This study moves beyond the question whether humanitarianism matters in third parties decisions about how to respond to civil wars, and examines for whom and when it matters. I argue that humanitarian motives, measured as the level of human rights violations in a conflict state, are more likely to play a role in deciding whether or not to intervene in a civil war when the third party is (1) not a major power, (2) a democracy, and (3) not faced with the power politics that characterize a bipolar international system like that of the Cold War era. To test this argument, I examine dyadic data on 174 civil wars between 1946 and 2008. In accordance with my theoretical expectations, I find that weaker states are more likely than major powers to intervene in conflicts that exhibit high levels of human rights violations. I also find support for propositions (2) and (3), but these findings are sensitive to alternative model specifications.
Keywords Humanitarianism, imperialism, human rights, intervention, civil war, conflict, violence
Aim and Scope
Broadly speaking, this study aims at examining the reasons why foreign countries (or third parties) militarily intervene in civil conflicts. To better illustrate its argument, I begin with a brief discussion of two contrasting examples: the Sierra Leone Civil War (1991-2002) and the Casamance Conflict in Senegal (1982-2014). Sierra Leone and Senegal are very similar on a number of dimensions. Both countries are located in West Africa, possess ample natural resources including diamonds and gold (but no oil), and are former colonies of Western powers. In addition, both Sierra Leone and Senegal have small economies without significant ties to the West, are militarily weak, and have little geopolitical importance. Therefore, if we only focus on such material factors, as Realists often do, it seems somewhat puzzling that the civil war in Sierra Leone triggered a military intervention by its former colonizer, Great Britain, whereas the one in Senegal did not (Fall 2010; Gberie 2005; Williams 2001).
If we extend our focus beyond material factors, one can see that the Sierra Leone Civil War differed from the Casamance Conflict in one important respect, which might help explain the difference in intervention outcomes: the level of violence against civilians. While all intrastate wars are tragic and involve substantial human suffering, there is substantial variation in the extent and nature of the atrocities committed by the warring parties. Some civil conflicts experience widespread and even systematic violence against civilians such as rape, torture, and targeted killings, whereas in others this kind of violence is relatively rare (Cohen 2013; Humphreys and Weinstein 2006; Weinstein 2006). The Sierra Leone Civil War is an example of the former type of conflict. More than 50,000 people died as a result of the war, the majority of them civilians. The years between 1997 and 2000 were marked by systematic atrocities committed against the civilian population, to the extent that some observers called it genocidal violence (Gberie 2005). In contrast, the Casamance Conflict was an intense but rather localized civil war. Fighting was largely restricted to the southern part of Senegal, and both warring parties showed relatively great restraint in their targeting of civilians (Clark 2011).
A third party’s desire to alleviate the suffering of and protect civilians abroad is referred to as “humanitarianism”, and military interventions that are primarily motivated by such concerns are called “humanitarian interventions” (Wheeler 2000). The British intervention in Sierra Leone in 2000 is said to be primarily the result of Britain’s concerns over the high level of violence against civilians during the conflict, and is therefore often viewed as an example of humanitarian intervention (Dorman 2009; Stewart 2008). However, the extent to which particular military interventions can be classified as humanitarian, and even the question whether genuine humanitarian interventions are at all possible, is hotly debated among publics, policymakers, and scholars alike (Connaughton 2002). Essentially, one side of the debate argues that states have a moral obligation to protect civilians in foreign countries from violence and abuse, and that this obligation is what generally drives military interventions. The other side asserts that the humanitarian justifications voiced by the governments of intervening states are nothing but lies that serve to appease the public and hide the true, selfish motives of the intervener.
