Does peacekeeping only work in easy environments? An analysis of conflict characteristics, mission profiles, and civil war recurrence

ABSTRACT

Peacekeeping is widely considered to be an effective means of preventing civil war recurrence. However, as peace has collapsed in a considerable number of cases despite peacekeeping efforts, we are left with the question which combinations of peacekeeping environments and peacekeeping missions lead to lasting peace. This article compares 22 peacekeeping missions between 1990 and 2012. While prominent United Nations documents assume that the success of post-conflict peacekeeping primarily depends on the features of the mission itself, the analysis shows that characteristics of the terminated civil war have a strong influence on whether peace endures. Restrained peacekeeping, defined by low troop density, non-robustness, and a lead nation that is not a permanent member of the Security Council, only succeeds in preserving peace in conducive environments. Inconclusive war endings, evenly distributed military capabilities at war’s end, ethnic conflicts, and high intensity create a particularly difficult context for peacekeeping.

KEYWORDS:

• Intrastate conflicts
• peace processes
• peace operations
• post-conflict peacebuilding
• qualitative comparative analysis

Peacekeeping is widely regarded as an effective instrument for preventing the resumption of civil war. Walter (2002) argues that the conclusion and implementation of peace agreements require security guaranties from third parties—an assurance that peacekeeping missions are able to provide. Fortna (2008, p. 179) states that “peacekeeping is an extremely effective tool for maintaining peace.” Numerous studies support these findings (e.g., Hultman, Kathman, & Shannon, 2016, pp. 240–245; Joshi, 2013, pp. 376–377; Mason, Gurses, Brandt, & Quinn, 2011, pp. 185–187). In recent years, however, an increasing portion of contributions has found that the effect of peacekeeping is not statistically significant (e.g., Hegre & Nygård, 2015, pp. 996–997; Joshi & Quinn, 2015, p. 25; Walter, 2015, pp. 1253–1255). As civil war has recurred in a considerable number of cases despite peacekeeping efforts, it is important to investigate which conditions allow peacekeeping missions to effectively prevent relapses into civil war.

Research on the conditions for peacekeeping success has not kept pace with studies that have investigated the average effect of peacekeeping missions on peace survival. The few comparative studies that do exist examine missions during armed intrastate conflicts and deployments prior to and after such instances of violence (Bratt, 1997; Durch, 2006; Heldt, 2001; Kim, 2004; Koko & Essis, 2012; Pushkina, 2006). Considering this mixed set of cases, one may question whether their insights are applicable to post-civil war peacekeeping. Another shortcoming is that most studies isolate variables from one another. While some contributions do investigate the interplay between security guaranties by third parties and powersharing (Walter, 2002) or between peacekeeping and war outcome (Caplan & Hoeffler, 2017, pp. 147–149), the quantitative methods they apply only allow for a small number of interaction effects to be considered. Identifying which combinations of conditions lead to success or failure requires a more systematic consideration of the interaction among various aspects of peacekeeping missions, along with the peacekeeping environment. In light of this, I apply Qualitative Comparative Analysis (QCA)—a method designed to address the interplay between conditions—to 22 peacekeeping missions deployed after civil wars, all of which ended in the post-Cold War era.

This article rejects the widespread assumption that it is only the characteristics of the peacekeeping mission itself that decide its outcome—a misjudgment that can be found in prominent United Nations (UN) policy documents. In identifying success factors for peacekeeping, the Capstone Doctrine for instance lists impartiality, limited use of force, legitimacy, credibility, and the promotion of national and local ownership (United Nations, 2008, pp. 31–40). The only factor it mentions beyond the mission profile is consent of the conflict parties, which is present in all cases considered in this article. The so-called HIPPO-Report (Report of the High-Level Independent Panel on Peace Operations, 2015, para. 10) also ignores contextual factors, stating that “there are outer limits for UN peacekeeping operations defined by their composition, character and inherent capability limitations.” Academic contributions also conclude that the mission environment is of minor importance for peacekeeping success. According to Martin-Brûlé (2017, p. 2), “outcomes are better explained by the type of intervener and the strategy employed than by the type of context.”

This article theoretically and empirically demonstrates that the peacekeeping environment is, in fact, quite relevant in terms of the impact of peacekeeping missions. Whether a civil war was fought between ethnically defined parties, whether it was of high intensity, ended inconclusively or ended with a military balance effectively distinguishes cases of peace survival from those with civil war recurrence. For cases in which most of the aforementioned civil war characteristics were present, peace collapsed when the given peacekeeping mission operated with a restrained approach. A “deliverable mandate” or a “tailored mission,” as demanded by the Capstone Doctrine (United Nations, 2008, p. 38) and the HIPPO-Report (Report of the High-Level Independent Panel on Peace Operations, 2015, para. 50–51), can only be achieved when the role of contextual conditions is recognized.

The article commences with a theoretical discussion of how characteristics defining the peacekeeping environment and the peacekeeping mission are related to peace survival and civil war recurrence. It then outlines the key terms, the sample, method, and operationalization of the conditions. Using QCA, the next section investigates the characteristics of terminated civil wars. The QCA application also presents a summary of the peacekeeping environment along with the features of the peacekeeping mission, highlighting that restrained peacekeeping fails in adverse environments. The conclusion discusses implications for peacekeeping missions and further research.

