oliver, and Tom Woodhouse. 1996. Humanitarian Intervention in Contemporary Conflict. Cambridge, MA: Polity Press. Ray, James Lee. 1990. “Friends as Foes: International Conflict and Wars between Formal Allies.” In Prisoners of War?: Nation-States in the Modern Era, edited by Charles Gochman and Alan Sabrosky, 73–92. Lexington, MA: Lexington Books. Regan, Patrick M. 1996. “Conditions of Successful Third-Party Intervention in Intrastate Conflicts.” Journal of Conflict Resolution 40 (2): 336–59. —. 1998. “Choosing to Intervene: Outside Interventions in Internal Conflicts.” The Journal of Politics 60 (3): 754–79. —. 2000. Civil Wars and Foreign Powers: Outside Intervention in Intrastate Conflict. Ann Arbor: MI: The University of Michigan Press.

Foreign Intervention in Civil Wars

157

—. 2002. “Third Party Interventions and the Duration of Intrastate Conflicts.” Journal of Conflict Resolution 46 (1): 55–73. Reisman, Michael, and Myres S. McDougal. 1973. “Humanitarian Intervention to Protect the Ibos.” In Humanitarian Intervention and the United Nations, edited by Richard B. Lillich, 167–95. Charlottesville, VA: University Press of Virginia. Richardson, Lewis F. 1960. Statistics of Deadly Quarrels. Pacific Grove, CA: Boxwood. Ross, Michael. 2004. “What Do We Know About Natural Resources and Civil War?” Journal of Peace Research 41 (3): 337–56. —. 2005. “Booty Futures.” Manuscript. Accessed September 25, 2008. http://www.sscnet.ucla.edu/polisci/faculty/ross/bootyfutures.pdf. Russett, Bruce M. 1975. International Regions and the International System: A Study in Political Ecology. Santa Barbara, CA: Greenwood Press. Sambanis, Nicholas. 2001. “Do Ethnic and Nonethnic Civil Wars Have the Same Causes?: A Theoretical and Empirical Inquiry (Part 1).” Journal of Conflict Resolution 45 (3): 259–82. Simpson, Chris. 2002. “Obituary: Jonas Savimbi, UNITA’s Local Boy.” BBC News (February 25). Accessed December 27, 2007. http://news.bbc.co.uk/2/hi/africa/264094.stm. Singer, J. David. 1963. “Inter-Nation Influence: A Formal Model.” American Political Science Review 57: 420–30; reprinted in J. David Singer, ed. 1979. The Correlates of War I: Research Origins and Rationale, 48–67. New York: Free Press. Siverson, Randolph M., and Harvey Starr. 1990. “Opportunity, Willingness, and the Diffusion of War.” American Political Science Review 84 (1): 47–67. Siverson, Randolph M., and Joel King. 1980. “Attributes of National Alliance Membership and War Participation, 1815–1965.” American Journal of Political Science 24 (1): 1–15. Small, Melvin, and David Singer. 1976. “The War Proneness of Democratic Regimes.” Jerusalem Journal of International Relations 1 (1): 50–69. —. 1982. Resort to Arms: International and Civil Wars, 1816–1980. Thousand Oaks, CA: Sage Publications. Smith, Alastair. 1996. “To Intervene or Not to Intervene: A Biased Decision.” Journal of Conflict Resolution 40 (1): 16–40. Smith, Tony. 1994. “In Defense of Intervention.” Foreign Affairs 73 (6): 34–46.

158

References

Starr, Harvey, and Benjamin A. Most. 1976. “The Substance and Study of Borders in International Relations Research.” International Studies Quarterly 20 (4): 581–621. —. 1978. “A Return Journey: Richardson, ‘Frontiers’ and Wars in the 1946–1965 Era.” Journal of Conflict Resolution 22 (3): 441–67. Steinbruner, John, and Jason Forrester. 2004. “Perspectives on Civil Violence: A Review of Current Thinking.” In Military Intervention: Cases in Context for the Twenty-first Century, edited by William J. Lahneman, 1–27. Lanham, MD: Rowman & Littlefield. Stone, Julius. 2006. Aggression and World Order: A Critique of United Nations Theories of Aggression. Clark, NJ: The Lawbook Exchange. Talentino, Andrea Kathryn. 2005. Military Intervention after the Cold War. Athens, OH: Ohio University Press. Tillema, Herbert K. 1994. “Cold War Alliance and Overt Military Intervention, 1945–1991.” International Interactions 20 (3): 249–78. Trachtenberg, Marc. 1993. “Intervention in Historical Perspective.” In Emerging Norms of Justified Intervention, edited by Laura W. Reed and Carl Kaysen, 15–36. Cambridge, MA: Committee on International Security Studies. Wafadar, K. 1981. “Afghanistan in 1980: The Struggle Continues.” Asian Survey 21 (2): 172–80. Walzer, Michael. 1977. Just and Unjust Wars: A Moral Argument with Historical Illustrations. New York, NY: Basic Books. Welsh, Jennifer M., ed. 2004a. Humanitarian Intervention and International Relations. Oxford: Oxford University Press. —. 2004b. “Introduction.” In Humanitarian Intervention and International Relations, edited by Jennifer M. Welsh, 1–10. Oxford: Oxford University Press. Woodwell, Douglas. 2004. “Unwelcome Neighbors: Shared Ethnicity and International Conflict during the Cold War.” International Studies Quarterly 48 (1): 197–223.