Refugee flows and state contributions to post-Cold War UN peacekeeping missions

Abstract

Why do states participate in United Nations peacekeeping missions? Extant theory revolves around two benefits states derive from participation: the financial benefits participants receive for their contributions and the ability to further liberal norms abroad. Yet, these theories have received limited empirical support. In addition, they cannot explain where and when leaders send troops. To rectify these limitations, I focus on a more straightforward benefit states receive from limiting conflict and providing stability to war-torn regions. Namely, stabilizing conflict areas helps stem the externalities caused from ongoing conflicts, such as refugees. This explanation of participation in peacekeeping predicts who is most likely to send support (those fearing refugee inflows), the size of contribution (increasing relative to the inflows), and where they send assistance (to the specific conflict area producing the externalities). Statistical analysis of all UN peacekeeping missions since 1990 provides support for this argument and reveals that refugee flows better predict who is likely to contribute troops, and to what conflict, than the size of those contributions. While other realpolitik considerations strongly influence which states participate in UN peacekeeping, neither democracy nor state need is a statistically significant explanator of who contributes. However, state need does predict the size of contribution.

Keywords

civil war peacekeeping refugees United Nations

Introduction

Why do states contribute soldiers to United Nations peacekeeping missions? Since the end of the Cold War, the United Nations has made a concentrated effort to use its peacekeeping missions to protect civilians and promote human rights in post-conflict areas. As Secretary-General Ban Ki-moon (2011) argued, ‘All of us share a fundamental responsibility to do more to protect civilians caught up in the horrors of war.’ The Security Council has explicitly called upon the international community to protect civilians and promote human rights in a number of its peacekeeping mandates, including missions in Afghanistan (UNAMA), the Democratic Republic of Congo (MONUSCO), Sierra Leone (UNAMSIL), and Sudan (UNMIS), among others. The UN Department of Peacekeeping Operations (DPKO, 2013b) has placed the protection of civilians and the promotion of human rights at the core of its mission, stating, the ‘challenging mandate [of civilian protection] is often the yardstick by which the international community, and those whom we endeavour to protect, judge our worth as peacekeepers’. Additionally, ‘Human rights is a core pillar of the United Nations. All staff in peace operations have the responsibility to ensure the protection and promotion of human rights through their work’ (DPKO, 2013a). Understanding why states become involved in these peacekeeping missions can help shed light on why states are willing to protect foreign civilians, and promote human rights, outside their borders.

Despite the UN’s humanitarian rhetoric, I argue that participation in peacekeeping plays a role similar to military intervention for many states. Drawing on the war contagion and third-party intervention literatures, I argue that states provide peacekeeping support when a conflict threatens to produce externalities for the country. States fearing refugee inflows from crisis regions are more likely than other states to send military support to guarantee peace and stability in the area. This argument makes three contributions to our understanding of peacekeeping. First, the peacekeeping literature on troop contributions tends to focus on who contributes to missions (e.g. Berman & Sams, 2000; Blum, 2000; Bobrow & Boyer, 1997; Neack, 1995; Victor, 2010). My explanation predicts not only who is most likely to send support (those fearing refugee inflows), but also the size of contribution (increasing relative to the inflows) and where they send assistance (to the specific conflict area producing the externalities). Second, contrary to past work, I find that large-scale humanitarian crises do not increase participation in peace operations (e.g. Bove & Elia, 2011; Choi, 2013). Rather, direct externalities from the conflict help determine which states participate in a mission and how much they contribute. Lastly, contrary to past quantitative research (e.g. Lebovic, 2004), my work supports recent qualitative studies that argue that regime type is a poor predictor of which states participate in peacekeeping missions (e.g. Bellamy & Williams, 2013).

The United Nations frequently struggles to fulfill the level of personnel required by its peacekeeping mandates (see Bellamy, Williams & Griffin, 2004). Without significant international support, war-torn countries may be condemned to a perpetual cycle of instability, violence, and human rights abuse (Doyle & Sambanis, 2000, 2006). Understanding the factors that make states more willing to support peacekeeping may help us craft better policy to attract broad-based international support to help stabilize struggling post-conflict situations.

Existing scholarship

Extant scholarship on UN peacekeeping troop contributions focuses on two main motivations for states to provide troops to UN peacekeeping missions. First, states contribute to UN peacekeeping missions to capitalize on the monetary and tactical incentives the United Nations provides to participating countries. Poor states, which have difficulty funding their militaries, receive two benefits from contributing soldiers to UN peacekeeping missions. Contributing states receive $1,028 per soldier per month. This money is paid directly to the contributing country, allowing the government to use the funds to support their militaries or other projects within their states or to pay off their supporters (Berman & Sams, 2000; Victor, 2010). In addition, soldiers from less developed countries receive tactical training from officers of more sophisticated militaries. Upon returning to their home countries, these soldiers can disperse the information, strengthening the tactical ability of their country’s military (e.g. Bobrow & Boyer, 1997). While this research provides a very plausible explanation for why states participate in peacekeeping missions, scholars have not found empirical support for these claims. For example, Blum (2000) finds that there is no relationship between need and contributions. These results suggest that need may not be the significant motivating factor in many states’ decisions of whether to participate in peacekeeping missions.

Second, scholars have also focused on the role humanitarianism plays in inducing countries to contribute to UN peacekeeping operations. Since the end of the Cold War, the United Nations has refocused the attention of its peacekeeping missions to promote democracy and human rights while protecting civilians in war-torn areas (Ban, 2011; DPKO, 2013a,b; Ratner, 1995). Given this new focus on promoting liberal institutions in post-conflict countries, democracies have begun to participate more frequently, and contribute more troops, in UN missions. Participation in peacekeeping provides democracies the means to both respond to domestic pressures to provide humanitarian relief abroad and help further liberal norms of conflict resolution in other countries (e.g. Daniel, 2011; Lebovic, 2004; Marten, 2004). In this manner, peacekeeping allows democracies to further their interests and promote the status quo from which they benefit (Neack, 1995).

While this research has considerable merit in explaining why many liberal democracies participate in peacekeeping, it does not provide an understanding of why many illiberal autocratic states take this same behavior. It is clear from these accounts that states that routinely abuse their citizens, violate human rights, and use violence at home should not work to maintain peace and protect civilians abroad. Yet they do. Illiberal states, such as Pakistan, primarily conduct the UN’s peacekeeping operations. For example, Nigeria, which regularly ranks as one of the world’s worst violators of human rights (see Cingranelli, Richards & Clay, 2014), has participated in 14 United Nations peacekeeping missions throughout the world (DPKO, 2012), attempting to stabilize war-torn countries and provide protection to non-combatants in the area. As we look more closely at these illiberal states, we find that none of the conventional explanations for why states participate in peacekeeping can fully account for this puzzling behavior. Such violators do not respond to domestic humanitarians. While individuals in these states may have humanitarian leanings, they do not appear to have much influence or willingness to shift non-humanitarian domestic policy. Human rights abuses, such as torture and extrajudicial killings, are common in these states.