The reason why the issue of humanitarian intervention is so hotly debated is that it has important implications for how we evaluate not only the suitability of intervention as a potential tool for conflict resolution but also the morality of the states that engage in it (Orford 2003). In part due to the controversial nature of the issue, previous scholarship on the role of humanitarianism in third parties’ decisions about how to respond to civil conflicts has reached an impasse. A vast number of studies argue that there are multiple examples of humanitarian intervention, whereas an equally large amount of research finds that third parties intervene in civil conflicts only when it is in their own strategic interest to do so. This study moves beyond the question whether humanitarianism matters in third parties' decisions about how to respond to civil wars and instead examines for whom and when it matters, thereby making three contributions to the literature. First, building on a rich tradition of research, I develop an expected utility theory of third-party intervention in civil wars. By conceptualizing humanitarianism as a distinct preference of interveners, as well as theorizing the conditions under which this preference is stronger or weaker, this study enriches our understanding of the relationship between humanitarianism and third-party intervention.
Second, by transforming the UCDP/PRIO Armed Conflict Dataset into a dyadic format, I provide scholars interested in the determinants of third-party intervention with a dataset that does not only cover more years than those used by other recent analyses of the causes of military intervention (e.g., Biddle et al. 2012; Kathman 2011; Koga 2011; Pickering and Kisangani 2009), but, since it is based on the regularly updated UCDP/PRIO Armed Conflict Dataset, also is easier to maintain and keep up to date than other datasets on the subject. In addition, this alternative data can help shed further light on the causes of military intervention, some of which are still highly sensitive to alternative model specifications.1 Third, through its extended data coverage and inclusion of previously omitted variables, the study at hand provides robust empirical evidence on the motives of third-party interveners, which in turn has important implications for our normative assessment on military interventions. As I will show, the level of violence against civilians in an interstate conflict does affect third parties' decision whether or not to intervene, but only for certain third parties and in certain environments.
The remainder of this study proceeds as follows. First, I briefly review the existing literature on the relationship between humanitarianism and third-party interventions in civil conflicts. In the next section, I develop an expected utility theory of third-party intervention in civil wars. Under an expected utility framework, a third party will only militarily intervene in a civil conflict if the perceived benefits outweigh the expected costs, which in turn are determined by the preferences of the third party. These preferences, then, are key to explaining military interventions in civil wars. I argue that humanitarianism is a distinct preference of third parties, as protecting civilians abroad from violence satisfies normative pressures that compel them to do what is considered morally right. How strong of a preference humanitarianism is for a particular third party depends on a number of factors: humanitarian considerations are more likely to play a role in deciding whether or not to intervene in a civil war when the third party is (1) not a major power, (2) a democracy, and (3) not faced with the power politics that characterize a bipolar international system like that of the Cold War era.
Next, I lay out the research design of this study. I use rare events logistic regression to examine dyadic data on 174 civil wars between 1946 and 2008. The results of this analysis, which provide partial support for my theory, are presented in section five. In accordance with my expectations, major powers differ from weaker states across models, with only the latter being more likely to intervene in civil wars that exhibit high levels of violence against civilians. I also find support for my other two propositions, but these findings are sensitive to alternative model specifications. In some of the models, democracies are actually less likely to intervene in civil wars with high levels of violence against civilians than autocracies. A concluding section then summarizes the main findings of this study, connects them with the broader theoretical debate, and indicates possible direction for future research on the subject.
Humanitarian Considerations in Third-Party Interventions
Before elaborating on the conditions under which third parties intervene in civil conflicts, it is necessary to define a concept central to this study: intervention. At the most general level, a third- party intervention is the involvement of a state, group of states, or organization(s) not directly associated with one of the warring parties in a civil war with the aim of influencing the outcome of the conflict (Regan 2000; 2002). According to Regan and Aydin (2006), such an involvement can take different forms: third-party intervention should be understood as a continuum, with diplomacy/mediation as the least intrusive form of intervention and military commitments as the most intrusive. Economic sanctions as a form of intervention lie somewhere in the middle of this continuum (Escriba-Folch 2010). This study exclusively focuses on the most intrusive form of third-party interventions, that is, military interventions.2 Keeping with the definition used by UCDP/PRIO, I understand military intervention as an independent state entering a civil conflict with troops to actively support one of the warring parties; this excludes instances where a third- party simply supports one of the warring parties with money, weaponry, or other resources. Therefore, I use the terms “intervention”, “third-party intervention”, and “military intervention” synonymously in this study.