Peacekeeping and civil war recurrence

The Brahimi Report, a prominent UN document published in 2000, makes the case for peacekeeping contexts to be taken seriously. In defining the environments of peace operations marked by greater or lesser degrees of danger, it discusses the type of conflict, the extent of casualties and destruction, the number of conflict parties, their aims and sources of income, and the role of neighboring countries. The report concludes that it “is vitally important that negotiators, the Security Council, Secretariat mission planners, and mission participants alike understand which of these political-military environments they are entering” (Report of the Panel on United Nations Peace Operations, 2000, para. 20–25, citation para. 26).¹

Practitioners have also acknowledged the importance of conditions beyond the peacekeeping mission itself. According to former U.S. diplomat Jett (2019, p. 209), “[t]he fundamental—and often missing—factor is a true interest by the parties in a lasting peace.” As Guéhenno (2015, p. 292), former UN Under-Secretary-General for Peacekeeping Operations, stresses, “the most critical element for the success of a peacekeeping operation is the political process that must underpin it.”²

Academic contributions likewise suggest that the peacekeeping environment is relevant for the outcome. Many quantitative papers consider contextual factors as control variables, including type of conflict, intensity, and the type of war termination (e.g., Fortna, 2008, p. 119; Hultman et al., 2016, pp. 240–245; Koko & Essis, 2012, p. 72). One study by Heldt (2001, p. 127) sets itself apart from others by concluding: “It is … the source of conflict or context, rather than the nature of peacekeeping missions, that determines success.” Doyle and Sambanis (2006, pp. 63–68) introduce an analytical triangle, highlighting the interplay between the level of hostility, local capabilities, and international assistance, including peacekeeping. They argue that the three “dimensions … substitute for each other—more of one substitutes for less of another” (Doyle & Sambanis, 2006, p. 64). Thus, “the more difficult the peace process will be … the more international assistance/authority will be needed” (Doyle & Sambanis, 2006, p. 65). Drawing from their approach, I assume that the profile of the peacekeeping mission does not suffice on its own for explaining why peacekeeping succeeds or fails in preventing civil war recurrence.

In arguing that the peacekeeping context matters, I summarize what previous studies have identified as the “causal mechanisms of peacekeeping.” I then address the question of how these mechanisms, and the prospects of peace survival, are affected by the peacekeeping environment. I subsequently discuss attributes of peacekeeping missions with reference to these causal mechanisms. Finally, the theoretical section deals with the interplay between the peacekeeping environment and the profile of the peacekeeping mission.

Causal mechanisms of peacekeeping

Fortna (2008, pp. 82–102) identifies four “causal mechanisms of peacekeeping” which I complement by insights from other authors. Causal mechanisms are understood here as the ways in which peacekeeping preserves peace. Fortna (2008, p. 86) stresses that “these mechanisms overlap in practice.”

The first mechanism alters the conflict parties’ incentives for war and peace by interpositioning peacekeepers between the warring parties. As one must first assault the peacekeepers in order to attack the opponent, strong peacekeepers can deter or counter such aggression (Walter, 2002, pp. 26–27). Peacekeeping also influences the incentives mentioned above by providing the option of a “face-saving” withdrawal, allowing one conflict party to claim that it has not given in to its opponent (James, 1969, p. 2).

Second, peacekeeping reduces uncertainty and fear. In consenting to peacekeeping, a conflict party increases its costs for resorting to the military option, thus signaling its intention for peace (James, 1969, pp. 92–93). Furthermore, peacekeepers provide security guaranties and thereby alleviate the commitment problem that can lead a peace process to collapse (Walter, 2002, p. 7, 14–27). Such a problem exists when neither party to the conflict can credibly commit to abiding by the provisions of an agreement, as each side knows that the best outcome would be to breach the agreement while the adversary implements it (Fearon, 1995, pp. 401–403). Peacekeepers can also serve as bodyguards for vulnerable representatives of the conflict parties, granting them sufficient assurance to participate in the peace process (Rikhye, 1984, p. 95).

Through the third causal mechanism, peacekeepers prevent unintended escalations by containing violent incidents at the local level (James, 1969, p. 294). Fourth, peacekeepers establish a safe environment for civilian peacebuilding.

The peacekeeping environment matters

The difficulty of peacekeeping varies by context: some environments provide greater incentives for resuming war, along with more uncertainty and fear, and increase the risks of an unintended escalation more than others.

Previous studies have highlighted two dimensions of the peacekeeping environment: the first relates to the characteristics of the terminated civil war and the second includes post-war developments beyond peacekeeping. I limit the analysis to the first dimension, as it would exceed the scope of this article to introduce multiple conditions for each dimension as well as the war characteristics, the peacekeeping mission itself, and other post-war developments. Another reason for opting to focus on this dimension is that, when peacekeeping initiators plan a mission, they can rely more on their knowledge of the civil war characteristics than on their expectations of post-war developments.

I discuss characteristics of the terminated war with regard to their direct relationship to civil war recurrence and their indirect impact on the causal mechanisms by which peacekeeping missions are deemed to preserve peace.

Ethnic conflicts are taken to be particularly prone to recurrence (e.g., Kreutz, 2010, p. 248) due to the prolonged and larger amount of grievances as well as opportunities for mobilization (Denny & Walter, 2014). As ethnic bonds are especially stable and can more easily be mobilized than other types of identity, these create stronger incentives for resuming war while also sustaining fear on both sides. Unintended escalation is more likely as ethnically defined parties to the conflict are more willing to interpret an incident as an assault on the entire group rather than as an isolated event.

Most previous studies have found that higher intensity makes civil war recurrence more likely (e.g., Toft, 2010, p. 56). More intense conflicts deepen cleavages and cause greater destruction, thereby intensifying incentives for war such as grievances, the desire for revenge, and opportunities. Additionally, more intense conflicts affect a greater number of people, resulting in more widespread fear and belief in the enemy’s malicious intentions, which exacerbates the commitment problem and the risk of unintended escalation.