Clearly, the extant literature has difficulty in predicting who contributes to peacekeeping and who does not. Furthermore, existing arguments about peacekeeping cannot explain where states send troops and how large a contribution they make. Regime type and military need may be able to describe aggregate levels of contribution, but they cannot explain where states focus their efforts. These explanations expect that democracies and militarily needy states will be more likely to participate. However, contributions appear strategic. For example, while Ghana ranks among the top contributors to UN peacekeeping missions, it has not contributed troops to any post-Cold War missions outside Africa (DPKO, 2012). In addition, not all democracies behave similarly. For instance, while France has committed nearly 40,000 soldiers to UN peacekeeping missions since the end of the Cold War, Venezuela has refused to contribute (DPKO, 2012).

Stabilizing war zones

While the peacekeeping literature focuses on different goods states can obtain for participating in peacekeeping missions, such as monetary subsidies or support from domestic humanitarian groups, I propose to refocus our understanding of what states can obtain in participating in these projects. Drawing from the civil war and military intervention literatures, I propose to focus on the more straightforward benefit states receive from limiting conflict and providing stability to war-torn regions – namely, stabilizing conflict areas helps stem the externalities caused from ongoing conflict and may allow refugees to return to their home countries. As Kathman (2010, 2011) and Regan (1998, 2002) argue, the externalities produced by a conflict may draw third parties into the war to end the conflict and stabilize their countries’ borders.

War causes a number of externalities for nearby states, including a decline in economic growth and the diffusion of conflict across borders. One of the most common concerns states face from nearby conflicts is the threat of refugee inflows. Refugee inflows strain host country resources and create or exacerbate tensions and grievances within the area. In particular, inflows can hurt the economy (Murdoch & Sandler, 2002, 2004), cripple health systems (Ghobarah, Huth & Russett, 2003, 2004), strain food supplies (Jenkins, Scanlan & Peterson, 2007), and shift demographics within the country (Saideman & Ayres, 2000). For these reasons, conflicts diffuse between states as the refugees’ home and host states often engage in militarized conflict with one another over the flows of people (e.g. Salehyan, 2008; Salehyan & Gleditsch, 2006).

This conflict literature notes that refugee inflows can induce states to invade their neighbors to stem these externalities. I extend this argument to explain peacekeeping behavior. Peacekeeping missions can help reduce the likelihood that a conflict will diffuse by stabilizing the war-torn country and protecting civilians. Upon entering a conflict, peacekeepers separate belligerents from one another and police the interactions between combatants and civilians. By separating the warring parties, peacekeepers help ensure that accidents or misunderstandings between the sides do not reignite fighting, keeping levels of violence in the area low. With lower levels of violence within the state, civilians are less likely to flee the country as refugees. Policing interactions between combatants and civilians also helps reduce refugee flows. By ensuring that combatants do not abuse civilians during the ceasefire, peacekeepers can make remaining in the country a more favorable option for individuals in the country. The focus peacekeepers place on lowering levels of violence within the state helps reduce the number of refugees leaving the country and pulling nearby states into the conflict (Beardsley, 2011).

Some scholars have suggested that states engage in humanitarian interventions to help stem the externalities they expect to receive from nearby conflicts (e.g. Snyder, 2010). Peacekeeping is one such type of intervention. Since peacekeepers can help stabilize a war-torn area and reduce the number of refugees leaving the area, contributing to a successful peacekeeping mission benefits a state by lowering the externalities it may receive from the conflict, helping reduce the likelihood it is drawn into a nearby conflict. A state’s desire to contribute peacekeeping troops should thus be linked to the benefits it receives from securing its borders. Note that this argument is different from one that focuses on the total number of displaced persons fleeing a conflict. Rather than focusing on forced migration as a global humanitarian concern (e.g. Bove & Elia, 2011), this argument suggests that direct dyadic flows from the conflict country to a third-party state helps explain the third party’s willingness to contribute to the peacekeeping mission.

Peacekeeping contributions

The negative effect refugees have on a host country will induce a state that is expecting inflows to take action to stop these externalities (e.g. Kathman, 2010, 2011; Regan, 1998, 2002; Salehyan, 2008). Tying this argument to peacekeeping, and suggesting that preventing refugee inflows drives participation in peacekeeping missions, allows me to explain who will participate, to which conflicts they will direct their efforts, and how much they will contribute to the mission. The answer to each of these questions – who, where, and how much – should be determined by the state’s motivation to prevent refugees from entering its territory.

Given the threat states face from refugee inflows, regimes experiencing these externalities should be more likely than other countries to work to provide peace and security and protect foreign civilians in other states so that the refugees do not continue to enter its territory. If the state can eliminate the source of the refugee problem, then it will be better off for not having to face the tensions created by these arrivals as they continue to enter the country. Due to the fear of the consequences of these inflows, many states are willing to take efforts abroad to stabilize war-torn countries and protect civilians in the area. For example, in defending his decision to engage in peacekeeping in Haiti, US President William Clinton directly referenced the need to halt the conflict to stop refugee flows due to the costs they were imposing on America:

We have a particular interest in stopping brutality when it occurs so close to our shores […] As long as Cedras rules, Haitians will continue to seek sanctuary in our nation. This year, in less than two months, more than 21,000 Haitians were rescued at sea by our Coast Guard and Navy. Today more than 14,000 refugees are living at our naval base in Guantanamo. The American people have already spent $177 million to support them. (White House, 1994)

Countries seeking to stop refugee inflows can participate in the UN’s multilateral peacekeeping efforts to alleviate the externalities they experience from nearby ongoing conflicts. Multilateral peacekeeping is a useful tool for states because it allows members to aggregate their capabilities through the UN towards a common goal of stabilizing the target country. By providing a stable structure through which to coordinate actions, and a supportive administrative apparatus to handle conflicts that arise as their members interact, the United Nations helps member states overcome their collective action problems in jointly pursuing their policy goals in peacekeeping (e.g. Abbott & Snidal, 1998). By uniting the capabilities of its member states and providing a centralized mechanism through which to operate, UN peacekeeping can help states enhance the efficiency of their efforts in stemming refugee inflows from surrounding conflicts. In this manner, multilateral action through the UN can help its members stop the spread of refugees and war (Beardsley, 2011).¹

However, a state must be strategic in how it distributes goods (e.g. Bueno de Mesquita et al., 2003). The state is unlikely to dedicate resources to situations in which it will not receive a return on its investment. Resources dedicated to low-return projects are wasteful for the state because it no longer has these goods at its service and it has not fortified its position. In terms of dispatching peacekeeping troops, this implies that a state will be unlikely to participate in operations that will not have an effect on preventing refugees from entering its territory. Troops sent to areas that are unlikely to produce refugees for the state cannot help the state achieve its goal of stopping inflows.

To help clarify this logic, imagine a hypothetical scenario in which Ghana chooses between contributing to a peacekeeping mission in Haiti or a mission in the Ivory Coast, noting that Haiti is outside of Ghana’s region and the Ivory Coast is inside Ghana’s region. Further, imagine that Haiti and the Ivory Coast are experiencing humanitarian crises of equal scale. However, only the conflict in the Ivory Coast is causing externalities for Ghana, despite both conflicts producing similar levels of total refugees.