A wealth of research has addressed the reasons why third parties militarily intervene in intrastate wars.3 Within this literature, a prominent and, arguably, controversial question is whether humanitarian considerations matter in third parties' decisions about how to respond to civil conflicts. In examining this question, a large body of research has found that military interventions are, at least in part, driven by the interveners' desire to alleviate the suffering of and protect civilians from violence. For example, Regan (1998; 2000) has found that civil conflicts which produce high numbers of refugees exhibit a greater likelihood of outside intervention. According to Regan, this finding suggests that “humanitarian issues do seem to matter in decisions about how to respond to civil conflicts” (1998, 775). Examining the United States' responses to human rights violations in 153 countries between 1981 and 2005, Choi (2013) finds that the level of human rights violations in a civil war is a much more important determinant for U.S. interventions than the level of economic and strategic importance of a country in conflict.
Similarly, research by Gilligan and Stedman (2003) as well as Beardsley and Schmidt (2012) suggests that the United Nations' intervention behavior adheres more closely to the humanitarian mission laid out in its Charter than critics of the organization often claim. Specifically, they find that measures of the severity and escalatory potential of a conflict are significantly better predictors of the extent of UN involvement in international crises than variables that measure the purely private interests of the five permanent members of the UN Security Council. A last example for this strand of literature elaborated on here is Koga (2011), who finds no significant relationship between the amount of lootable resources (such as oil and gas) in a conflict state and the likelihood of third-party intervention. This indicates that despite often claimed otherwise, third parties do not intervene in conflict states with the goal of exploiting or gaining access to its natural resources. She does find a weak relationship between non-lootable resources (such as secondary diamonds and illicit drugs) and the likelihood of third-party intervention, but only for authoritarian regimes. This significantly weakens popular arguments about the selfish nature of military interventions in civil wars.
On the other hand, there are numerous studies asserting that humanitarian concerns are mostly window-dressing, and that in reality interventions primarily serve the national interest of the intervener. For example, Ross (2004) finds that opportunities to loot resources influence the decisions of some third parties to intervene on behalf of rebel groups. Gent (2008) argues that interveners are not interested in conflict resolution, as evinced by their lack of diplomatic efforts. Rather, the critical factors “are that the third party has preferences over a policy and that the outcome of this policy will be determined either by the party that wins the conflict or through a negotiated settlement between the two parties. Given this, third parties will prefer to intervene in civil conflicts when a military intervention will be most effective at increasing the likelihood of a more favorable policy outcome” (Gent 2008, 716).
Similarly, research has found that the existence of certain types of ties between the third party and the warring parties increases the likelihood of a military intervention. More specifically, the likelihood of a third party to intervene in a civil conflict increases when it is in an alliance with the government of a country in conflict or with one of the neighboring states facing the risk of contagion, when it has economic ties with one of these parties, when it has ethnic/ideological ties with one of the belligerents, and/or when the conflict takes place in one of its former colonies (Carment and James 1995; Davis et al. 1997; Kathman 2011). The presence or absence of such ties, which represent private interests of the third party, is said to outweigh potential humanitarian considerations.
The studies mentioned in this review a by no means a comprehensive account of the overall literature on the relationship between humanitarianism and third-party intervention. Yet, they illustrate the fact that there is overwhelming support for both sides of the argument, which suggests that there is truth to both views. In light of the evidence, it seems possible that the relationship between humanitarianism and third-party intervention is conditional, meaning that concerns over high levels of violence against civilians matter under certain conditions but not under others. In the following section, I develop a theory that explores these conditions. Therefore, it contributes to the ongoing debate about the possibility and prevalence of genuine humanitarian intervention.