With regard to war outcome, a plurality of contributions concludes that victories are better suited for preserving peace than peace agreements since military triumphs destroy the organizational capabilities of the defeated side (e.g., Licklider, 1995, p. 685). Other studies find victories and peace agreements to have an equal effect on preventing civil war recurrence when compared to ceasefires (Doyle & Sambanis, 2006, pp. 87–88) or other war outcomes (Meernik, Nichols, & King, 2010, pp. 317–318) such as “low activity”—continued violence that does not surpass the threshold of 25 fatalities per calendar year, as used by the Uppsala Conflict Data Program (UCDP).³ A higher degree of conclusiveness distinguishes peace agreements and victories from other types of war termination (cf. Meernik et al., 2010, p. 324). The war outcome offers clues to the quality of what Guéhenno (2015, p. 292) calls the underpinning political process. After failing to conclude a peace agreement, each former warring party has less trust in the good will of its enemy. They perceive mere ceasefires and war termination due to low activity as less binding than peace agreements, which not only regulate military behavior but also address the political issues underlying the conflict. Furthermore, in the absence of political concessions provided by peace agreements, the parties to the conflict have less to lose and resuming the civil war is therefore less costly. Inconclusive war endings create a particularly high level of uncertainty about future arrangements, which impedes peacebuilding activities. Finally, after inconclusive war terminations, the parties to the conflict tend to possess more capabilities for resuming fighting.

Some peace agreements are signed under the condition of military asymmetry while others are concluded under a military balance. The distribution of capabilities is deemed to be important considering that military balances cause “conflicting expectations of what … war will be like” and therefore makes war more likely (Blainey, 1973, p. 56). In contrast, clear asymmetry implies concordant perceptions, which tend to preserve peace (Wagner, 1993, pp. 242–246). Moreover, military balances aggravate commitment problems and further unintended escalation: The parties will perceive that they are more likely to obtain a decisive advantage by violating an agreement.

Overall, this discussion about the characteristics of the terminated civil war with reference to the causal mechanisms of peacekeeping suggests that the context does, in fact, matter for the outcome of peacekeeping.

The characteristics of terminated civil wars mentioned above were drawn from studies that understand causality in probabilistic terms and marginally investigate the interplay of conditions. As such, these contributions do not provide insights into which of the possible combinations of these war characteristics are related to peace survival or civil war recurrence. It would be far-fetched to assume that one disadvantageous condition (e.g., an ethnic conflict or a highly intense civil war) cancels out the negative impact of another (e.g., an inconclusive war ending or a military balance at war’s end) or impedes another adverse condition from arising. It would be safe to assume that, as the number of adverse war characteristics increases, the peacekeeping environment becomes more difficult. When such disadvantageous characteristics are predominant, I refer to the affected environment as “difficult” or “adverse,” and otherwise as “easy” or “conducive.” Peacekeeping succeeds when there are no more adverse than advantageous characteristics of the terminated civil war.

Hypothesis 1: Peace survives when adverse civil war characteristics, an ethnic conflict, high intensity, termination through a ceasefire or “low activity,” and a military balance at war’s end, do not outnumber advantageous conflict characteristics.

Restrained peacekeeping only works in conducive environments

I now use the mechanisms of peacekeeping to discuss the features of a mission. How well a mission alters the incentives for war and peace, reduces uncertainty and fear, prevents unintended escalations, and establishes a safe environment for civilian peacebuilding depends on the credibility and capability of the peacekeepers. Credibility and capability are interrelated: credibility is shaped by some aspects of capability such as the absolute or relative size of the deployment, the robustness of the mandate, troop training, equipment, and the status of the key participating states (Walter, 2002, p. 41).

All else being equal, the more troops a peacekeeping force contains, the more capable it will be at deterring or countering an aggressive act, at guaranteeing security by establishing large buffer zones, at signaling good intentions, at containing local incidents, and at protecting individuals and peacebuilding efforts (Hultman et al., 2016, pp. 234–237).

Robust missions are mandated to use force for purposes beyond self-defense. Robustness aims to alleviate uncertainty by signaling more resolve for guaranteeing security and by making the peacekeepers less dependent on the good will of the parties to the conflict. By giving their consent for robust deployments, the parties can signal their good intentions. Robust mandates reduce incentives for war, in general, and for escalating local incidents, in particular, by increasing the military costs of attacking. Moreover, robust mandates facilitate a pro-active approach to establishing a safe environment for civilian efforts. On the other hand, robust missions require better trained and equipped troops, they can provoke more resistance, including attacks on civilian peacebuilders, and they are riskier for peacekeepers (Hunt, 2018, pp. 151–160; Tardy, 2011, pp. 160–162).

Another discussion addresses the degree to which peacekeeping success depends on the participation of particular actors. Some scholars argue that success is enhanced when permanent members of the UN Security Council (P5) assume a dominant role in the mission: these states introduce high military capabilities and political resolve (Bratt, 1997, p. 54; Walter, 2002, p. 41).

Many studies discuss whether UN peacekeeping is less successful than missions carried out by other actors (e.g., Rikhye, 1984, p. 239). UN missions are deemed to be more impartial but also exhibit less morale and resolve, less expertise about the affected region, poorer organization, and lower military quality compared to deployments by regional organizations or coalitions of willing states (Durch, 2006, pp. 582–588; Malan, 2018, p. 39; Martin-Brûlé, 2017, p. 10). As discussed, high capabilities and resolve help fulfill the tasks related to the causal mechanisms of peacekeeping.