In this hypothetical scenario, if Ghana contributes troops to PKO A in Haiti, these soldiers will be able to help ameliorate the humanitarian crisis in Haiti, but they will be unable to alleviate the violence that is driving the refugees into Ghana. While these troops may limit violence in Haiti and help Haitian civilians, from Ghana’s point of view and its concerns, these resources are not used in a beneficial manner, as they do not help Ghana stem the tide of refugees entering its territory from the Ivory Coast. With the particular problem of preventing refugee inflows as motivation, and a strategy of directing its resources towards relieving this problem, a state expecting refugee inflows from a specific conflict will focus its resources on the actual area of concern. In this hypothetical, then, Ghana should be more likely to participate in peacekeeping in the Ivory Coast than in Haiti, despite both crises being equal humanitarian crises, because only the crisis in the Ivory Coast directly threatens Ghana’s security.

Hypothesis 1: A state is more likely to contribute to peacekeeping missions in situations from which it is experiencing refugee inflows.

In addition to influencing the likelihood of participation, refugee inflows should also affect the size of a state’s contribution. Conflicts producing higher levels of refugees are more destabilizing to the region and require stronger third-party action to help end the conflict (e.g. Fortna, 2008). A state interested in ending refugee outflows from the target country must consider the force size necessary to successfully stabilize the conflict country and stem the refugee flows it is receiving. As the conflict produces more refugee inflows for a state, the state is likely to scale its contribution in relation. Severe conflicts that produce a large number of refugee outflows will require more third-party troops to keep the peace and prevent the violence that causes refugee outflows (e.g. Hultman, Kathman & Shannon, 2013). Therefore, as a state receives more refugees from a particular conflict, the state is likely to contribute more troops to the peacekeeping mission that is dedicated to that war-torn country.

Hypothesis 2: A state is likely to contribute more troops to peacekeeping missions in situations from which it is experiencing higher volumes of refugee inflows.

Empirical analysis

In the sections above, I presented a theory to explain why states become involved in peacekeeping. To test the two hypotheses I derive from this argument, I focus my empirical analysis on state contributions of troops to United Nations peacekeeping missions in the post-Cold War era. When participating in UN peacekeeping missions, states can direct their contributions to the troubled area of interest. States’ ability to target their efforts at particular areas allows me to explore the strategic relationship between the state and the other territory. Thus, rather than focus on each state’s aggregate contributions to the UN’s peacekeeping efforts (e.g. Bobrow & Boyer, 1997; Lebovic, 2004), I use a directed-dyad-year research design that allows me to examine each state’s contribution to each mission throughout the mission’s existence. The unit of analysis is the third-party-mission-year for all states not already party to the ongoing conflict for all UN peacekeeping missions to civil war countries between 1990 and 2011.

Data

I generate two testable hypotheses from my theory. Hypothesis 1 concerns which states contribute to each UN peacekeeping mission. Hypothesis 2 concerns the number of troops contributed by each state. Therefore, there are two dependent variables in this analysis.

Participation

The first dependent variable is whether a state participates in a given peacekeeping mission. Kathman (2013) provides data on country participation in peacekeeping missions. I assign an indicator variable of 1 to each state that participated in a given mission; 0 to all others.

Contribution size

The second dependent variable is the size of a state’s contribution to a given peacekeeping mission. Kathman (2013) provides these data on the total number of troops a state contributes to each UN peacekeeping mission. Contribution sizes range from 0 to over 7,000 (Pakistan in UNOSOM).²

Refugee flows

The main independent variable in each of my models is the size of the refugee flows between the country receiving the peacekeeping mission and each potential contributor state in each year of an ongoing mission. By measuring dyadic refugee flows from each conflict area to each potential donor state, I am able to calculate the direct externalities received by each third party from the conflict, rather than capturing a general measure of conflict severity for the international community. I collect data on directed refugee flows for each dyad from the United Nations High Commissioner for Refugees (UNHCR, 2012). Dyadic flows of refugees are highly skewed. In these data, refugee inflows range from 0 for many dyads (e.g. Haiti to Guyana in 2000) to 1.9 million from Afghanistan to Pakistan in 2010. Due to the skew of this variable, I use its log³ in my models, Log(Refugee inflows). I lag this variable by one year to establish the sequence of causality captured in the data.

Log(Refugee inflows) is my main variable of interest in the models examining state contributions to peacekeeping missions. I expect that as the potential donor state experiences refugee inflows from the war-torn state receiving the peacekeeping mission it will become more likely to participate in the peacekeeping effort. Note that because of the dyadic nature of this inflow variable, we can separate the possibility that all states respond to severe conflicts that produce large refugee flows from the idea that states are more likely to intervene in these conflicts when the refugees affect them directly.

I include 14 control variables along with my main variable of interest. I lag all variables one year.

Democracy

I include whether the state is a Democracy based on the –10 to 10 Polity IV scale (Marshall & Jaggers, 2010). Extant explanations of peacekeeping expect that democracies will contribute more than other regimes to externalize their domestic norms of peace (Lebovic, 2004) or to please humanitarian factions within their state (Jakobsen, 1996; Western, 2002). If these explanations are correct, we should expect that as a state’s Polity score increases (the state becomes more democratic), the state will participate increasingly in the peacekeeping missions.

Financial need

I include a variable to capture financial need. States may participate in peacekeeping to help fund their militaries (e.g. Berman & Sams, 2000; Victor, 2010). If this argument is correct, poorer states should contribute more troops to collect the per solider per month reimbursement provided to participating countries. To test this theory, I include each state’s gross domestic product (GDP) per capita (World Bank, 2012). Given the high rightward skew of GDP per capita levels, I use the log of this variable in my models, Log(GDPpc).

Strength

I include a measure of each state’s military strength based on military personnel. Stronger states may be more likely to intervene in civil violence, provide peacekeeping, and protect civilians, to snuff out smaller conflicts before they can expand and draw the major powers back into the battlefield against one another (e.g. Mueller, 2004). Data on military personnel come from the Correlates of War Project (Singer, 1988). Given the high rightward skew of troop levels, I use the log of this variable in my models, Log(Military personnel).

Distance

I include the log of distance, Log(Distance), between each potential contributor and each war-torn country to control for the possibility that states are more likely to send troops to nearby conflicts because of a fear of conflict contagion or other geopolitical concerns, regardless of their refugee inflows. Bennett & Stam (2000) provide data on distance between countries.

Population

I include the log of each potential contributing state’s population, Log(Population), as more populous states have more people to potentially contribute to peacekeeping missions. The World Bank (2012) provides population data.

Joint ethnicity

I include an indicator variable of 1 to all dyads in which the majority of the population in the third-party and war-torn states share a common ethnicity; 0 for all others. States may intervene in a conflict to support a regime with which it shares a common ethnicity. Additionally, controlling for joint ethnicity helps ensure that the refugee flow variable is not capturing situations of states aiding those with a common heritage. Data on joint ethnicity are provided by the CIA World Factbook (2013).

Alliance

I include a variable for whether each potential contributor and the target state are allies, coded 1 if the dyad is an ally; 0 otherwise. Allies share common strategic and security interests (Russett & Oneal, 2001) and may be more willing to engage in peacekeeping in a partner state to provide support for the unstable regime (Beardsley & Schmidt, 2012). Data on alliances are provided by Gibler (2009).