An Expected Utility Theory of Third-Party Interventions
In this section, I develop an expected utility theory of third-party interventions, drawing on a rich tradition of research (Brito and Intriligator 1985; Bueno de Mesquita 1980; 1983; Gilpin 1983). Under an expected utility framework, individual decision-makers are assumed to order alternatives in terms of their preferences. The order of preferences is transitive so that if A is preferred to B and B is preferred to C, then A must be preferred to C. Individuals know the intensity of their preferences, with that intensity of preference being known as utility. Individuals consider alternative means of achieving desirable ends in terms of the product of the probability of achieving alternative outcomes and the utility associated with those outcomes. Decision-makers, being rational, always select the strategy that yields the highest expected utility (Bueno de Mesquita 1988). From this set of assumptions, it follows that a third party will only militarily intervene in a civil conflict if the perceived benefits outweigh the expected costs, which in turn are determined by the preferences of the third party. These preferences, then, are key to explaining military interventions in civil wars.
Determining a third party's preferences regarding intervention is a difficult task. The potential to infer a state's preferences from a specific outcome is limited. It is very well possible that a state has a certain interest but cannot materialize this interest, since the outcome is also determined by the preferences and actions of other states. Other strategies to determine a state's preferences are making assumptions or building on observations (for example using national elites as a proxy), but these strategies also exhibit important limitations. Thus, the best strategy for examining a state's interests is deduction from prior theory (Frieden 1999). A wealth of theoretical literature has taken up the task of determining the preferences of states.4 It is generally argued that the order of preferences varies across states, and that this order is determined by a state's position (or rank) vis-a-vis other states, its domestic attributes, and the international environment it is faced with (Moravcsik 1997).
I argue that humanitarianism is a distinct preference of states. By intervening in a civil conflict to protect civilians from further violence, third parties satisfy normative pressures that compel them to do what is considered morally right (March and Olsen 1998). Simply put, saving civilians abroad poses an intangible benefit for third parties. This intangible benefit, then, can also result in more tangible benefits, for example an improved reputation that can materialize as increased trade relations and the like. The challenge, then, is to develop a theoretical framework that illustrates how the strength of humanitarianism as a preference varies with each of the three aforementioned dimensions: a state's position (or rank) vis-a-vis other states, its domestic attributes, and the international environment it is faced with.
As for the first dimension, it has long been argued that the preferences of major powers are systematically different from those of other states (Waltz 1979; Keohane 1984; Dittmeier 2013). Major powers (or great powers) are those states whose material power, including their military prowess, economic capabilities, and geopolitical location, are disproportionately greater than those of other states (Snyder et al. 1972). One potential reason as to why their preferences are different from those of other states is that the main interest of major powers is to protect the status quo (i.e. the current distribution of relative power within the international system). The great powers are those benefitting the most from the current international order, and therefore have a strong incentive to fight off any challenge to this order. This includes the interference in other states' internal affairs with the goal of preventing regional hegemons from becoming great powers on a global scale (Dittmeier 2013). This concern for protecting the status quo leaves only little room for humanitarian concerns, which is why states that are not major powers can be expected to have a greater preference for humanitarianism.
A second, complementary argument is that weaker states are aware of the fact that they will likely never come close to reaching similar material capabilities as the great powers. Therefore, to establish themselves vis-a-vis the great powers, they try to improve their international standing by pioneering issues that are generally seen as morally right. This is the reason why many progressive norms, such as legal sex quotas, develop first among the margins of international society, rather than among the great powers (Towns 2010; 2012). The same argument can be made for humanitarian intervention. One only needs to look at the history of humanitarian intervention at the United Nations. In 2005, the UN codified humanitarian intervention in paragraphs 138 and 139 of its charter, which came to be known as the Responsibility to Protect (or R2P). Despite critics often arguing that R2P is a tool of Western imperialism, it was actually the African Union that championed the concept at the UN. The Western powers were initially reluctant to make humanitarian intervention a legal obligation of the international community, but eventually gave in to the lobbying by states such as South Africa and Rwanda (Bellamy 2009; 2011). Therefore, I expect states that are not major powers to be more likely to intervene in civil wars as a result of humanitarian considerations.
H1: Non-major powers are more likely than great powers to intervene in conflicts that exhibit high levels of violence against civilians.