In the following, high troop density, a robust mandate and dominance by a P5 member are what defines a strong peacekeeping posture. In contrast, a low troop-to-population ratio, a mandate confined to self-defense, and a non-P5 state assuming a leading role in the mission constitute restrained peacekeeping or peacekeeping “lite.”⁴ Depending on which features preponderate, a mission is labeled strong or restrained.

Compared to a strong posture, restrained peacekeeping has less influence on the incentives for war and peace, it is less effective at reducing uncertainty and fear, it is more limited in containing local escalations, and it is less capable of safeguarding civilian efforts. When an easy environment compensates for the smaller potential of a restrained peacekeeping mission, peace survives. But when the characteristics of a difficult environment preponderate (ethnic conflict, high intensity, an inconclusive war termination, and a military balance at war’s end), the profile of the peacekeeping mission must compensate. Consequently, peacekeeping lite does not suffice for preventing civil war recurrence in difficult environments.

Hypothesis 2: Restrained peacekeeping only prevents civil war recurrence when advantageous conflict characteristics are not outnumbered by adverse characteristics.

Strong peacekeeping is needed to compensate for adverse contexts

Compared to restrained peacekeeping, a comprehensive mission with a robust mandate led by a powerful state has greater influence on the former warring parties’ incentives for war and peace and on the level of uncertainty and fear. It has more capacities to respond to local escalations and support peacebuilding activities. Thus, peacekeeping missions with a strong posture can compensate for a difficult environment where the parties to the conflict are more willing and more capable to resume war.

Hypothesis 3: For contexts in which adverse civil war characteristics preponderate, peace only survives when a strong peacekeeping mission is deployed.

Hypothesis 2 seems to imply Hypothesis 3. But, conceivably, the empirical analysis only supports the former and not the latter. I therefore state two separate hypotheses.

Drawing from Doyle and Sambanis (2006, p. 64), I argue that difficult contexts, in contrast to easy environments, need to be compensated by missions with a strong posture to ensure lasting peace. While strong peacekeeping is not needed in an easy environment, it is also not a threat to peace survival here. As indicated above, a particularly strong mission can provoke violent resistance from the conflict parties. In an easy environment, however, the parties have fewer causes and opportunities for such a resistance.

Hypothesis 4: Peace survives when peacekeeping missions with a strong posture operate in easy environments.

Research design

Key terms

Peacekeeping is the deployment of armed troops or unarmed military observers in a third country with the aim of preventing the resumption of hostilities (cf. Fortna, 2008, p. 5). This article only deals with post-civil war peacekeeping and its military component.

Civil war is defined as armed intrastate conflict in which at least 1,000 lives have been lost in battles or assaults throughout the period of fighting. Apart from the number of battle-related deaths from state-based conflicts, this definition also considers victims of one-sided violence against civilians and those due to non-state conflicts, provided that one of the parties of the state-based conflict participated in such violence. The sample is based on UCDP datasets.⁵

A civil war is considered terminated when the threshold of 25 fatalities has not been surpassed for at least one year—for state-based conflicts, non-state conflicts, and one-sided violence. Civil war recurrence is conceptualized in a strict sense, meaning that the new war involves the same warring parties, concerns the same issues and leads to at least 1,000 fatalities in a state-based conflict, a related non-state conflict or one-sided violence.

Cases

The dynamics of terminating civil wars and building peace have significantly changed since the end of the Cold War (Merz, 2012, p. 204). In the current era, civil wars have more often ended in peace agreements (Kreutz, 2010, p. 246). At the same time, peacekeeping missions have become more frequent and robust, comprise more personnel, and exhibit a multidimensional range of activities (see e.g., Koops, MacQueen, Tardy, & Williams, 2015, pp. 2–3, 262–263; Bellamy & Williams, 2015, pp. 14–15, 29–30). In focusing on civil wars terminated after 1990, this article stands by the claim made by Wallensteen (2015, p. 97) that “the Cold War experiences are … irrelevant for post-Cold War undertakings.” It considers civil war terminations until the end of 2009, allowing time for observing whether peace collapsed in the subsequent years for each case considered.

Relying on data compiled by Fortna (2008), Mullenbach’s dataset for Third Party Peacekeeping Missions (2013) and case-specific literature, I identify 22 post-civil war peacekeeping missions in 17 countries. I code cases as single missions when an operation has been renamed or when the executing organization has changed. Developments are traced until the resumption of civil war, or for the first seven post-war years, if peace has endured. This time span captures all seven resumptions that occurred until 2012, while losing only two cases of peace survival with an investigation period shorter than seven years. Civil war recurred despite the presence of a peacekeeping mission in six cases and recurred following the withdrawal of peacekeepers in one case.

Table

Cases with peace survival

Cases with civil war recurrence

Bosnia–Herzegovina (Croats) 1993–1994 Bosnia–Herzegovina (Serbs) 1992–1995 Burundi (National Council for Defence of Democracy-Forces for the Defence of Democracy) 1994–2004 Croatia 1995^6 Democratic Republic of the Congo (DRC) (Mouvement de Libération du Congo (MLC)) 1998–2002 El Salvador 1980–1991 Georgia (Abkhazia) 1992–1993 Guatemala 1965–1995 Indonesia (Aceh) 1999–2005 Liberia 2000–2003 Mozambique 1977–1992 Nicaragua 1982–1990 Sierra Leone 1991–2000 Tajikistan 1992–1996 Yugoslavia (Kosovo) 1998–1999

Angola (National Union for the Total Independence of Angola) 1975–1995 DRC (Rassemblement Congolais pour la Démocratie du Peuple (RCD)) 1998–2004 DRC (Congrès national pour la défense du peuple (CNDP)) 2006–2009 Liberia 1989–1996 Sri Lanka (Tamil Tigers) 1983–2001 Sudan (Sudan People’s Liberation Army SPLA)) 1983–2004 Yugoslavia/Croatia 1991–1993

Method

Peace survival versus civil war recurrence does not necessarily depend on a single condition or one combination of conditions. Various paths may lead to stable peace or resumed civil war. QCA has been designed to uncover different causal paths that may each combine several conditions. This method presumes that causation can be “conjunctural.” While sociology was the first area that widely applied QCA, this method is also gaining recognition in peace research and security studies (e.g., Binder, 2015; Haesebrouck, 2018). QCA prefers the notion of “conditions” over “independent variables” and “outcome” over “dependent variable.”