Colonial ties

I include an indicator of 1 to each dyad in which the target state was a former colony of the potential contributor; 0 for all others. Colonial powers may be more likely to intervene in countries with which they have a history of colonial involvement (see Kathman, 2011). Data on former colonies comes from Tir et al. (1998).

Total refugee outflows

I include a variable capturing the total refugee outflows from each conflict each year (UNHCR, 2012). Conflicts that create severe externalities for the international community may be a significant threat to international security. Given this concern, and the humanitarian motives such crises provide potential contributors, crises that produce large numbers of forced migrants may be more likely to see intervention (e.g. Bove & Elia, 2011). Given the skew of this variable, I use its log in my models, Log(Total refugee flows).

Mass killing

I include an indicator variable equal to 1 for all conflicts in which the government engaged in a campaign of genocide or politicide; 0 for all others (Harff, 2003). As genocide is considered the crime of crimes in the international community, such episodes of mass killing may attract more international intervention than will other conflicts (see Schabas, 2000).

Other contributors

I include the count of other states contributing to a mission in each year. Contributions by other states may reduce the number of refugees fleeing the conflict into a state’s territory. If the refugee situation can be handled without its involvement, a state is unlikely to waste its soldiers on the conflict. Given the skew of this count, I include its log, Log(Other contributors), in my models.

UN Security Council permanent five member

I include an indicator variable equal to 1 for the five permanent members of the United Nations Security Council (UNSC); 0 for all others. The UNSC must authorize UN peacekeeping missions. Including the variable P5 member controls for the possibility that the permanent five members (China, France, the Russian Federation, the United Kingdom, and the United States) may only allow peacekeeping missions in those areas in which they have a strategic interest in ending the conflict (Diehl, 2008).

Past contributions

When testing Hypothesis 2, I include the lagged dependent variable in my time-series models. Not only does including this variable help with problems of autocorrelation in the data, but it also controls for the possibility that some states may participate more in peacekeeping because they have already built up the infrastructure to do so in past missions. In addition, it controls for the possibility that some states may be overstretched militarily and unable to participate in any additional mission.

Time polynomial

To control for temporal dependency in the data, I include the number of years, years squared, and years cubed since a state last contributed to a peacekeeping mission (see Carter & Signorino, 2010).

Models

The dependent variable Participation is binary. I therefore use a logistic regression to analyze the relationship between Log(Refugee inflows) and Participation.

The dependent variable Contribution size is the yearly count of each state’s contribution to a given peacekeeping mission. This count variable is not distributed normally and has a variance that is greater than its mean. It appears to follow a negative binominal distribution, in which most contributors provide nominal numbers of troops (if any) and only a few countries provide hundreds or thousands of soldiers to the mission. When faced with a negative binominal distribution, it is sometimes appropriate to employ a negative binomial model to estimate the parameters of interest.

However, a significant number of states never participate in UN peace operations. Therefore, it may be the case that two populations exist. The first is a ‘certain zero’ population that never contributes to peacekeeping. States in this population will always have a contribution size of zero. The second is a population that sometimes participates. States in this population will have contribution sizes that vary depending on the incentives the state faces each mission. A Vuong test suggests that this two-population process is likely at work in the data.

Given the two-population data generating process, I model the relationship between Log(Refugee inflows) and Contribution Size using a zero-inflated negative binomial model (ZINB) (e.g. Green, 1994). The first stage of this model estimates how likely it is that a state never participates in a given peacekeeping mission using a logistic regression. It models whether there are ‘certain’ zeros in the data – actors who never participate. Taking into account the predictions of this first stage and the actors who never participate in peacekeeping, the second stage uses a negative binominal model to estimate how many troops a given state is likely to contribute to a particular mission.

Results

My theory suggests that states are strategic in where they send their troops. States experiencing inflows of refugees should be judicious in where they make their contributions. As the results in Table I indicate, strong support exists for my argument that refugee flows help determine, in part, which states participate in each UN peacekeeping mission. Model 1 is a parsimonious test of whether refugee inflows increase the likelihood that a state participates in peacekeeping. It includes only the refugee inflow variable and the polynomial of time since each state last contributed to a mission to control for time dependence. I find that states facing inflows of refugees from a particular conflict are more likely to contribute troops to that area than are other states. As refugee inflows increase by one standard deviation, a state is 17% more likely to participate in the mission. This supports my argument and Hypothesis 1.

Table I.

Logit analysis of contributions to each UN Peacekeeping Mission

	Model 1: Base	Model 2: Controls	Model 3: CINC	Model 4: No distance	Model 5: Mission FE
Dependent variable = Participation	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)
Log(Refugee inflows) _t– ₁	0.159* (0.016)	0.047* (0.022)	0.047* (0.022)	0.078* (0.021)	0.030* (0.015)
Democracy _t– ₁		0.007 (0.011)	0.006 (0.011)	0.007 (0.011)	0.011 (0.007)
Log(GDPpc) _t– ₁		–0.032 (0.047)	–0.027 (0.072)	–0.063 (0.046)	–0.032 (0.028)
Log(Military personnel) _t– ₁		0.006 (0.089)		0.041 (0.087)	0.013 (0.046)
Log(CINC) _t– ₁			0.001 (0.146)
Log(Distance) _t– ₁		–0.151* (0.043)	–0.149* (0.042)		–0.185* (0.028)
Log(Population) _t– ₁		0.187* (0.085)	0.190 (0.151)	0.142 (0.085)	0.194* (0.046)
Joint ethnicity _t– ₁		–1.301* (0.545)	–1.303* (0.543)	–1.223* (0.568)	–1.516* (0.394)
Alliance _t– ₁		0.784* (0.203)	0.784* (0.203)	0.899* (0.198)	1.101* (0.139)
Colonial ties _t– ₁		1.497* (0.603)	1.499* (0.602)	1.596* (0.580)	1.702* (0.238)
Log(Total refugee flows) _t– ₁		0.004 (0.024)	0.004 (0.024)	–0.004 (0.025)	0.045 (0.046)
Mass killing _t– ₁		0.417* (0.153)	0.414* (0.153)	0.436* (0.154)	–0.509* (0.209)
Log(Other contributors) _t– ₁		0.936* (0.049)	0.934* (0.048)	0.914* (0.048)	–0.043 (0.049)
P5 _t– ₁		–0.136 (0.236)	–0.134 (0.235)	–0.158 (0.235)	–0.076 (0.132)
Years until contribution	–1.101* (0.047)	–0.919* (0.073)	–0.926* (0.074)	–0.917* (0.073)	–1.051* (0.073)
Years until contribution²	0.078* (0.010)	0.044* (0.020)	0.045* (0.020)	0.046* (0.020)	0.066* (0.019)
Years until contribution³	–0.002* (0.000)	–0.000 (0.001)	–0.000 (0.001)	–0.001 (0.001)	–0.001 (0.001)
Constant	–1.888* (0.065)	–5.630* (1.281)	–5.674 (3.828)	–5.896* (1.267)
N	48,762	27,410	28,104	27,410	21,891
Log pseudolikelihood	–7,557.6068	–3,875.5992	–3,889.7545	–3,890.8951	–3,501.617

*p < 0.05. Errors clustered by the potential contributor–mission dyad.