As for the second dimension, a wealth of research argues that the preferences of democracies are systematically different from those of authoritarian regimes. Two reasons for this have been proposed, a normative logic and an institutional logic. Proponents of the normative logic argue that the key effect of democracy is to socialize political elites to act on the basis of democratic norms. The essence of these democratic norms is their universalism, i.e. the belief that every human, regardless of nationality and geographic location, is entitled to a set of basic rights, and that governments obligated to both respect and protect these rights (Doyle 1997; Owen 1997). It comes as no surprise, then, that democratic regimes have, on average, shown much greater respect for the rights of their citizens than autocracies (Hafner-Burton 2014). Because democratic leaders are committed to these norms they try, as far as possible, to adopt them in the international arena as well. In fact, it has been argued that the safety of civilians abroad is an integral part of democracies’ identities and therefore of their foreign policies (Risse and Sikkink 1999).
According to the institutional logic, democratic institutions and processes make leaders accountable to a wide range of social groups that may, in a variety of circumstances, worry about the human suffering in civil wars abroad. Accountability derives from the fact that political elites want to remain in office, that there are opposition parties ready to capitalize on unpopular policies, and that there are regular opportunities for democratic publics to remove elites who have not acted in their best interests. Moreover, several features of democracies, such as freedom of speech and open political processes, make it fairly easy for voters to rate a government's performance. In short, monitoring and sanctioning democratic leaders is a relatively straightforward matter (Lake 1992; Russett 1993). Therefore, when faced with public pressure to intervene in a civil conflict abroad to protect the civilian population, democratic leaders are more likely to respond to this pressure than their authoritarian counterparts.
Support for this argument comes from the literature on the so-called “CNN effect”, which has shown that democracies are more likely to intervene in a civil conflict that is extensively covered by their domestic media. The reason for this is that the images of the fighting create demands among the domestic population to end the human suffering in the conflict state, and as foreign policy makers are at least somewhat responsive to public opinion, it will make them more likely to intervene in civil conflicts which are extensively covered by the media (Shaw 1996; Robinson 2000). One example is the 1992 US intervention in Somalia, which has demonstrated the power of the media to influence domestic groups and thus move governments: “by focusing daily on the starving children in Somalia, a pictorial story tailor-made for television, TV mobilized the conscience of the nation's public institutions, compelling the government into a policy of intervention for humanitarian reasons” (Cohen 1994, 9-10). Taken together, then, the normative and institutional logics provide a compelling explanation why democracies are expected to have a stronger preference for humanitarianism than autocracies.
H2: Democracies are more likely than authoritarian regimes to intervene in conflicts that exhibit high levels of violence against civilians.
As for the third dimensions, it has been argued that the preferences of states vary with their international environment. The most drastic change to the international environment during the last century has arguably been the end of the Cold War (Lebow and Risse-Kappen 1996; Wenger and Zimmermann 2003). During the Cold War, states were, on average, faced with a much greater chance of being annihilated than they are today, which incentivized many smaller countries to align with one of the two blocs, and the powerful states to engage in bellicose policies such as proxy wars and arms races. In this environment, many states could simply not afford to let their actions be guided by humanitarian considerations rather than power politics. In other words, the bipolar international system during the Cold War was associated with a greater salience of security-related preferences. The Cold War therefore largely suppressed humanitarian considerations among states and replaced them with different rationales for behavior. With the end of the Cold War, states were generally faced with less serious threats to their immediate survival, which awarded humanitarianism a more prominent place among their preferences (Moyn 2010).
1 The analyses by Aydin (2010) and Findley and Teo (2006) are good examples in this context, as they reach contradictory conclusions regarding the effect of previous military interventions on the likelihood of future interventions.
2 One might object to this narrow focus and wonder why my analysis does not include diplomatic and economic interventions as well. The reason is that diplomatic and economic interventions are less costly than military ones, and therefore represent a less serious commitment to protecting human rights. In other words, only focusing on military interventions poses a “hard test” for my theory: if third parties care about human rights to the extent that they would undertake costly military interventions in order to protect them, it seems reasonable to assume that they will also undertake less costly measures.
3 For excellent reviews of this literature, see Biddle et al. (2012) and Kathman (2011).
4 See, for example, Finnemore (1996), Krasner (1978), Moravcsik (1997), and Snyder (2013).