QCA serves to identify necessity and sufficiency for a particular outcome (Ragin, 1987; Schneider & Wagemann, 2012). An ethnic conflict, for instance, proves necessary for civil war recurrence if each resumed civil war is an ethnic conflict. An ethnic conflict will prove sufficient for civil war recurrence if each ethnic conflict results in resumed civil war. To stress the interest in necessity and sufficiency, users increasingly label QCA as a set-theoretic method.

As the outcome is understood as dichotomous (civil war recurrence vs. peace survival), I follow a recommendation by Ragin (2014, p. 90), QCA’s developer, and use crisp-set QCA for which all conditions are binary.

Selection and operationalization of conditions

The conceptual section has introduced conditions that fulfill two requirements. First, they are based on sound theoretical reasoning and, second, they prove promising in light of the empirical data. The second requirement is needed as even the best theory will prove useless if the values of the corresponding condition remain constant or nearly constant for all cases considered.⁷ Such a condition cannot contribute to the explanation, even if combined with other factors: The QCA algorithm will exclude it due to redundancy (Berg-Schlosser & De Meur, 2009, p. 28).

Table A1 in the online appendix⁸ summarizes the raw data for all conditions and the outcome. I use data compiled by a larger research project on civil war recurrence in assigning the values for all conditions related to the characteristics of the terminated civil war. The project’s dataset is based on case descriptions⁹ that document all sources and coding decisions (Gromes & Ranft, 2016). All codings for these conditions rely on assessments in the case-specific literature, with any exceptions indicated.

The condition “ethnic conflict” designates whether a civil war was waged between ethnically defined conflict parties. The condition on relative war intensity is dichotomized at 2.5% of the pre-war population (“intensity2.5”).

For the war outcome, this analysis does not juxtapose victories and peace accords as is common in previous studies. Such a juxtaposition ignores that the war outcomes ceasefire and “low activity” occur just as frequently as victories in my sample. The QCA application compares conclusive and inconclusive war endings. The condition “no accord no victory” is coded with 1 if the war ended in a ceasefire or “low activity” and with 0 in cases of a peace agreement or victory.

The condition “balance” aggregates several military balance indicators at the end of the civil war: whether one side won the war, the extent of rebel-controlled territory, the rebels’ relative fighting capacity, whether both sides could have continued fighting at the end of the war, and whether one side killed or captured its opponent’s political leader. Relevant datasets and case-specific sources were used to assess these aspects.¹⁰ Combined, the indicators define a spectrum with three segments of equal scope: the middle segment marks a military balance (value 1), while the two outer sectors represent military imbalance favoring the government or the rebels (value 0).

Table A1 in the online appendix documents the properties of the peacekeeping missions, including the maximum number of deployed troops or unarmed observers (according to Mullenbach, 2013). As 10,000 troops in a small state signal more resolve and relative military capability than the same number in a large country, I relate the number of troops to the mean size of the population in the post-war period and to the size of the target country. If the conflict is limited to parts of a country, only the affected population or area is considered. No theoretical considerations guide the dichotomization of these conditions for which the raw data is metrically scaled. QCA literature recommends identifying leaps in the raw data to avoid thresholds that would otherwise classify cases with similar values into different categories. Moreover, dichotomization must ensure that both categories contain a considerable number of cases (Rihoux & Ragin, 2009, pp. 28, 42, 93, 125). Values for the troop-to-population ratio (“troops per capita”) are dichotomized using a threshold of 300 troops per 100,000 inhabitants. The threshold for dichotomizing the troop-to-space ratio (“troops space”) is set at 100 troops per 1,000 km².¹¹

The dichotomization of the condition “robust” is based on Fortna’s research and on relevant sources such as UN Security Council resolutions for recent cases. The dominant role of a P5 state (“P5”) is given when such a state is identified as a mission’s lead state in Mullenbach’s peacekeeping dataset. A UN mission (“UN”) is provided when the UN conducts the operation. The compilations by Fortna and Mullenbach, along with documents found on the UN website, were used as the main sources for coding this condition.

Analysis of the peacekeeping environment

QCA must first identify necessary conditions and then sufficient conditions (Schneider & Wagemann, 2012, p. 115). As Table A4 in the online appendix shows, civil war termination through a peace agreement or victory proves quasi necessary for peace survival. Nicaragua is the exception that confirms the rule: although no formal peace agreement settled the political issues at stake, the war ended conclusively. When the civil war was terminated through a ceasefire, the former Sandinista government had already accepted the process of political transition and lost power in elections.

A military balance at war’s end only minimally fulfills the requirements of a necessary condition for civil war recurrence.

I now turn to the analysis of sufficiency.¹² This article claims that the peacekeeping context matters for the success or failure of peacekeeping missions. This assumption would be supported if we find that the conditions of the peacekeeping environment effectively distinguish cases of peace survival and instances of civil war recurrence. As will be shown, this criterion is fulfilled.