Model 2 re-estimates Model 1 and includes the control variables.⁴ Model 2 indicates that after controlling for confounding effects and alternative arguments, refugee inflows increase a state’s likelihood to contribute to a peacekeeping mission by 5%.⁵ Substantively, this effect is produced, for example, by moving from an inflow of 1,000 refugees to 10,000 refugees. In the post-Cold War era, the contribution of peacekeepers from European countries has decreased, while it has increased from African and Asian countries. While this pattern may be evidence of the UN broadening the base of its peacekeeping contributors (see Perry & Smith, 2013), these patterns can also be explained, in part, by my argument. Africa and Asia have seen more civil war and forced migration than Europe in the post-Cold War era. In response, the UN has authorized more peacekeeping missions in these areas (DPKO, 2012), and nearby refugee-receiving states have responded by contributing more military personnel to these missions to help decrease refugee flows into their territories (Kathman, 2013).⁶

Model 2 also provides mixed support for the existing scholarship on state contributions to UN peacekeeping missions. First, I find that democracies are no more likely to participate in any given UN peacekeeping mission than are non-democracies (see Lebovic, 2004). Daniel (2011) suggests that the UN should draw more heavily from democracies interested in promoting a Western agenda globally when trying to fulfill its mandated troop levels, and Bellamy & Williams (2013) note that it is important for these Western democracies to contribute, as they have better military capabilities to assist the UN. However, my results indicate that democratic interests are similar to those of non-democracies: they are not likely to become involved in peacekeeping unless the crisis threatens them directly. Second, I find that poor states are no more likely to participate in UN missions than are their wealthier peers. These results fail to support a common refrain in the peacekeeping literature (Berman & Sams, 2000; Bobrow & Boyer, 1997; Bove & Elia, 2011; Victor, 2010). Similarly, military personnel levels and membership in the P5 do not have a significant influence on the likelihood of a state participating in a peacekeeping mission. Additionally, while Fortna (2008) finds that peacekeeping is more likely when conflicts generate greater refugee crises and Bove & Elia (2011) find that the total size of the humanitarian crisis in terms of forced migration is an important influence on member states’ willingness to contribute to the mission, I find that this is not the case. Instead, total refugee flows from the conflict are an insignificant predictor of third-party participation, whereas dyadic flows from the conflict country to a potential contributor are much more likely to increase contribution.

Model 2 indicates that several variables do have a positive effect on a state’s likelihood of participation. States with larger populations are 21% more likely to contribute, suggesting that states with large populations may use peacekeeping as a way to export labor.⁷ Similarly, I find that states allied to the war-torn country are 119% more likely to participate in peacekeeping. This result indicates that military considerations influence how states react to conflicts (e.g. Mueller, 2004). I also find that mass killing of civilians in the war-torn country increases the likelihood that a state will contribute to the peacekeeping mission by 52%. This finding supports recent work suggesting a growing norm of civilian protection in the post-Cold War era (e.g. Hultman, 2013). Related, the contributions of other states do not deter a state from contributing to peacekeeping. Instead of free riding on other members, a state becomes 155% more willing to share the burden of peacekeeping as others contribute to a mission.⁸ Lastly, I find that colonial ties have the largest effect on which states contribute to each mission, increasing the likelihood of contribution by 347%.

Conversely, third parties further from the conflict are less likely to participate in peacekeeping. This pattern also reflects the dynamics of non-UN military intervention into civil wars (e.g. Kathman, 2010, 2011; Regan, 1998, 2002). Surprisingly, states that share a joint ethnicity with the war-torn country are 73% less likely to contribute to the peacekeeping mission. Figure 1 graphs the substantive results for the significant predictors of peacekeeping participation from Model 2.

Figure 1.

Percentage change in the likelihood of participation

One potential concern with this analysis is that relying on the number of soldiers a state possesses as a measure of military capability misses the fact that troops are not the only source of a state’s military capabilities. This may explain why the military personnel variable is statistically insignificant in Model 2. To investigate this concern, I re-estimate Model 2 using the Correlates of War CINC scores (Singer, 1988) to proxy total state strength in place of military personnel. However, as Model 3 indicates, CINC scores are also a statistically insignificant measure of state strength.

A second concern may be that refugees tend to flee to nearby states. Thus, Model 2 may suffer from issues of multicollinearity, as refugee flows may be highly correlated with distance between states. To ensure that the results of my refugee variable are robust, I re-estimate Model 2 after dropping Log(Distance). As Model 4 indicates, the effects of refugee inflows are robust to the inclusion or exclusion of distance.

Lastly, there may be unobservable mission-specific effects that are uncontrolled for in Model 2. For instance, backroom negotiations during the drafting of the mission mandate or political pressures on member states during the mission may influence a state’s willingness to contribute to a specific mission. To determine whether unobserved mission-specific influences are affecting my results, I re-estimate Model 2 and include mission fixed effects. In Model 5, I once more find that the effect of refugee inflows is robust.

Models 1–5 support Hypothesis 1 and my claim that refugee inflows influence a state’s willingness to participate in UN peacekeeping missions. I now evaluate Hypothesis 2: a state is likely to contribute more troops to peacekeeping missions in situations from which it is experiencing higher volumes of refugee inflows. Once more, I begin with a parsimonious test of my argument. Model 6 is a zero-inflated negative binomial (ZINB) regression that contains two stages. The first stage estimates how likely a state is to never participate. It includes only the refugee inflow variable and a time polynomial since each state’s last contribution to control for time dependence. In the first stage of the model, I find that refugee inflows decrease the likelihood that a state will never participate in a UN peacekeeping mission by 78%. The second stage estimates the likely size of a state’s contribution to a given mission. It includes only the refugee variable and the size of the state’s lagged contribution. In the second stage, I find that states facing refugee inflows from a particular conflict are likely to contribute more troops to that area than are other states. However, this result is not statistically significant. Therefore, the ZINB does not support Hypothesis 2.⁹

Model 7 re-estimates Model 6 and includes the control variables (see Table II). After controlling for confounding effects and alternative arguments, Model 7 indicates that states receiving refugee inflows contribute 8% more troops to the peacekeeping mission than do other states. This result is significant at the 0.1 level. However, as Models 8 and 9 indicate, this result is not robust to different model specifications.

Table II.

