The ability of international actors to help stabilize conflict-affected states is critically important for protecting civilians and maintaining global security. Civil wars cause many more casualties than interstate conflicts, but ending them, especially through settlements, is difficult (Walter Reference Walter2002). On issues such as human rights, electoral integrity, and international trade, external actors can help enforce agreements between domestic actors (Donno Reference Donno2013; Hafner-Burton Reference Hafner-Burton2013; Keohane Reference Keohane1984; Matanock Reference Matanock2020; Simmons Reference Simmons2009; Simmons Reference Simmons2010). Scholars and policymakers see a similar role for the international community in securing settlements to civil conflicts.
While the international community engages in various forms of intervention, peacekeeping after civil conflict is one of the most invasive, and operations led by the United Nations (UN), in particular, have increased in frequency and intensity (Fortna and Howard Reference Fortna and Howard2008). There is an emerging consensus that this type of intervention “works” in improving prospects for post-conflict peace (Doyle and Sambanis Reference Doyle and Sambanis2006; Fortna Reference Fortna2008; Gilligan and Sergenti Reference Gilligan and Sergenti2008; Gizelis and Benson Reference Gizelis and Benson2019; Walter Reference Walter2002). Existing work shows that third-party involvement can help combatants overcome the challenge of credibly committing to a negotiated settlement (Walter Reference Walter2002).
Despite a rich literature on peacekeeping, how interveners mitigate these commitment problems and change incentives remains unclear—and is crucial to understand (Walter, Howard, and Fortna Reference Walter, Howard and Fortna2019). Some studies suggest that peacekeeping missions primarily work by threatening military force (for example Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2014; Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2016; Walter Reference Walter1997), while other research points to alternative instruments (for example, Fortna Reference Fortna2008; Howard Reference Howard2019).Footnote 1 Overall, however, reviews of the peacekeeping literature explicitly note that it is “inconclusive” about the role military force plays in these operations (Fortna and Howard Reference Fortna and Howard2008, 292, 295). Recent work that seeks to test the use of different peacekeeping instruments is qualitative in nature and relies on select case studies (Howard Reference Howard2019).
This article specifies a theory of incentive change and quantitatively tests potentially critical but different peacekeeping instruments using new cross-national data. Building on the existing literature, we take credible commitment problems to be a major barrier to securing post-conflict peace, and one that peacekeeping missions—which rely on multilateral coordination between the UN and its partners, including member states and other international actors—seek to directly address. We identify two main instruments that the UN and its partners use to change the incentives of domestic actors to credibly commit to comply with a settlement: “conditional incentives” (CI) and “military coercion.” We specify the former “soft” peacekeeping instrument—where interveners condition incentives such as foreign aid on combatants' settlement compliance—and contrast it with the latter “hard” peacekeeping instrument—where interveners enforce compliance through the threat or use of military force. With few exceptions, peacekeeping missions tend to use information to identify compliance, but they use different instruments or tools to change incentives to comply. We assess whether these instruments are effective at preventing conflict recurrence using data we collected on UN peacekeeping missions' instruments in post-conflict peace periods from 1989 to 2012. We use a matching analysis to assess their relationship with the durability of post-conflict peace, following and extending the influential work of Gilligan and Sergenti (Reference Gilligan and Sergenti2008). We show that, contrary to conventional wisdom, military coercion is neither commonly used nor necessary for peace. Peacekeeping missions often rely on conditioning economic, legal, and political incentives on settlement compliance. Our study is not meant to be a conclusive analysis of all peacekeeping instruments, or the conditions under which they are effective—objectives for future studies—but an important step in theorizing and testing how peacekeeping missions broadly work.
As international involvement in intrastate conflict continues to expand, this research has important implications for scholars and practitioners. Global efforts to prevent the recurrence of civil wars have remained “strikingly deficient,” according to UN officials (New York Times 2014). Better understanding peacekeeping instruments is still essential for improving prospects for stable peace. Moreover, our finding that CI can reduce the risk of repeat conflict suggests that under certain conditions, peacekeeping may be able to succeed without incurring the casualties or other costs associated with military coercion. Indeed, most post-conflict countries receive considerable foreign assistance that could be leveraged to enforce compliance with peace processes. While deploying peacekeeping missions composed of large numbers of armed troops may, at times, be crucial to protect civilians or prevent the spread of violence in the midst of active conflict (Beardsley and Gleditsch Reference Beardsley and Gleditsch2015; Carnegie and Mikulaschek Reference Carnegie and Mikulaschek2020; Fjelde, Hultman, and Nilsson Reference Fjelde, Hultman and Nilsson2019; Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2019), our results suggest that threatening force is not necessary to secure peace between combatants in post-conflict contexts. This article contributes to a wider understanding of how international intervention works, while opening new policy avenues for its use.
Instruments to Keep Peace Following Civil Conflict
Scholars criticize international peacekeeping at times, including for being counterproductive (Diehl, Reifschneider, and Hensel Reference Diehl, Reifschneider and Hensel1999; Toft Reference Toft2009), failing to reach its potential (Autesserre Reference Autesserre2014), or struggling to improve governance (Lake and Fariss Reference Lake and Fariss2014). Generally, however, many studies have established that these interventions can improve prospects for post-conflict peace, as civil war recurrence is less likely when peacekeeping missions are present (for a comprehensive review of this literature, see Walter, Howard, and Fortna Reference Walter, Howard and Fortna2019). Yet, exactly how peacekeeping works remains under debate (Fortna and Howard Reference Fortna and Howard2008, 292; Walter, Howard, and Fortna Reference Walter, Howard and Fortna2019).Footnote 2 We build on existing work but fully specify two different peacekeeping instruments and test them. The first is military coercion, which depends on the threat or use of force. Initial evidence from existing studies, however, questions the extent to which peacekeeping missions actually employ this instrument. The second is conditional incentives (CI), which depend on economic, legal, and political benefits that can be withdrawn from former combatants. We describe these instruments after reviewing commitment problems and the roles of peacekeeping missions outlined in the existing literature.
In post-conflict settings, most studies focus on the role of international actors in ameliorating commitment problems faced by combatants following civil wars (Walter Reference Walter2002), though there are multiple pathways to peace. Since conflict is often costly, combatants can typically identify mutually beneficial deals. However, commitment problems occur when one side becomes even temporarily weaker during a peace process, thereby incentivizing their opponent to take advantage to gain benefits. Concerns about these implementation problems—even suspecting a violation or misperceiving unintentional acts as voluntary noncompliance—can derail an agreement (see, for example, Fearon and Laitin Reference Fearon and Laitin2007; Toft Reference Toft2009; Walter Reference Walter2002). Commitment problems can cause combatants to return to conflict by encouraging surprise attacks to avoid early warnings, preemptive attacks due to concerns over noncompliance, or retributive attacks when an opponent alters aspects of the deal (Fortna Reference Fortna2008, 82–5).
The commitment problem is therefore an informational problem and also, especially, an incentive problem. To solve it, all sides must know noncompliance will be identified and sanctioned, so that complying becomes more beneficial than defecting.Footnote 3 In terms of information, peacekeepers can reduce distrust by clarifying all parties' obligations, investigating perceived noncompliance, and notifying all parties of their findings. In other words, the presence of peacekeeping missions can reveal credible information about each side's resources, intentions, and behavior, potentially reducing conflict recurrence by overcoming information asymmetries but also by identifying noncompliance that can otherwise contribute to commitment problems (see, for example, Fortna Reference Fortna2008; Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2016; Lindley Reference Lindley2007; Mattes and Savun Reference Mattes and Savun2010; Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2017).
Yet, if we take commitment problems seriously—both as a major reason for conflict recurrence in post-conflict settings, and as a challenge many peacekeeping missions try to address—then securing peace requires identifying and changing combatants' incentives for noncompliance. Peacekeeping missions must make it costlier to defect than to comply with a settlement through credible conditionality. This is particularly important since commitment is most problematic when relative capabilities shift post-conflict, at least temporarily, as each side's fighting capabilities are translated into power in state institutions (Walter Reference Walter2009, 258–9). Peacekeeping missions therefore require a form of conditionalityFootnote 4 to change these incentives and overcome these commitment problems (Fortna Reference Fortna2008).
This question of how peacekeeping missions can effectively provide conditionality is receiving more attention but remains debated. Recent studies that examine the impact of peacekeeping at the subnational level often examine a combination of different instruments or underlying mechanisms without differentiating the effect of each on conflict avoidance or reduction. For instance, Ruggeri, Dorussen, and Gizelis (Reference Ruggeri, Dorussen and Gizelis2017) attribute the conflict-reducing effects of peacekeeping deployments to a blend of coercive and non-coercive instruments that potentially help combatants overcome commitment problems. Another set of studies explore the effect of peacekeeping missions on civilian casualties, finding some mixed effects and attributing the results to different instruments (for example, Fjelde, Hultman, and Nilsson Reference Fjelde, Hultman and Nilsson2019; Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2013; Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2019; Hunnicutt and Nomikos Reference Hunnicutt and Nomikos2019). Some cross-national research similarly takes a broad perspective that considers but does not fully distinguish between peacekeeping mechanisms (for example, Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2016; Mattes and Savun Reference Mattes and Savun2009). These studies join theoretical work that suggests a variety of possible pathways through which peacekeeping missions can secure peace (for example, Fortna Reference Fortna2008). Recent work by Howard (Reference Howard2019) uses several in-depth case studies of peacekeeping to propose persuasion, inducements, and coercion as possible instruments.Footnote 5 This article builds on that work and explores how aspects of the last two instruments address commitment problems.Footnote 6
Drawing on this research, we specify two main instruments through which peacekeeping missions are likely to change incentives for compliance with peace processes: military coercion and CI. Both require information to identify compliance—but they use different tools to change incentives for compliance based on that information.
For both instruments, multilateral coordination is a central component. To implement peacekeeping missions, the UN relies on its partners—member states and other bilateral and multilateral organizations—for personnel and resources that can be conditioned on compliance; these actors together comprise the “peacekeeping missions” that we examine. The UN coordinates the recruitment and deployment of troops, police, and other personnel from member states, but some impose restrictions on what their personnel can do. Japanese peacekeepers, for example, have not been permitted to engage in combat operations, and other countries restrict their personnel to safer operating areas or keep them beholden to the commands of their home governments (see, for example, Bove and Ruggeri Reference Call, Stedman, Rothchild and Cousens2016, 685–7; Ishizuka Reference Ishizuka, Bellamy and Williams2013, 408–9; Polman Reference Polman2003, 40, 52; Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2018, 7). The UN also coordinates the provision of resources to states that receive peacekeeping missions, both by distributing foreign assistance through its agencies directly and by working with “Groups of Friends” of the Secretary-General—which are especially interested donor governments and international organizations (for example, the European Union)—to shape a country's receipt of aid and other benefits, such as membership in certain organizations (see, for example, Whitfield Reference Whitfield2007). Most peacekeeping missions do eventually receive personnel, though partners do not send as many individuals as quickly as the UN prefers (see, for example, Lundgren, Oksamytna, and Coleman Reference Lundgren, Oksamytna and Coleman2020; Passmore, Shannon, and Hart Reference Passmore, Shannon and Hart2018). Likewise, most of the contexts to which missions are sent garner substantial foreign assistance, as countries recovering from conflict tend to receive more aid than similar, non-conflict-affected countries (Collier and Hoeffler Reference Collier and Hoeffler2004, 1136).
While multilateral coordination is not the central focus of this article, we argue that UN peacekeeping missions rely on this coordination to mobilize personnel and resources, making it a critical feature of these missions regardless of what instruments they use to incentivize compliance. Others have examined the challenges of coordination in peacekeeping, including specifically in mobilizing troops (for example, Lundgren, Oksamytna, and Coleman Reference Lundgren, Oksamytna and Coleman2020; Passmore, Shannon, and Hart Reference Passmore, Shannon and Hart2018). However, generally, the UN plays a pivotal role not only in monitoring the behavior of combatants, but also in leveraging its ties to member states and other organizations to develop, deploy, and ensure support for peacekeeping missions. In contrast to domestic civil society—which is often weak and polarized at the end of civil wars, limiting its ability to monitor and incentivize compliance (Wantchekon Reference Wantchekon2004, 17, 27)—or other foreign actors—which do not require the same buy-in from multiple stakeholders (Dobbins et al. Reference Dobbins2005, 243–5; Kreps and Wallace Reference Kreps and Wallace2009; Osborn Reference Osborn2013, 48–50)—the UN is more capable of exercising oversight and shaping a response that is less partisan to the particular conflict. This is one reason why the UN deploys the most peacekeeping missions globally (Mullenbach Reference Mullenbach2013). We now turn to the two primary instruments these missions use to incentivize compliance with peace processes.
Military Coercion
Some studies—including Walter (Reference Walter1997, 340–1) and Hultman, Nygård, and Hegre (Reference Hultman, Nygård and Hegre2016)—suggest peacekeeping missions work by threatening military force. In this conception, peacekeeping missions recruit troops that can use military coercion to provide security guarantees to each side, which raises the cost of aggression, so that a secure settlement can then offer combatants a path to survival and an opportunity to achieve their political goals. Rebels are usually required to disarm more extensively than the government, so they may especially require guarantees from peacekeeping missions that have the capability to protect and the resolve to sanction government noncompliance (Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2019). The threat of force by peacekeepers, rather than its use, may shape combatants' behavior (Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2013, 389; Fjelde, Hultman, and Nilsson Reference Fjelde, Hultman and Nilsson2019), but if deterrence fails, force must be employed to make enforcement credible (Fortna Reference Fortna2008, 87–9, 102; Schelling Reference Schelling1966; Walter Reference Walter2002).
While the existing literature often emphasizes this military coercion instrument, many empirical studies indicate that peacekeeping missions tend not to have access to it. The high costs of troops can render the sustained application of military methods untenable, diminishing mission credibility (Fortna Reference Fortna2008; Gilligan and Sergenti Reference Gilligan and Sergenti2008). Even when troops are deployed, they may be hesitant to risk casualties or provoke combatants. Many missions are allowed to employ force only to protect themselves or civilians, not to pursue combatants, supply “compellence-based” security guarantees, or seek the military defeat of spoilers or other noncompliers (Howard Reference Howard2019, 21, 186; United Nations 2008, 34–5). Military coercion may also be disproportionate to the violation; in general, UN actions are supposed to be “calibrated in a precise, proportional and appropriate manner” (United Nations 2008, 35). Thus, the troops that are deployed after fighting has stopped are generally not enforcement missions—instead, they typically serve primarily as “referees,” supplying information and coordinating among different actors (Kjeksrud and Ravndal Reference Kjeksrud and Ravndal2011, 9–11; United Nations 2008, 19, 31–5). In post-conflict contexts, deploying troops who seek to credibly threaten force is especially uncommon (Fortna Reference Fortna2008, 88; United Nations 2008, 34–5). While the UN could use a mission without a mandate for force as a “trip wire” for the use of force should violations be detected (Fortna and Martin Reference Fortna and Martin2009), the evidence does not support it: “the international community has not responded to the tripping of a peacekeeping wire with forceful intervention” (Fortna Reference Fortna2008, 80, 88).
The pervasiveness of military coercion is therefore questionable—and existing work has struggled to assess its use. Walter (Reference Walter2002), Doyle and Sambanis (Reference Doyle and Sambanis2006), and Fortna (Reference Fortna2008) find that “traditional” peacekeeping—observation and verification missions authorized under Chapter VI of the UN Charter, which requires receiving host-state consent—is equally effective in reducing conflict recurrence compared to Chapter VII peace enforcement missions. However, distinguishing only between Chapter VI and VII overlooks extensive variation in the tools available to, and employed by, different operations.Footnote 7 Other work attempts to gauge the impact of different types of peacekeeping by focusing on the constitution of a mission, finding that increases in personnel, particularly troops, are associated with greater cooperation, fewer civilian casualties, and less conflict recurrence (Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2016; Maekawa, Arı, and Gizelis Reference Maekawa, Arı and Gizelis2019; Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2013).Footnote 8 However, this may just be driven by the overall size of the mission.Footnote 9 For instance, Hultman, Nygård, and Hegre (Reference Hultman, Nygård and Hegre2016) find that mission size and troop presence are highly correlated, so it is not clear that military coercion specifically is driving their results.Footnote 10 In many cases, troops carry weapons but do not employ or even threaten force. Indeed, while peacekeeping has become more militarized over time, it often reflects group-preserving decision making among permanent members of the UN Security Council, not increasing reliance on military coercion (Howard and Dayal Reference Howard and Dayal2018). Analysts of UN missions have long noted that “the use of military contingents in a peacekeeping mission is wrongly associated with the use of force” (IPS 1997, 72).
To further explore this instrument, we examine the mandates of the UN missions dispatched to post-civil war contexts from 1946 to 2012.Footnote 11 We find that only 51 per cent (18 of 35) were actually authorized to employ force under their UN mandates.Footnote 12 The evidence also suggests that even when peacekeeping missions are authorized, they often elect not to employ force. One UN study of eight peacekeeping operations—mostly robust peace enforcement missions—found that peacekeepers intervened in only 20 per cent of reported combatant attacks on civilians (101 of 507), despite being authorized to do so (UN General Assembly 2014). When peacekeepers did respond, “a show of force to deter the progress of actual or intending attackers” was rare (UN General Assembly 2014, 8, 21).
This casts doubt on the notion that peacekeeping missions in post-conflict settings primarily provide credible conditionality through military coercion. We therefore specify another peacekeeping instrument that can act as a complement to, or a substitute for, military coercion in changing incentives.
Conditional Incentives (CI)
We specify a CI instrument for altering incentives by conditioning economic, political, and legal incentives on compliance during peace processes. To overcome commitment problems, combatants—particularly leaders who negotiate and sign settlements—must anticipate conditionality on their compliance. Yet, leveraging military coercion is often so costly that combatants may not expect outside actors to use or sustain it. Withdrawing aid or enacting similar punishments is often more proportionate to many violations of peace settlements that tend to be political in nature, and, for many missions, these actions are not as costly as military coercion, so they can significantly raise combatants' expectations of international enforcement. We describe how CI are supplied and the effects they have in specific cases before empirically testing this instrument alongside military coercion.
In peacekeeping missions, verification on the ground can be combined with external assistance to reward compliance and punish noncompliance. Post-conflict contexts often feature UN “Groups of Friends” (initially, “Friends of the Secretary-General”), as described earlier, which are informal collections of interested states that complement international troop deployments by supplying technical expertise and humanitarian, development, and peacebuilding resources (Whitfield Reference Whitfield2007, 3–4). Since the end of the Cold War, donor governments and international organizations have often offered this assistance, so conditioning it on domestic actors' behavior tends not to raise costs for the UN and its partners. The employment of UN Groups of Friends during peacekeeping missions increased sevenfold over the 1990s (Whitfield Reference Whitfield2007, 4), when donor-funded economic development and democracy and governance assistance also expanded—which was typically conditioned on compliance with constitutional rules and procedures by recipients (Bjornlund Reference Bjornlund2004, 24; Carothers Reference Carothers1999, 6, 85).Footnote 13 While early research suggested that donors might be unwilling to cut off aid—the so-called “Samaritan's Dilemma” (for example, Buchanan Reference Buchanan and Phelps1975)—systematic reviews show conditionality is often enacted (Wright and Winters Reference Wright and Winters2010). In post-conflict settings, foreign aid tends to be “the main repository of ‘sticks’ and ‘carrots’ to keep a peace process on track” for all parties (Arnault Reference Arnault2006, 12; see also Cil and Huth Reference Cil and Huth2019). Yet, most conflict-related research focuses on the “sticks,” such as economic sanctions, and pays less attention to the “carrots” (Regan and Aydin Reference Regan and Aydin2006). However, aid for all parties—including incentives formally provided in exchange for disarmament, demobilization, reintegration, reconstruction, and party building—can make compliance more beneficial and violations less so (Cil and Huth Reference Cil and Huth2019; Flores and Nooruddin Reference Flores, Nooruddin, Coyne and Mathers2011; Matanock Reference Matanock2017a; Matanock Reference Matanock2020). The tools that these peacekeeping missions use in post-conflict contexts, then, may be economic (foreign aid and sanctions), political (electoral incentives and diplomatic threats), or legal (accession processes).