Hypothesis 1 specifies the general claim by stating that peace survives when advantageous civil war characteristics are not outnumbered by adverse characteristics. Table 1 presents the results of the analysis of sufficiency with regard to peace survival. Before discussing these results I outline how to read them.

Table 1. Sufficiency analysis for peace survival.

Conditions: ethnic conflict, intensity2.5, balance, no accord no victory
Solution consistency: 1.000, solution coverage: 0.933 (Burundi not covered)
Paths	Consis-tency	Consistent cases	Raw coverage	Unique coverage
1) ∼intensity2.5*∼no accord no victory	1.000	Bosnia-H. (Serbs), Bosnia-H. (Croats), Croatia 2, DRC MLC, El Salvador, Georgia, Indonesia, Liberia 2, Sierra Leone, Tajikistan, Yugoslavia	0.733	0.200
2) ethnic conflict*∼balance *∼no accord no victory	1.000	Croatia 2, Georgia, Guatemala, Indonesia, Yugoslavia	0.333	0.067
3) ∼ethnic conflict *balance* ∼no accord no victory	1.000	DRC MLC, El Salvador, Mozambique	0.200	0.067
4) ∼ethnic conflict* ∼intensity2.5 * ∼balance	1.000	Nicaragua, Sierra Leone, Tajikistan	0.200	0.067

The solution consists of four paths, which, combined, cover 14 out of 15 cases of peace survival, implying solution coverage rate of 93.3%. The paths depict combinations of conditions that consistently relate to the outcome. An asterisk indicates which conditions are combined and the tilde designates the negation of the condition: “∼intensity2.5,” for instance, represents conflicts with relatively low intensity.

Path 1 describes civil wars of relatively low intensity terminated through a peace agreement or victory. The paths result from filtering configurations (i.e., combinations of conditions) consistently related to the outcome under investigation. These configurations are then compared to identify dispensable conditions. In Path 1, the condition that indicates ethnic conflicts proves expendable, as some covered cases were ethnic conflicts while others were not. The same holds true for a military balance at war’s end.

The consistency (100%) in Path 1 indicates the portion of cases that experienced lasting peace. Configurations must exhibit consistency of at least 75% to be considered in the sufficiency analysis (Schneider & Wagemann, 2012, p. 127).

Path 1 describes 73% of all cases of lasting peace (raw coverage). Since it overlaps all other paths, the portion of cases it exclusively covers (unique coverage) is 20%. Due to this overlap, the sum of the raw coverages of all paths is higher than the solution coverage (0.933) and even higher than 1.

Path 2 covers ethnic conflicts with a military imbalance at the end of a war terminated through a peace agreement or victory. On Path 3, non-ethnic conflicts with a military balance at the end of a war terminated through a peace agreement or victory are consistently related to peace survival. Path 4 represents non-ethnic conflicts of lower intensity and a military imbalance at war’s end. The type of war termination is not part of this configuration as, for this aspect, Nicaragua (ceasefire) deviates from Sierra Leone and Tajikistan (peace agreement respectively).

I now discuss the results. Expressed in set-theoretic terms, with regard to sufficiency, Hypothesis 1 states that cases with a peacekeeping environment in which no or no more than two adverse civil war characteristics are given are a subset of all cases with peace survival. This implies that if Hypothesis 1 is correct, the paths to lasting peace in Table 1 should include no or only a few disadvantageous civil war characteristics. Indeed, Path 1 and Path 4 exhibit no condition in an adverse form and, on Paths 2 and 3, only one civil war characteristic appears in its disadvantageous form. The single uncovered case of peace survival is Burundi, an ethnic conflict of high intensity with a military balance at war’s end terminated through a peace agreement. In this deviating case, three conditions were given in their disadvantageous form. The reason why peace lasted here will be discussed later on.

Overall, the analysis of peace survival supports Hypothesis 1. This hypothesis would be additionally corroborated if the analysis with regard to civil war recurrence shows that cases in which adverse civil war characteristics preponderate are a subset of all cases with the resumption of civil war. Paths consistently related to the collapse of peace should exhibit a preponderance of adverse civil war characteristics. The results presented in Table 2 correspond to this.

Table 2. Sufficiency analysis for civil war recurrence.

Conditions: ethnic conflict, intensity2.5, balance, no accord no victory
Solution consistency: 1.000, solution coverage: 0.714 (Liberia 1, Sudan not covered)
Paths	Consis-tency	Consistent cases	Raw coverage	Unique coverage
1) ethnic conflict*balance*no accord no victory	1.000	Angola, DRC CNDP, DRC RCD, Sri Lanka, Croatia 1	0.714	0.714

The single path covers five out of seven cases of resumed civil war and describes ethnic conflicts with a military balance at the end of a war terminated through a ceasefire or due to low activity. The path combines three disadvantageous conditions.

In total, the paths to peace survival or civil war recurrence cover 19 out of the 22 cases. As expected, conducive environments are associated with peace survival, whereas contexts in which adverse civil war characteristics preponderate relate to civil war recurrence. These findings support Hypothesis 1.¹³

Analysis of different peacekeeping profiles in different environments

Since Tables 1 and 2 do not directly show the presence of a peacekeeping mission, one should keep in mind that the sample exclusively consists of peacekeeping cases. Next, the analysis combines civil war characteristics with different peacekeeping postures. With a total of 22 cases, the number of conditions that can be concurrently investigated is limited to five (Marx & Dusa, 2011, p. 114); including all four contextual conditions and all four peacekeeping conditions is inadmissible. One established approach for reducing the number of conditions is summarizing them (Ragin, 2000, pp. 321–328). To do so, I define a more abstract condition that distinguishes between more or less challenging peacekeeping environments. Based on the numbers in Tables 1, 2, and A4 (in the online appendix), the summarizing condition “adverse context” obtains the value 0 if no more than two out of four civil war characteristics are given in their disadvantageous form, and otherwise a value of 1.¹⁴ The integrated analysis combines “adverse context” with four peacekeeping conditions that relate to the mandate (“robust”), troop density (“troops per capita”), and actor (“P5” and “UN”). None of these peacekeeping conditions proves necessary for peace survival (see Table A4 in the online appendix).