Zero-inflated negative binominal of contributions to each UN peacekeeping mission

	Model 6		Model 7		Model 8		Model 9
	Stage 1: Never participate	Stage 2: Size of contribution	Stage 1: Never participate	Stage 2: Size of contribution	Stage 1: Never participate	Stage 2: Size of contribution	Stage 1: Never participate	Stage 2: Size of contribution
	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)	Coefficient (Standard error)
Log(Refugee inflows) _t– ₁	–0.217* (0.068)	0.032 (0.036)	–0.139* (0.056)	0.078† (0.046)	–0.165* (0.052)	0.043 (0.045)	–0.121* (0.053)	0.070 (0.046)
Democracy _t– ₁			–0.009 (0.027)	–0.019 (0.025)	–0.022 (0.027)	–0.031 (0.024)	–0.008 (0.027)	–0.020 (0.026)
Log(GDPpc) _t– ₁			–0.023 (0.123)	–0.296* (0.099)	0.205 (0.205)	–0.361 (0.185)	–0.043 (0.117)	–0.281* (1.10)
Log(Military personnel) _t– ₁			0.189 (0.309)	0.382 (0.206)			0.203 (0.304)	0.374 (0.210)
Log(CINC) _t– ₁					–0.401 (0.417)	0.422 (0.346)
Log(Distance) _t– ₁			–0.085 (0.106)	0.063 (0.133)	–0.118 (0.106)	0.056 (0.128)
Log(Population) _t– ₁			–0.215 (0.235)	–0.150 (0.179)	0.347 (0.418)	–0.272 (0.349)	–0.250 (0.226)	–0.129 (0.195)
Joint ethnicity _t– ₁			0.803 (0.732)	–0.338 (0.527)	0.603 (0.710)	–0.757 (0.444)	0.830 (0.726)	–0.413 (0.509)
Alliance _t– ₁			–0.097 (0.399)	0.102 (0.294)	–0.070 (0.409)	0.190 (0.296)	–0.068 (0.398)	0.082 (0.291)
Colonial ties _t– ₁			–0.422 (0.755)	0.089 (0.538)	–0.540 (0.764)	0.014 (0.494)	–0.367 (0.738)	0.095 (0.531)
Log(Total refugee flows) _t– ₁			0.142* (0.038)	0.048 (0.045)	0.151* (0.036)	0.061 (0.046)	0.139* (0.038)	0.052 (0.042)
Mass killing _t– ₁			–1.354* (0.513)	–0.945* (0.294)	–1.323* (0.515)	–0.923* (0.303)	–1.349* (0.522)	–0.963* (0.288)
Log(Other contributors) _t– ₁			–1.671* (0.122)	–0.669* (0.160)	–1.669* (0.120)	–0.665* (0.155)	–1.665* (0.121)	–0.666* (0.161)
P5 _t– ₁			0.329 (0.641)	–0.682 (0.492)	0.646 (0.609)	–0.499 (0.497)	0.352 (0.633)	–0.703 (0.498)
Log(Past contribution) _t– ₁		0.584* (0.028)		0.562* (0.047)		0.560* (0.047)		0.557* (0.048)
Years until contribution	2.314* (0.419)		1.039* (0.190)		1.092* (0.187)		1.038* (0.187)
Years until contribution²	–0.273* (0.077)		–0.082 (0.044)		–0.094* (0.043)		–0.083* (0.044)
Years until contribution³	0.010* (0.004)		0.003 (0.002)		0.004 (0.002)		0.003 (0.002)
Constant	–2.921* (0.570)	2.600* (0.192)	5.084 (3.186)	6.983* (3.075)	–7.422 (10.423)	13.622 (9.361)	5.137 (3.173)	7.027* (3.096)
α	9.895* (0.597)		5.224* (0.402)		5.207* (0.392)		5.178* (0.394)
N	6,217		3,821		3,824		3,821
Log pseudolikelihood	–15,284.170		–9,136.139		–9,139.376		–9,137.390

*p < 0.05. # Errors clustered by the potential contributor–mission dyad. † Log(Refugees inflows) is significant at the 0.1 level in Model 7.

Comparing Models 7–9 to the logit analysis reveals that while several factors influence a state’s participation and contribution size in similar ways, others affect these decisions differently. For example, not only is democracy an insignificant predictor of participation in post-Cold War peacekeeping missions, but it also has no influence on the size of a state’s contribution. The same is true for membership in the P5 and state strength. Similarly, I find that total refugee outflows from the conflict state do not affect a third party’s decision about whether to contribute to the mission or how many troops to provide. This result stands in contrast to previous work (e.g. Bove & Elia, 2011) and highlights the importance of focusing on dyadic refugee flows.

Several factors that increase the likelihood of a state’s participation in a given mission have no influence on the size of its contribution. For instance, while colonial ties have the largest influence on which states participate in a mission, they do not increase the size of the contribution the third party provides. Similarly, while a large population and alliance ties each increase the likelihood that a state will contribute to the peacekeeping mission, neither of these variables is statistically associated with the size of a state’s contribution. I also find that while distance and joint ethnicity each decrease the likelihood that a state participates in a given peacekeeping mission, these variables have no statistically significant influence on the size of a participant’s contribution. Conversely, while I find that wealth does not affect which states participate in a mission, Models 7 and 9 indicate that poorer states provide between 24% and 26% more troops to each mission than do their wealthier peers.¹⁰

Two variables are statistically significant in both the logit analysis (Models 2–5) and the ZINB models (Models 7–9), but produce divergent effects. First, I find that while the presence of mass killing in a post-conflict state increases the likelihood that a state will contribute peacekeepers to the mission, mass killing decreases the size of the peacekeeping force the state is willing to contribute to the area. This finding raises questions concerning whether states are truly internalizing norms about protecting civilians from violence or whether they are sending nominal levels of troops as cheap talk in supporting such a norm. Second, while the presence of other contributors increases the likelihood that a state will participate in a peacekeeping mission, other contributors lead to a decrease in the number of troops each state contributes. Figure 2 graphs the substantive results for the significant predictors of peacekeeping contribution size from Model 7.

Figure 2.

Percentage change in the size of contribution

Discussion and conclusion

Our current understanding of UN peacekeeping troop contributions focuses on which states participate in missions, but has difficulty explaining to which conflicts they send troops and the size of contributions. To rectify these limitations, I draw from literature on conflict and military interventions to focus on the benefits states receive from limiting conflict and providing stability to war-torn regions. States engage in efforts to stabilize war-torn countries when the conflict produces externalities threatening to affect the country directly. By ending violence, a state hopes to stem the potential externalities arriving in its territory. This argument helps explain why weak and strong states and democracies and non-democracies are all willing to participate in some peacekeeping missions, but not others.

My findings indicate that states facing refugee inflows are likely to participate in peacekeeping missions to those countries generating the refugees. These results indicate that states participate in peacekeeping missions when it is in their interest, rather than to further an idealistic commitment to the global community. I also find that while other realpolitik considerations, such as alliance and colonial ties, strongly influence which states participate in UN peacekeeping, some of the primary explanations for why states contribute troops are unsupported. For example, neither democracy (e.g. Lebovic, 2004) nor need (e.g. Berman & Sams, 2000; Victor, 2010) is a significant explanator of which states participate in peacekeeping. However, state need does predict the size of a state’s contribution, which refugee flows do not explain well.