While the existing literature focuses less on this CI instrument compared to military coercion, case studies illustrate how some peacekeeping missions have used it to change the incentives of governments and rebels. Some of the “early” UN peacekeeping missions, for instance, intentionally avoided force but used CI (Miller Reference Miller2013). During El Salvador's 1990s peace process, the government failed to register voters, mainly among rebel supporters, tipping the balance of power; in response, UN peacekeepers verified the complaints and the US froze disbursement of US$70 million in aid—pressure that ultimately forced the government to comply (Fortna Reference Fortna2008, 90; Howard Reference Howard2008, 94, 115–16; LeoGrande Reference LeoGrande, Grant, Nijman and NY1998, 108). Aid provided by the UN and its partners has also been useful for changing rebels' incentives: rebels in Mozambique considered reneging on the 1992 Rome Accords after losing post-conflict elections, but they decided to remain peacefully engaged because the international community established a substantial trust fund for political party development conditioned on compliance (Turner, Nelson, and Mahling-Clark Reference Turner, Nelson, Mahling-Clark and Kumar1998, 72, 161–4). CI are not unique to smaller peacekeeping missions, and they may help explain the success of larger operations, even those mandated to use force. The United Nations Transitional Authority in Cambodia (UNTAC), for example, was largely effective in preventing further conflict, but its authority “never derived from UNTAC's ability to coerce [with force],” despite its large number of troops, and was based instead on its ability to provide information and “deliver on certain [non-military] promises” (Howard Reference Howard2008, 177–8).
While many missions rely on economic tools, political tools, such as electoral incentives, are also common. Returning to El Salvador, after an explosion at an illegal arms cache was attributed to the rebels, the government suggested that their party be banned from registering to participate in elections until it disarmed as agreed, and the UN Secretary-General criticized the rebels and, notably, did not contradict the threat, while the US State Department also discussed this possibility—being kept from participating in elections pushed the rebels to immediately reveal and destroy other weapons caches, and even to deem some of their candidates as “too damaged to run” in the elections (UN Doc S/26005, June 29, 1993, and interviews cited in Call Reference Call, Stedman, Rothchild and Cousens2002, 560; Howard Reference Howard2008, 117–18; “FMLN Legalization—Contingent on Demobilization,” 1992, 1–2; Stanley and Holiday Reference Stanley and Holiday1997, 27, 32). This approach of using electoral incentives has been widespread when both government and rebel parties participate in elections (Matanock Reference Matanock2017b). In Guatemala, for example, the UN established a trust fund and coordinated direct support from organizations, such as the Organization of American States (OAS) and its Group of Friends. Some support, especially from leftist Spanish foundations, provided specific assistance for the political process and for each of the parties, and the broader mission made clear that these benefits were conditional on compliance with the peace accords (Azpuru et al. Reference Azpuru2004, 24–30; Heard Reference Heard1999, 25; Stanley and Holiday Reference Stanley, Holiday, Rothchild, Stedman and Cousens2002, 32; Verstegen Reference Verstegen2000, 62).
Finally, to illustrate the use of legal tools, in Croatia, the United Nations Transitional Administration for Eastern Slavonia, Baranja and Western Sirmium (UNTAES)—a highly complex and heavily militarized operation—convinced its partners to stake International Monetary Fund (IMF) loans and membership in the North Atlantic Treaty Organization (NATO) and the European Union (EU) on adherence to the peace process. Then, using peacekeepers to monitor implementation, UNTAES was able to “achieve compliance … in the sense that Croatia subsequently refrained from taking unilateral military action, as it had done previously despite the presence of [prior peacekeeping missions]” (Remmert Reference Remmert2017, 22).
These examples show how the prospects of being denied aid, the ability to run candidates in elections, or membership in beneficial organizations can be sufficient to overcome incentives for noncompliance with peace settlements. Losing resources or access to political power may not be as costly as facing violence, but these cases demonstrate that they were, at least in some instances, costly enough to encourage former rebels or governments that began violating a settlement to change course. Indeed, due in part to the appropriate proportionality and ease of implementing CI, international officials overseeing peacekeeping missions have emphasized the importance of employing them (for example, Soto and Castillo Reference De Soto and Del Castillo1994). This instrument also builds on existing work showing that the international community often uses conditionality when providing foreign aid in post-conflict contexts, especially around elections, to influence the behavior of domestic actors (Cil and Huth Reference Cil and Huth2019; Fortna Reference Fortna2008, 89–93, 102; Girod Reference Girod2012; Matanock Reference Matanock2017a; Matanock Reference Matanock2020).Footnote 14
Empirical Implications
Hypothesis 1: If military coercion is effective in prolonging peace, then post-conflict environments that receive peacekeeping missions mandated to employ military coercion should experience less conflict recurrence than other cases.
Hypothesis 2: If CI reduce conflict recurrence, then post-conflict environments that receive peacekeeping missions that employ CI should experience less conflict recurrence.
We evaluate the efficacy of each peacekeeping instrument independently, as they potentially complement each other. We therefore mainly focus on the efficacy of each instrument in prolonging peace compared to similar post-conflict contexts without peacekeeping missions, though we also compare the instruments. We use cross-national data on post-conflict peace periods, assessing UN missions—the most prevalent type of peace operation—with different mandates to test these hypotheses.