Table 3 presents the integrated results of the sufficiency analysis for peace survival. If Hypothesis 2 is correct, there should be one or more paths on which restrained peacekeeping operates in a non-adverse environment. Hypothesis 3 would be corroborated if no path exists to lasting peace with a difficult context and peacekeeping lite. Paths exhibiting strong peacekeeping in easy environments would be in accordance with Hypothesis 4.

Table 3. Integrated sufficiency analysis for peace survival.

Conditions: adverse context, UN, P5, troops per capita, robust
Solution consistency: 1.000, solution coverage: 0.933 (Burundi not covered)
Paths	Consis-tency	Consistent cases	Raw coverage	Unique coverage
1) ∼adverse context*∼P5* ∼troops per capita*∼robust	1.000	El Salvador, Guatemala, Indonesia, Mozambique, Nicaragua	0.333	0.333
2) ∼adverse context*∼UN*P5* troops per capita	1.000	Bosnia-H. (Serbs), Bosnia-H. (Croats), Georgia, Tajikistan, Yugoslavia	0.333	0.333
3) ∼adverse context*UN*∼P5* robust	1.000	Croatia 2, DRC MLC, Liberia 2, Sierra Leone	0.267	0.267

The solution is perfectly consistent and covers all cases of peace survival except Burundi, which exhibits an adverse context. All three paths exhibit a conducive peacekeeping environment. On Path 1, none of the attributes of a strong peacekeeping posture is a necessary part of this configuration that is sufficient for lasting peace. On Path 3, robust UN missions in conducive contexts consistently relate to peacekeeping success. As Paths 1 and 3 show, restrained peacekeeping works in favorable contexts—a finding that supports Hypothesis 2.

Path 2 represents cases combining a conducive context with peacekeeping missions featuring high troop density and led by a P5 state. A strong peacekeeping posture is not counterproductive in conducive environments. This is in line with Hypothesis 4.

As implied by Hypotheses 2 and 3, no configuration exists combining a difficult context and restrained peacekeeping that is consistently related to peace survival. However, there is no path that supports Hypothesis 3 in stating that peace needs a strong mission to endure in an adverse environment.

Table 4 presents the results of the integrated analysis of sufficient conditions for civil war recurrence.¹⁵ All resumptions took place in an adverse context. On the first three paths, none of the attributes of a strong peacekeeping posture are a necessary part of configurations sufficient for resumed civil war. This accords with Hypotheses 1, 2, and 4.

Table 4. Integrated sufficiency analysis for civil war recurrence.

Conditions: adverse context, UN, P5, troops per capita, robust
Solution consistency: 0.875, solution coverage: 1.000
Paths	Consis-tency	Consistent cases	Raw coverage	Unique coverage
1) adverse context*UN*∼P5* ∼troops per capita	0.800	Angola, DRC CNDP, DRC RCD, Sudan deviant case Burundi	0.571	0.429
2) adverse context*∼P5*∼troops per capita*∼robust	1.000	Angola, Sri Lanka	0.286	0.143
3) adverse context*UN*∼P5* ∼robust	1.000	Angola, Croatia 1	0.286	0.143
4) adverse context*∼UN*∼P5* troops per capita*robust	1.000	Liberia 1	0.143	0.143

Path 4 exhibits a robust mission with high troop density. At first sight, this would suggest that even a strong peacekeeping posture fails in adverse contexts—a finding that would reject Hypothesis 3. However, the first peacekeeping mission in Liberia described by Path 4 had already withdrawn when civil war recurred. As it was pulled too early, this does not demonstrate that strong peacekeeping also fails in challenging environments.

While peacekeeping lite has nearly always failed in difficult contexts, the question as to whether peacekeeping with a strong posture can preserve peace in challenging environments remains open. The data does not contain such a configuration.

In total, the QCA results support Hypotheses 1, 2, and 4 but cannot confirm Hypothesis 3.

The deviating case of Burundi indicates that peace can survive in adverse contexts even when restrained peacekeeping is in place. To make sense of this deviation, one can consider conditions beyond civil war characteristics and the peacekeeping profile. A look at further post-war developments documented in the dataset for post-civil war orders reveals some differences between Burundi and the other cases with the same configuration (Path 1 in Table 4). These can be used to explain why peace lasted in Burundi. Here, the former warring parties did not maintain separate forces throughout the post-civil war period. Moreover, Burundi was less autocratic than the other cases and exhibited a particularly far-reaching form of powersharing, including veto rights.

The finding that contextual conditions matter raises the question of whether peacekeeping makes a difference at all. If peace proves just as stable in conducive environments without peacekeeping as in conducive contexts with peacekeeping missions, one may conclude that peacekeeping is not needed. However, such an inference is inadmissible as the contextual conditions in cases with peacekeeping systematically differ from those in cases without such missions. In the dataset used for this analysis, a peacekeeping mission followed each civil war terminated by a peace agreement. Moreover, peacekeepers were deployed in 80% of the cases with a military balance at war’s end, whereas peacekeeping was carried out in less than one-third of cases with a military imbalance at this time (for more details, see Tables A13–A22 in the online appendix). The findings presented here do not support the interpretation that peacekeeping only works in contexts where it is not needed.

Conclusion

As comparative studies have mainly addressed the question of “does peacekeeping work?,” this article shifts the focus to combinations of conditions that allow peacekeeping to succeed. The findings stand opposed to the Capstone Doctrine (United Nations, 2008, pp. 31–40) and to the Report of the High-Level Independent Panel on Peace Operations (2015, para. 10) that discuss prospects for success (almost) exclusively with regard to peacekeeping profiles. This article shows that contextual conditions matter for peacekeeping success. Missions with a restrained approach succeed in advantageous contexts but fail in difficult environments. Such peacekeeping lite is defined by the presence of most or all of the following: a non-robust mandate, low troop density, and a lack of dominance by a P5 state. An adverse context is given when the following characteristics of the terminated civil war are predominant: ethnic conflict, high intensity, a military balance at war’s end, and civil war termination via ceasefire or due to low activity.

A conclusive civil war termination (a peace agreement or a military victory) is necessary for peace survival. For cases with peacekeeping, this finding contradicts the conventional wisdom that victories are better suited for preserving peace than are peace agreements.

Although missions with a strong posture do no harm in relatively conducive environments, decision-makers should opt for deploying restrained peacekeeping in such contexts. These missions demand fewer resources, are less risky than strong peacekeeping, and prove equally effective. Whether or not a strong peacekeeping posture will preserve peace despite an adverse context remains an open question, as my data lacks this configuration.

The considerable number of failed peacekeeping missions should inspire more studies on conditions that lead to success or failure. Further studies could extend the analysis to recent cases and armed conflicts with fewer than 1,000 fatalities. As my contribution limits itself to one dimension of the peacekeeping context (characteristics of the terminated civil war), future studies could (also) consider post-war conditions beyond peacekeeping, such as powersharing, the level of democracy or armed conflict in neighboring countries. With regard to the peacekeeping profile, I incorporated conditions that distinguish between restrained and strong peacekeeping. Alternatively, studies could focus on the composition or other aspects of peacekeeping missions. Nevertheless, any way one analytically models the peacekeeping profile, the highlighted civil war characteristics will continue to exert a strong influence on the peacekeeping outcome—at least for the cases and the specifications used here.

Acknowledgments

I would like to thank Evgeniya Bakalova, Matthias Dembinski, Julian Junk, Dirk Peters, Vera Rogova, Chris Ross, Judith Thorn, Jan Philipp Vatthauer, Theresa Werner, Simone Wisotzki, the editor, and the reviewers for their constructive comments. I also thank Nina Zimmer for her research assistance.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Thorsten Gromes is a Senior Researcher at the Peace Research Institute Frankfurt. He has taught at the Universities of Marburg, Frankfurt, Erfurt, and Leipzig. His current main interests are post-civil war societies and humanitarian military interventions.

ORCID

Thorsten Gromes http://orcid.org/0000-0002-3136-0901

Additional information

Funding

This work was supported by the German Research Foundation [grant number: GR 3357/2-1, GR 3357/2-2, BR 878/24-1, and BR 878/24-2].

Notes

1 It also highlights the difference between peacekeeping during and after armed conflict. More recent UN papers confirm the key difference between the environment of traditional peacekeeping and more challenging contexts with no peace to keep (Report of the High-Level Independent Panel on Peace Operations, 2015, para. 23, 104). This difference is not relevant for this article, which exclusively addresses post-civil war peacekeeping.

2 Guéhenno (2015, p. 88) also writes “the main problem is the lack of political will” among the member states.

3 http://www.pcr.uu.se/research/ucdp/definitions.

4 Implicitly, these definitions also consider whether the UN conducts a mission. As Table A1 in the online appendix shows, no UN mission is dominated by a P5 member.

5 http://ucdp.uu.se/downloads/.

6 According to UCDP, war-related violence in Croatia lasted from July 1991 until a ceasefire came into force in late 1993. In 1995, the conflict re-escalated but the number of fatalities remained narrowly below the threshold of 1,000. Strictly speaking, no civil war recurred and only one peacekeeping mission took place. Such an approach, however, ignores the drastic change in the situation in 1995 when Croatia’s government defeated the forces of the self-proclaimed Republika Srpska Krajina. To account for the fact that peacekeeping after this victory operated on a very different basis, I code two missions in Croatia.

7 Authorization by the UN Security Council, consent by the conflict parties, and peacekeeping outlasting the first post-civil war election are (nearly) constant conditions. The online appendix (Table A3) documents these and other conditions that are not considered in the analysis.

8 https://www.hsfk.de/fileadmin/HSFK/hsfk_publikationen/peacekeeping-appendix.pdf.

9 https://www.hsfk.de/publikationen/working-paper-33-case-descriptions/.

10 For details see: https://www.hsfk.de/fileadmin/HSFK/hsfk_publikationen/Codebook-post-civil-war-power-and-compromise.pdf.

11 Table A2 in the online appendix sorts the values for all metric conditions.

12 The online appendix presents all truth tables. All analyses were done with fsQCA 2.5. As I exclude “logical remainders,” the complex solution corresponds to the intermediate solution and the parsimonious solution.

13 Tables A7 and A8 in the online appendix replace relative intensity (“intensity2.5”) with absolute intensity (“50000 deaths”). The results are largely the same for peace survival and match perfectly with regard to civil war recurrence.

14 While “adverse context” is necessary for civil war recurrence, an easy peacekeeping environment is necessary for peace survival, as Table A4 in the online appendix shows.

15 Tables A11 and A12 in the online appendix show that the results hardly change when the troop-to-population ratio is replaced by the troop-to-space ratio.