In the post-Cold War era, a significant portion of UN peacekeeping activities have become focused on protecting civilians and promoting human rights in the conflicts to which the UN responds (Ban, 2011; Ratner, 1995; DPKO, 2013a,b). I argue that a state’s motivations for participating in these missions may not be humanitarian. These non-humanitarian motivations may have implications for the civilians receiving protection from these states. The goal for these states is to stop the flows of refugees from entering their territory. In order to ensure they succeed in their mission, these states may order their soldiers to use any means necessary to stop flows of people and weapons from crossing their border. They may also simply ignore abuses by their own troops. If the troops policing the inflows feel it necessary to use high levels of force to achieve their objectives, some states may be unlikely to flinch at these techniques. I believe it is important to investigate the effect these motivations have on the human security provided to the peacekept while missions are underway.

The dynamics I present in this article are important to our understanding of international relations, as they highlight the importance of organizational design (Koremenos, Lipson & Snidal, 2001). Recent research has focused on how major powers can force multilateral organizations closer to their ideal policy points (Voeten, 2001) and legitimize their actions through the organization (Thompson, 2009). What this scholarship misses, however, is how, and whether, the major powers can maintain control of their projects once they are implemented by the organization and its smaller members. Since the end of the Cold War, Western powers have worked to focus peacekeeping and peacebuilding missions on democratizing the failing states (e.g. Ratner, 1995). If participants in these missions are focused on their own aims of preventing refugee flows, rather than the UN’s goal of establishing sustainable democratic governments, it is not surprising that many of these missions fail in establishing long-lasting democracies. Furthermore, if organizations like the UN are unable to effectively regulate the behavior of their members on the ground, then states will have an easier time extracting policy concessions from newly formed states, severely threatening the democratic viability of the regime from the beginning (Bueno de Mesquita & Downs, 2006).

My work also provides several avenues for future research. Here, I will focus on the most immediate. Several studies have noted that since the end of the Cold War, more regional organizations have established peacekeeping operations outside of UN authorization (e.g. Bellamy & Williams, 2005; Daniel, Taft & Wiharta, 2008). While this article focuses on UN peacekeeping, it would be fruitful to examine whether refugee flows influence contributions to these missions as well. Some anecdotal evidence suggests that they do. For example, Sierra Leone President Momoh made clear why his country contributed to ECOMOG’s efforts in Liberia: ‘The massive influx of refugees into our country with its attendant economic and social consequences is just one of the many grave responsibilities we are now called upon to shoulder [through ECOMOG]’ (quoted in Gberie, 2005: 57). Similarly during the Liberian Civil War, Obed Asamoah, Ghana’s foreign minister, argued that the heavy economic toll of the refugee situation on Liberia’s neighbors made rapid intervention by ECOWAS an imperative (Adibe, 1997). Given these anecdotes, I encourage scholars to continuing working on the important subject of peacekeeping to unpack fully the dynamics of third-party involvement in war-torn areas.

Footnotes

Replication data

I conducted all analysis using Stata 13. The dataset and do-files for the empirical analysis in this article, along with the online appendix, are located at http://www.prio.org/jpr/datasets and at .

Acknowledgements

I would like to thank Vincent Arel-Bundock, Christian Davenport, Trevor Johnston, Lisa Koch, Shaun McGirr, James Morrow, Paul Poast, Allan Stam, Sonja Starr, Jessica Steinberg, Jana von Stein, Matthew Wells, Alton Worthington, the three anonymous reviewers, and the editor of JPR for their helpful feedback on previous drafts of this article.

Notes

References

Abbott

Kenneth

Snidal

Duncan

(1998) Why states act through formal international organizations. Journal of Conflict Resolution 42(1): 3–32.

Adibe

Clement

(1997) The Liberia conflict and the ECOWAS-UN partnership. Third World Quarterly 18(3): 471–488.

Ban

Ki-moon

(2011) Remarks to Security Council’s open debate on the protection of civilians in armed conflict (https://www.un.org/apps/news/infocus/sgspeeches/search_full.asp?statID=1371).

Beardsley

Kyle

(2011) Peacekeeping and the contagion of armed conflict. Journal of Politics 73(4): 1051–1064.

Beardsley

Kyle

Schmidt

Holger

(2012) Following the flag or following the charter? Examining the determinants of UN involvement in international crises, 1945–2002. International Studies Quarterly 56(1): 33–49.

Bellamy

Alex

Williams

Paul

(2005) Who’s keeping the peace? Regionalization and contemporary peace operations. International Security 29(4): 157–195.

Bellamy

Alex

Williams

Paul

(2013) Providing Peacekeepers: The Politics, Challenges, & Future of United Nations Peacekeeping Contributions. Oxford: Oxford University Press

Bellamy

Alex

Williams

Paul

Griffin

Stuart

(2004) Understanding Peacekeeping. Cambridge: Polity.

Bennett

D Scott

Stam

Allan

(2000) EUGene: A conceptual manual. International Interactions 26(2): 179–204.

10.

Berman

Eric

Sams

Katie

(2000) Peacekeeping in Africa: Capabilities and Culpabilities. Geneva: United Nations Institute for Disarmament Research and Pretoria: Institute for Security Studies.

11.

Blum

Andrew

(2000) Blue helmets from the South: Accounting for the participation of weaker states in United Nations peacekeeping operations. Journal of Conflict Studies 20(1): 1–22.

12.

Bobrow

Davis

Boyer

Mark

(1997) Maintaining system stability: Contributions to peacekeeping operations. Journal of Conflict Resolution 41(6): 723–748.

13.

Bove

Vincenzo

Elia

Leandro

(2011) Supplying peace: Participation in and troop contribution to peacekeeping missions. Journal of Peace Research 48(6): 699–714.

14.

Bueno de Mesquita

Bruce

Downs

George

(2006) Intervention and democracy. International Organization 60(3): 627–649.

15.

Bueno de Mesquita

Bruce

Smith

Alastair

Siverson

Randolph

Morrow

James

(2003) The Logic of Political Survival. Cambridge, MA: MIT Press.

16.

Carter

David

Signorino

Curtis

(2010) Back to the future: Modeling time dependence in binary data. Political Analysis 18(3): 271–292.

17.

Central Intelligence Agency (2013) World Factbook (https://www.cia.gov/library/publications/the-world-factbook/docs/guidetowfbook.html).

18.

Choi

Seung-Whan

(2013) What determines US humanitarian intervention? Conflict Management and Peace Science 30(2): 121–139.

19.

Cingranelli

David L

Richards

David L

Chad Clay

(2014) The CIRI Human RightsDataset, version 2014.04.14 (http://www.humanrightsdata.com).

20.

Daniel

Donald CF

(2011) Partnering for troop supply. International Peacekeeping 18(5): 534–560.

21.

Daniel

Donald CF

Taft

Patricia

Wiharta

Sharon

(2008) Peace Operations: Trends, Progress, and Prospects. Washington, DC: Georgetown University Press.

22.

Diehl

Paul

(2008) Peace Operations. Malden, MA: Polity.

23.

Doyle

Michael

Sambanis

Nicholas

(2000) International peacebuilding: A theoretical and quantitative analysis. American Political Science Review 94(4): 779–801.

24.

Doyle

Michael

Sambanis

Nicholas

(2006) Making War and Building Peace: United Nations Operations. Princeton, NJ: Princeton University Press.

25.

Fortna

Virginia Page

(2008) Does Peacekeeping Work? Shaping Belligerents’ Choices after Civil War. Princeton, NJ: Princeton University Press.