A common methodological problem in quantitatively studying peacekeeping is that missions are not randomly assigned. Previous research has shown that peacekeeping missions are sent to areas most likely to experience recurrence (Fortna Reference Fortna2008; Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2018), but mission coerciveness or conditionality may not similarly select the “hardest” cases. While there exists no perfect remedy, we adopt several methods to deal with these concerns as rigorously as possible by assessing selection models of the deployment of particular types of peacekeeping and by employing matching (following Gilligan and Sergenti [Reference Gilligan and Sergenti2008] and Ruggeri, Dorussen, and Gizelis [Reference Ruggeri, Dorussen and Gizelis2017]).
Data and Measures
Dependent Variable
We use an existing dataset of post-conflict peace periods that we expand temporally. Gilligan and Sergenti (Reference Gilligan and Sergenti2008) specify a country-level unit of analysis with the dependent variable as the number of months until conflict recurs after an intrastate conflict that ended between 1989 and 2003.Footnote 15 We extend the data from 2003 to 2012. This increases the peace periods from 87 to 118 and those that received peacekeeping from 19 to 30.
We also code whether combatants reached a settlement as the peace period began, based on the Uppsala Conflict Data Program (UCDP) Peace Agreement Dataset (Högbladh Reference Högbladh, T and Themnér2011).Footnote 16 Of the 118 peace periods in our data, 43 featured settlements (36 per cent): 23 were among the 30 cases that received peacekeeping missions (77 per cent), and 20 were among the 88 cases that did not (23 per cent). Previous studies suggest peace periods resulting from settlements are more fragile than those following military victories because commitment problems are more intense (for example, Toft Reference Toft2009). Therefore, given that more peacekeeping cases in our data end through peace agreements—which are more likely to fail—examining all peace periods should bias against the effect of peacekeeping.Footnote 17 We also examine the content of these settlements—whether they include power sharing, security sector reform (SSR), and disarmament and demobilization (DDR) provisions—and other controls to assess balance across mission types (see Online Appendix 1.6).
Independent Variables
To explain variation in the duration of peace periods, we follow Gilligan and Sergenti (Reference Gilligan and Sergenti2008) in focusing on UN peacekeeping, as opposed to other types of intervention, for several reasons. First, UN peacekeeping missions have been the most widely deployed (Mullenbach Reference Mullenbach2013), and since they are more often empowered to use military coercion than regional intergovernmental organizations,Footnote 18 successful reliance on CI would be particularly surprising and have broader policy implications. Second, as discussed earlier, the UN is uniquely equipped to both provide credible information and elicit conditional responses—using its own tools and shaping the response of its partners—so its missions are therefore a suitable first test of the CI instrument. We do, however, control for the presence of non-UN peacekeeping missions (described later).
CI
For our first independent variable of interest, we coded UN peacekeeping missions' use of CI. Our coding was based on evidence of any economic, legal, or political “threatened or imposed” punishments or “promised or granted” rewards tied to combatants' compliance with the terms of a peace settlement. This builds on the coding of conditionality developed by Donno (Reference Donno2013, 203–4) regarding adherence to electoral rules.Footnote 19 If there were elections held during a peace period with peacekeeping missions, we referred to Donno's coding of conditionality. However, since some cases of post-conflict peacekeeping—approximately a third of peace periods in our sample—did not experience elections, we drew on qualitative evidence from individual cases to provide a broader coding of conditionality.Footnote 20 Using evidence from UN reports, UN Security Council resolutions, case studies of individual missions, and news databases, we coded each case “1” if the reports indicated evidence of conditionality on incentives from the UN or its donor partners, and “0” otherwise.
Conditionality can take different forms, as we illustrated earlier, but the measures we encountered while coding primarily included economic instruments, such as UN trust funds and reconstruction aid from Groups of Friends and other foreign donors working alongside the UN. For example, the United Nations Observer Mission in Angola (MONUA) used the threat of aid withdrawals and the enticements of trust funds to help peacekeepers “steer demobilization [of combatants] to completion” (Howard Reference Howard2008, 198). In Bosnia, donors linked the agreed-on return and protection of displaced civilians and cooperation with war crimes tribunals to economic assistance—and relied on peacekeepers to verify compliance (Vayrynen Reference Vayrynen and Cortright1997, 158). In Cote D'Ivoire, the UN Security Council threatened to impose sanctions against individuals “obstructing the work of ONUCI [the UN mission]” (BBC 2006).
Coding data on conditionality for peacekeeping is a primary contribution of this article.Footnote 21 We required that CI be enacted in response to UN monitoring and verification, which could, if anything, underestimate the effectiveness of CI because donors may instead employ conditionality in response to their own information, for instance. Similarly, in some cases, conditionality may be implicit rather than explicit. Our measure may therefore undercount the use of these incentives, which would likely bias against our findings: in many other instances, implicit conditionality may have deterred noncompliance. CI do not appear to be driven by a case's propensity for peace—something we examine more systematically later. Reward and punishment were promised for compliance and noncompliance, and our coding suggests that conditionality was invoked based on ex-combatant behavior. There does not seem to be a selection effect whereby conditionality occurs only in instances without expected violations (see Online Appendix 2.3b). Moreover, a settlement did not automatically mean CI: it was employed in 60 per cent of peace periods where combatants reached settlements.Footnote 22 Nonetheless, this correlation between peace agreements and CI should underestimate CI's effectiveness because settlements are less stable than military victories.
Military Coercion
For our second independent variable, we coded military coercion by assessing whether a peacekeeping mission was authorized to use or threaten military force, based on: (1) the mandate's language (drawing on Franke and Warnecke Reference Franke and Warnecke2009); (2) whether it was a Chapter VI or Chapter VII mission (from Doyle and Sambanis Reference Doyle and Sambanis2006; Fortna Reference Fortna2008); and (3) the primary purpose of the mission (drawing on Mullenbach Reference Mullenbach2013). If any of these criteria were met, we coded the mission as authorized for military coercion. This coding is more precise than previous attempts to distinguish between mission types, particularly because it goes beyond the “Chapter VI” versus “Chapter VII” distinction. We coded all Chapter VII missions as authorized for military coercion but also eleven Chapter VI missions because either the mandate language or mission purpose suggested its possible use.
Despite our skepticism that force is used even under Chapter VII missions, if there is any possibility for a UN mission to employ it, we code it as having a mandate for military coercion. This means that we code as military coercion many successful cases of peacekeeping that may really operate through CI to ensure any bias works against our argument. This approach is a generous estimate of military coercion because at least some missions may only be allowed to use force to protect UN personnel or civilians. As a robustness check, we also coded as military coercion only missions with Chapter VII authorization.Footnote 23 Since peacekeeping is authorized by the UN Security Council, changes to missions' mandates typically require authorization of a “new” operation, so coding by mission captures temporal variation in military coercion.
Analysis
We evaluate the empirical implications of each instrument by examining whether peacekeeping missions employing one or both instruments are associated with reductions in conflict recurrence. We primarily compare post-conflict peace periods that experienced different missions—those that did and did not employ CI, and those with and without mandates for military coercion—to peace periods that received no peacekeeping. We also compare outcomes in periods that received different missions, but the number of cases is small, so the results must be interpreted cautiously.Footnote 24 We employ several techniques to try to account for selection effects.
Descriptive Statistics
In our sample, 53 per cent of UN missions (16 of 30) employ CI, which is the same as the percentage of missions with mandates for military coercion (see Table 1). However, CI is nearly twice as common in noncoercive compared to coercive missions, suggesting CI and military coercion often act as substitutable tools.Footnote 25 This is notable given our generous coding of military coercion. Many missions coded as employing CI but not military coercion are so-called “observer” missions comprised of unarmed peacekeepers.
Mandates for military coercion and CI in post-conflict peacekeeping missions, 1989–2012
Note: P-value (Fisher's exact): 0.08.
Cross-tabulations of conflict recurrence rates, using a binary indicator of peace failure at any point in the peace period, suggest cases with CI and without military mandates both have statistically significantly lower rates of recurrence than those without peacekeeping missions (see Table 2). The same is true in comparing cases that received peacekeeping missions with CI and missions without CI (not shown; Fisher's exact = 0.03). The recurrence rate for post-conflict contexts receiving missions without CI is statistically the same as those receiving no missions (71 versus 67 per cent).
Conflict recurrence rates in all post-conflict peace periods, 1989–2012
Since military coercion and CI may act as complementary tools in some cases, we also examine categories of each combination of peacekeeping type: CI missions with mandates for military force (for example, the United Nations Mission in Bosnia Herzegovina [UNMIBH]); CI missions without mandates for military coercion (for example, the United Nations Observer Mission in El Salvador [ONUSAL]); non-CI missions with military coercion (for example, the United Nations Stabilization Mission in Haiti [MINUSTAH]); and non-CI, noncoercive missions (for example, the United Nations Mission for the Referendum in Western Sahara [MINURSO]).Footnote 26 While the sample sizes are small, cross-tabulations show peace periods receiving peacekeeping missions with CI—regardless of whether they were also mandated to use military coercion—tended to experience less recurrence than periods that did not experience peacekeeping (see Table 3). The results are statistically significant. Again, the recurrence rates for missions without CI were not significantly different than non-peacekeeping cases.
Conflict recurrence rates in all post-conflict peace periods, 1989–2012
Notes: The total number of peacekeeping cases in this table adds up to 32 instead of 30 because two cases (peace periods) are double counted—Croatia and Sierra Leone—since each one experienced two separate UN missions (including a coercive mission and a non-coercive one) in the same peace period. PKO = peacekeeping operation.
Selection Models
It is possible, however, that peacekeeping missions employing particular instruments are not distributed randomly. If CI missions are sent to post-conflict contexts where peace is easier to secure, then the associated reductions in conflict recurrence may not be driven by the use of that tool. While not the focus of this article, we identified possible correlates of peacekeeping deployments and ran selection models to assess whether they predict the deployment of particular types of peacekeeping compared to no peacekeeping. We also checked for correlate balance across mission types, including CI versus no CI, and military coercion versus no coercion.Footnote 27
Our correlates included those identified by Gilligan and Sergenti (Reference Gilligan and Sergenti2008) as potentially affecting conflict recurrence: battle deaths from the previous war (a measure of conflict intensity); the war's duration; a country's degree of ethnic fractionalization, population size, mountainous terrain, number of military personnel, level of democracy, gross domestic product (GDP) per capita before the war, and regional indicators. Previous studies have shown that these variables can influence peace durability and peacekeeping deployments, but we also incorporated a series of additional correlates to control for other confounding factors (Doyle and Sambanis Reference Doyle and Sambanis2006; Fortna Reference Fortna2008; Toft Reference Toft2009).Footnote 28 These included: the outcome of the previous conflict (examining peace agreements and outright military victories by one side); whether elections occurred during the peace period; whether the previous war was territorial, Marxist, or ethnic; combatants' relative capabilities; the number of rebel groups involved; whether peacekeeping missions were also sent during the conflict; and whether a non-UN mission—for example, led by the African Union—deployed during the peace period.Footnote 29 We also examined whether a peace agreement included provisions such as power sharing and autonomy. We looked at: how much military and economic aid a country received; its alliances; Groups of Friends; and colonial history. Finally, we included a time period indicator for whether a peace period started before 1995, during 1995–2001, or after 2001.Footnote 30
The selection models and balance tests, discussed in Online Appendix 1.6, suggest missions either with CI or mandates for military coercion are more likely following conflicts that are harder to settle, similar to findings for peacekeeping generally (Fortna Reference Fortna2008; Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2018). Missions with military coercion are associated with a higher number of rebel groups; CI missions are associated with fewer government troops and peace settlements without SSR provisions; and both instruments are less likely to be deployed following outright military victories. Other studies have linked each of these dynamics to less stable peace (Hartzell and Hoddie Reference Hartzell and Hoddie2007; Toft Reference Toft2009).Footnote 31 Both instruments also tend to be accompanied by post-conflict elections—a vehicle through which CI can be enacted (Matanock Reference Matanock2017a; Matanock Reference Matanock2017b)Footnote 32—along with Groups of Friends that can help coordinate either instrument.Footnote 33 Finally, CI missions were more common in countries with higher levels of GDP, indicating that they are not unique to aid recipients. CI missions and missions with mandates for military coercion are otherwise less associated with other variables, both in comparison to the deployment of the other type of mission and to peace periods without peacekeeping. Some of these results suggest that a reduction in conflict recurrence associated with peacekeeping may be an underestimate given the possibility that missions employing each instrument are sent to contexts where peace is more difficult to secure—similar to the argument made by Fortna (Reference Fortna2008)—including on some dimensions that distinguish missions with CI from those mandated for military coercion.
Matching
We also take a second step to address selection concerns. Following Gilligan and Sergenti (Reference Gilligan and Sergenti2008) and Ruggeri, Dorussen, and Gizelis (Reference Ruggeri, Dorussen and Gizelis2017), we turn to matching to control for the observable factors identified in our selection models as potentially confounding the effects of peacekeeping. We use genetic matching (Sekhon Reference Sekhon2011), which employs a genetic algorithm to maximize covariate balance between observations to create a “control” group of cases (post-conflict peace periods that did not receive peacekeeping missions) that are as similar as possible to “treated” cases (periods that received missions using each instrument). We matched peace periods one-to-one with replacement for four different treatments: UN missions with mandates for military coercion; missions without military coercion; missions that employ CI; and missions that do not employ CI. Given that the peacekeeping instruments could be complementary, we also broke missions into four categories (CI with a mandate for military force, and so on) and rerun the results. These treatments are dichotomous variables equal to “1” if a particular mission was present at any point during the peace period and “0” otherwise. For each treatment, we restricted the match to cases that receive no peacekeeping mission at all (the control category).Footnote 34
Matching offers a method to address some selection bias and endogeneity concerns. Yet, it is unable to fully resolve these issues. Matching is only able to approximate an as-if random research design for observed confounders (Dunning Reference Dunning, Collier and Brady2010), so, like most work on peacekeeping, we still cannot completely exclude the possibility that unobserved variables bias our results. However, we include a broad array of potential confounders in our analysis and conduct a battery of robustness checks. Given that the UN is unlikely to randomize its instruments—and in the absence of a plausible instrumental variable—this is the best technique we can employ.
Since the samples are relatively small, we conducted two separate matches on covariates identified in our selection models. The first matched cases on Gilligan and Sergenti's (Reference Gilligan and Sergenti2008) covariates described earlier, plus the number of rebel groups in the conflict ending in the peace period; the second replaced the regional indicators with indicators for war type (ethnic, territorial, or Marxist).Footnote 35 We include these covariates because they seemed to influence peacekeeping deployments in some of our selection models, and unlike some of the other variables—such as Groups of FriendsFootnote 36—they are pre-treatment (a requirement for matching). Both matches generate pairs of observations—one treated; one control—that differ as little as possible on these covariates (for a list, see Online Appendix 1.8). Balance statistics for all matched covariates show we achieved excellent balance according to standard indicators (see Online Appendix 1.6). Our sample size was close to the original: Gilligan and Sergenti (Reference Gilligan and Sergenti2008) had 19 treated observations and 68 possible controls; ours contains 14–17 treated and 101–5 possible controls, depending on the treatment. As a robustness check, we also included only peace periods that followed settlements and rematched (see Online Appendix 2.2).Footnote 37
To evaluate the four treatments—missions with and without CI, and with and without military coercion—we used peace duration, in months, as our outcome variable and ran Cox proportional hazards models. Table 4 presents the results, with values less than 1 indicating a decreased risk of renewed conflict and values greater than 1 indicating an increased risk. For all four treatments, the control category is peace periods that received no UN mission. Yet, our results are similar when we analyze a restricted sample of matched pairs containing only those peace periods that received CI missions assessed against non-CI missions. We included controls for the covariates on which we matched to adjust for any remaining imbalance,Footnote 38 along with an indicator for the alternative instrument than the one being analyzed (so we control for mandates for military coercion when assessing CI missions).
Effect of UN peacekeeping on conflict recurrence, by mandate type
Note: Control category = no peacekeeping. Statistically significant estimates denoted by * p = 0.10; ** p = 0.05; *** p = 0.01; **** p = 0.001. T-statistics in parentheses. ATT = average treatment effect for each treatment.
The results show that only CI missions—which reduced the hazard rate by 99 per cent—are associated with statistically significant reductions in conflict recurrence across specifications.Footnote 39 Missions with mandates for military coercion, and those without CI, were associated with an increased risk of renewed conflict in the first analysis, but neither association is statistically significant. The associations with missions without coercive mandates—which reduced the hazard rate by estimates of 48 and 82 per cent, respectively—are also not statistically significant.
Since hazard models can be unreliable (see, for example, Box-Steffensmeier and Zorn Reference Box-Steffensmeier and Zorn2001), we used GenMatch to estimate the average treatment effect for each treatment (ATT) compared to not receiving peacekeeping. Unlike the Cox models, these are not hazard rates, so positive coefficients indicate a longer peace period. Here, again, only CI missions and those without mandates for military coercion have statistically significant associations with peace.
We also estimated Cox models for the unmatched data using the same controls included in the models in Table 4. The results are similar, but the matched data show much larger associations compared to the unmatched data (see Online Appendix 1.7). This is consistent with Gilligan and Sergenti's (Reference Gilligan and Sergenti2008) findings, indicating that a failure to correct for the nonrandom assignment of peacekeeping underestimates its importance.
We also rematched cases using each possible combination of mission type: CI with and without military coercion; and military coercion with and without CI. For both matches, we added time period to the covariates because it correlates with mandates for military coercion (but not CI) (see Online Appendix 1.2; see also Howard and Dayal Reference Howard and Dayal2018). Balance statistics show each match achieved good balance; however, given the small sample size, we ran the same analyses but without controls (see Table 5).Footnote 40 While the results should be interpreted especially cautiously given the small sample size, they are consistent with our other findings. Compared to no peacekeeping, missions with CI, both with and without military coercion, reduced the risk of conflict recurrence by between 74 and 91 per cent, and the results were statistically significant in all but one model. Missions without CI had a much weaker association with the risk of recurrence (and were not statistically significant). As with the other analysis, we ran Cox models on the unmatched data, which also yielded similar results.
Effect of UN peacekeeping combinations on conflict recurrence
Notes: Two peace periods with peacekeeping—Croatia and Sierra Leone—are double-counted since they experienced two separate UN missions (a coercive mission and a non-coercive mission). Control category = peace periods receiving no peacekeeping. Statistically significant estimates denoted by * p = 0.10; ** p = 0.05; *** p = 0.01; **** p = 0.001. T-statistics in parentheses.
As noted earlier, our findings are similar when we confined the analysis to only those peace periods that received a peacekeeping mission (see Online Appendix 2.1). Peace periods that were treated with CI peacekeeping experienced a large (80 per cent) and statistically significant decrease in the risk of renewed conflict compared to the control group (peace periods that received peacekeeping without CI). However, peace periods treated with missions mandated to use military coercion experienced no statistically significant decrease. Finally, we evaluated the effect of CI-only missions (N = 11), compared to coercion-only missions (N = 10). While the small sample size precluded the use of matching or the inclusion of controls—meaning that, again, the results should be met with caution—we found that CI-only missions reduced conflict recurrence risk by 88 per cent (see Online Appendix 1.7).
Robustness Checks
As noted throughout our analysis, we ran myriad robustness checks, considering different comparisons (CI to no CI missions and so on), different measures, and different covariates. In addition, we reran the analysis in Table 4 controlling for seven additional correlates found in our selection models to be positively associated with all missions: whether missions were deployed during the preceding conflict; whether any non-UN mission was deployed post-conflict; whether a non-UN mission mandated for military coercion was deployed; whether there was prior peacekeeping in the country; whether a Group of Friends formed; whether a peace period had a settlement providing for participatory elections; and whether post-conflict elections were held during the peace period.Footnote 41 We did not match on these variables since they predict missions but not necessarily conflict recurrence, and most are post-treatment, so they cannot be used in matching. The results are substantively the same (see Online Appendix 2.0a).
We also repeated the analysis after incorporating governance-related factors that previous research has shown to be correlated with post-conflict peace, such as political and legal institutions that can constrain elites (Walter Reference Walter2015). We therefore matched on three covariates that measure legal accountability, political accountability, and transparency in the country in the first year of the peace period (Freedom House 2019; World Bank 2019). The results hold (see Online Appendix 2.0b).
We then conducted a series of additional checks, where we changed the genetic algorithm for population size for all matches,Footnote 42 included and excluded cases where we disagree with Gilligan and Sergenti's (Reference Gilligan and Sergenti2008) coding, used alternative coding for ambiguous cases of conditionality, removed potential “key” cases (for example, the former Yugoslavia), and included controls for how the previous conflict ended, past UN missions, and the presence of other peace operations (see Online Appendix 2.0c). We also reran the match exclusively on peace periods with peace agreements (see Online Appendix 2.2). Only CI missions consistently yield statistically significant results, where it is associated with a reduced risk of renewed conflict compared both to no peacekeeping and to other types of missions.
Conclusion
How does international intervention work? This article focuses on a prominent form of international involvement in civil wars—UN peacekeeping missions—and disaggregates the instruments used to influence prospects for post-conflict peace. In doing so, we contribute to an important and growing literature seeking to understand how intervention works. We challenge the notion that peacekeeping missions often rely on military coercion to help combatants overcome commitment problems in post-war settings, and we build on previous research to theoretically and empirically specify a different instrument—CI—which relies on rewards, pressure, and punishments that do not entail force.
We find cross-national evidence that UN peacekeeping missions that condition economic, legal, and political incentives on compliance with post-war settlements are associated with more enduring peace. We collect original data on peace operations that employ these CI, and, to analyze the effect of different types of peacekeeping, we extend data on post-conflict peace periods. We then follow other studies of UN peacekeeping (Fortna Reference Fortna2008; Gilligan and Sergenti Reference Gilligan and Sergenti2008; Ruggeri, Dorussen, and Gizelis Reference Ruggeri, Dorussen and Gizelis2017) and employ both selection models and matching to help address potential selection problems. Our analysis suggests that CI help prolong peace, both examined alone and when paired with mandates for military coercion, whereas military coercion alone does not. This casts doubt on the notion that effectively keeping peace between ex-combatants requires the threat of force. Our approach has limits—a small sample size and an inability to completely rule out unobserved sources of bias—so, like most work on peacekeeping missions, we cannot definitively claim causality. Yet, combined with the qualitative examples we presented earlier in the article, this is compelling evidence that CI are an effective peacekeeping tool. Our results thereby challenge studies suggesting more militarized interventions are needed to secure post-conflict peace (Hultman, Nygård, and Hegre Reference Hultman, Nygård and Hegre2016) but are consistent with recent work showing softer missions can prevent the outbreak of conflict (Beardsley Reference Beardsley2011; Howard Reference Howard2019).
More research is needed to confirm or invalidate more implications of our analysis, and to further assess the relative effectiveness of peacekeeping instruments under different conditions, including at the subnational level (Gizelis and Benson Reference Gizelis and Benson2019). However, this article takes a crucial step toward carefully theorizing and quantitatively testing a potentially important non-military instrument by which intervention helps secure settlements following civil conflicts. Our results suggest that, at least under certain conditions, peacekeeping can increase the durability of peace even if it does not possess a mandate to use force. Future research should therefore explore when different CI are used, by what types of actors, when they are effective, and when they fail, building on our analysis of conflict types and severity. Such studies should also explore alternative development and democratization outcomes.
Understanding the processes by which intervention works, especially these tools that we examine, is critical for refining theories of peacekeeping and identifying the ingredients of a successful operation for policymakers. Amid growing debate over the proper role of peacekeeping, UN officials—such as Jean Arnault (Reference Arnault2006), former head of several peace missions—recognize that non-military measures are critical tools for an effective intervention. Yet, it has seemingly become conventional wisdom that, in the words of one report, “robust peacekeeping involves the use of force” (Sartre Reference Sartre2011, 20). While such tactics may be necessary in some cases—particularly in the midst of active conflict, but perhaps also to protect civilians and mission personnel in post-conflict contexts—our research questions the assumption that more active and armed troops are always needed to help implement a peace agreement. Given that a primary goal of peacekeeping missions is often to prevent conflict recurrence between combatants, policymakers may be able to consider less costly options of intervention. This could enhance the viability of third-party intervention, since countries may be more likely to supply peacekeepers if they are not required to employ military coercion. The UN can focus on effectively employing alternative instruments to force—such as donor assistance or elections—to promote peace.
Acknowledgements
We are most grateful to Naazneen Barma, Kyle Beardsley, Andrew Bertoli, Jennifer Bussell, John Ciorciari, Bridget Coggins, Alex Cooley, Paul Diehl, Michael Doyle, Pierre Englebert, Jim Fearon, Page Fortna, Desha Girod, Kristian Skrede Gleditsch, Lise Morjé Howard, Susan Hyde, Bob Keohane, Steve Krasner, Andy Kydd, Dan Lee, Melissa Lee, Katerina Linos, Michaela Mattes, Alison Post, Bob Powell, Emily Ritter, Burcu Savun, Ken Schultz, Megan Shannon, Hannah Smidt, Art Stein, and Sarah von Billerbeck, the journal editors and anonymous reviewers, as well as others at the Annual Meetings of the American Political Science Association, the International Studies Association, the UC Conference on International Cooperation, the Online Peace Science Colloquium, and workshops at Brown, Denver, Essex, Maryland, MIT, Notre Dame, Princeton, Stanford, Wisconsin, and Yale for valuable comments that have improved our work. Of course, however, any omissions or other errors are our own.
Supplementary Material
Online Appendices are available at: https://doi.org/10.1017/S0007123421000491
Data Availability Statement
Replication Data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/ENAZAA
Financial Support
Our early research on this project was supported by funding from the Institute on Global Conflict and Cooperation at the University of California, San Diego.
Competing Interests
None.