26.

Gberie

Lansana

(2005) A Dirty War in West Africa. London: Hurst.

27.

Ghobarah

Hazem Adam

Huth

Paul

Russett

Bruce

(2003) Civil wars kill and maim people – long after the shooting stops. American Political Science Review 97(2): 189–202.

28.

Ghobarah

Hazem Adam

Huth

Paul

Russett

Bruce

(2004) Comparative public health: The political economy of human misery and well-being. International Studies Quarterly 48(1): 73–94.

29.

Gibler

Douglas

(2009) International Military Alliances, 1648–2008. Thousand Oaks, CA: CQ.

30.

Gleason-Roberts

Megan

(2012) Annual Review of Global Peace Operations 2012. Center on International Cooperation (http://cic.nyu.edu/sites/default/files/gpo_2012.pdf).

31.

Greene

William

(1994) Some accounting for excess zeros and sample selection in Poisson and negative binomial regression models. Working paper EC-94-10. Department of Economics, New York University.

32.

Harff

Barbara

(2003) No lessons learned from the Holocaust? Assessing risks of genocide and political mass murder since 1955. American Political Science Review 97(1): 57–73.

33.

Hultman

Lisa

(2013) UN peace operations and protection of civilians: Cheap talk or norm implementation? Journal of Peace Research 50(1): 59–73.

34.

Hultman

Lisa

Kathman

Jacob

Shannon

Megan

(2013) United Nations peacekeeping and civilian protection in civil war. American Journal of Political Science 57(4): 875–891.

35.

Jakobsen

Peter

(1996) National interest, humanitarianism or CNN: What triggers UN peace enforcement after the Cold War? Journal of Peace Research 33(2): 205–215.

36.

Jenkins, J Craig; Stephen Scanlan & Lindsey Peterson (2007) Military famine, human rights, and child hunger: A cross-national analysis, 1990–2000. Journal of Conflict Resolution 51(6): 823–847.

37.

Kathman

Jacob

(2010) Civil war contagion and neighboring interventions. International Studies Quarterly 54(4): 989–1012.

38.

Kathman

Jacob

(2011) Civil war diffusion and regional motivations for intervention. Journal of Conflict Resolution 55(6): 847–876.

39.

Kathman

Jacob

(2013) United Nations peacekeeping personnel commitments, 1990–2011. Conflict Management and Peace Science 30(5): 532–549.

40.

Koremenos

Barbara

Lipson

Charles

Snidal

Duncan

(2001) The rational design of international institutions. International Organization 55(4): 761–799.

41.

Lebovic

James

(2004) Uniting for peace? Democracies and United Nations peace operations after the Cold War. Journal of Conflict Resolution 48(6): 910–936.

42.

Marshall

Monty

Jaggers

Keith

(2010) Polity IV project: Political regime characteristics and transitions, 1800–2006 (http://www.systemicpeace.org/polity/polity4.htm).

43.

Marten

Kimberly

(2004) Enforcing the Peace: Learning from the Imperial Past. New York: Columbia University Press.

44.

Mueller

John

(2004) The Remnants of War. Ithaca, NY: Cornell University Press.

45.

Murdoch

James

Sandler

Todd

(2002) Economic growth, civil war, and spatial spillovers. Journal of Conflict Resolution 46(1): 91–110.

46.

Murdoch

James

Sandler

Todd

(2004) Civil wars and economic growth: Spatial dispersion. American Journal of Political Science 48(1): 138–151.

47.

Neack

Laura

(1995) UN peace-keeping: In the interest of community of self? Journal of Peace Research 32(2): 181–196.

48.

Perry

Chris

Smith

Adam

(2013) Trends in uniformed contributions to UN peacekeeping: A new dataset, 1991–2012. Providing for Peacekeeping 3: 1–11.

49.

Ratner

Steven

(1995) The New UN Peacekeeping: Building Peace in Lands of Conflict after the Cold War. New York: St Martin’s.

50.

Regan

Patrick

(1998) Choosing to intervene: Outside interventions in internal conflicts. Journal of Politics 60(3): 754–779.

51.

Regan

Patrick

(2002) Civil Wars and Foreign Powers: Outside Intervention in Intrastate Conflict. Ann Arbor, MI: University of Michigan Press.

52.

Russett

Bruce

Oneal

John

(2001) Triangulating Peace: Democracy, Interdependence, and International Organizations. New York: WW Norton.

53.

Saideman

Stephen

William Ayres

(2000) Determining the causes of irredentism: Logit analyses of minorities at risk data from the 1980s and 1990s. Journal of Politics 62(4): 1126–1144.

54.

Salehyan

Idean

(2008) The externalities of civil strife: Refugees as a source of international conflict. American Journal of Political Science 52(4): 787–801.

55.

Salehyan

Idean

Gleditsch

Kristian Skrede

(2006) Refugees and the spread of civil war. International Organization 60(2): 335–366.

56.

Schabas

William

(2000) Genocide in International Law: The Crime of Crimes. Cambridge: Cambridge University Press.

57.

Singer, David

(1988) Reconstructing the Correlates of War dataset on material capabilities of states, 1816–1985. International Interactions 14(2): 115–132.

58.

Snyder

Jack

(2010) Realism, refugees, and strategies of humanitarianism. In: Betts

Alexander

Loescher

Gil

(eds) Refugees in International Relations. Oxford: Oxford University Press.

59.

Thompson

Alexander

(2009) Channels of Power: The UN Security Council and US Statecraft in Iraq. Ithaca, NY: Cornell University Press.

60.

Tir

Jarsolav

Schafer

Philip

Diehl

Paul

Goertz

Gary

(1998) Territorial changes, 1816–1996: Procedures and data. Conflict Management and Peace Science 16(1): 89–97.

61.

United Nations Department of Peacekeeping (2012) Statistics (http://www.un.org/en/peacekeeping/resources/statistics/).

62.

United Nations Department of Peacekeeping (2013a) Human rights (https://www.un.org/en/peacekeeping/issues/humanrights.shtml).

63.

United Nations Department of Peacekeeping (2013b) Protection of civilians (https://www.un.org/en/peacekeeping/issues/civilian.shtml).

64.

United Nations High Commissioner on Refugees (2012) UNHCR statistical database (http://data.un.org/Data.aspx?d=UNHCR&f=indID%3AType-Ref#UNHCR).

65.

Victor

Jonah

(2010) African peacekeeping in Africa: Warlord politics, defense economics, and state legitimacy. Journal of Peace Research 47(2): 217–229.

66.

Voeten

Erik

(2001) Outside options and the logic of security council action. American Political Science Review 95(4): 845–858.

67.

Western

Jon

(2002) Sources of humanitarian intervention: Beliefs, information, and advocacy in the US decisions on Somalia and Bosnia. International Security 26(4): 112–142.

68.

White House (1994) Office of the press secretary, selected remarks prepared for delivery by President William Jefferson Clinton on US policy toward Haiti, 15 September (http://www.presidency.ucsb.edu/ws/?pid=49093).

69.

World Bank (2012) Data catalog (http://data.worldbank.org/data-catalog/).

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB