Negotiating peacekeeping consent: Information and peace outcomes

Abstract

The mixed record on civil war termination shows that it is a difficult task, one fraught with uncertainty and risk. Gaining consent for peacekeeping is one strategy policymakers and scholars forward to reduce these concerns. Formal and informal work argues that allowing peacekeeping serves as a costly signal of peaceful intentions; however, these models treat peacekeeping costs as exogenous. I argue that peacekeeping costs have an endogenous element and use consent for peacekeeping missions as a proxy measure. Three conclusions are evident. It is difficult to determine whether belligerents are insincere actors in a peace process or merely distrustful, but consent can tell us whether a ceasefire is precarious and therefore more likely to fail; peacekeeping is difficult but meaningful under some conditions, and reliable information can be taken from negotiating, not just war-fighting. These results qualify the extent to which peacekeeping, with its changing emphasis on consent, can improve its outcomes.

Keywords

civil war termination consent endogeneity peacekeeping

Hailed as a necessary component for mission success, consent is a fundamental principle of peacekeeping for institutions like the United Nations (UN). In fact, Secretary-General Boutros Boutros-Ghali wrote in a supplement to An Agenda for Peace (1992):

The United Nations can be proud of the speed with which peace-keeping has evolved […], but […] respect for certain basic principles of peace-keeping are essential to its success. Three particularly important principles are the consent of the parties, impartiality and the non-use of force except in self-defence. United Nations Secretary-General (1995: 9, emphasis added)

The literature treats consent for peacekeeping as ‘permission’, and formal analysis shows that consent can be a costly signal of peaceful intentions as long as the costs imposed by the mission are large enough to prevent bluffing but small enough that a belligerent is willing to bear them to gain the benefits of peace (Diehl, 1994; Fortna & Martin, 2009; Walter, 2002). An important assumption in these models is that the costs of peacekeeping are exogenous, meaning belligerents do not influence them; they merely accept or reject them. Real peacekeeping negotiations show that this is not how the process happens.¹

Often belligerents have a great deal of influence on the ‘costliness’ of the peacekeeping mission – specifically, they can set limits on peacekeeping. I argue that this is important because the belligerents leverage consent to influence what a mission looks like and therefore how effective the mission is if sent. They do this in anticipation of what they think the peace process will be like. The statistical analysis supports this logic and provides several implications. First, consent is informative only in that it reveals when one or more belligerents may be less committed to the peace process – it does not reliably reveal sincere types as previously suggested. Second, the analysis provides concrete potential for learning from negotiating processes not just battlefield outcomes. Third, these results show another explanation for why peacekeeping is difficult – the belligerents’ strategic motivation to minimize their peacekeeping costs may lead to mission failure even when peace is desirable. Fourth, the results suggest important policy implications around recent UN initiatives, particularly the ‘responsibility to protect’ (R2P), which may have the unintended consequence of causing peacekeeping to fail more often.

What role do peacekeepers play?

Walter (1997, 2002) shows how commitment problems plague civil war termination. She argues that civil wars that do not end in partition require one or more groups to disarm as part of the settlement, which creates a natural commitment problem since the existing settlement is predicated on the pre-implementation distribution of forces. Disarmament alters the balance, which creates opportunities for the side that does not disarm to gain an advantage that it could not gain on the battlefield. She argues that third-party security guarantees are the most effective way to ensure settlement implementation because they raise the costs of reneging.

Fortna (2004) applies this logic to the role peacekeepers play as security guarantors and compliance monitors. Doyle & Sambanis (2000, 2006), contemporaneously, explore the role of post-conflict peacebuilding efforts to support peace processes, with mixed results for peacekeeping. They consider the myriad functions peacekeepers perform, including monitoring and disarmament, protecting vulnerable populations, physically separating forces, and even monitoring features of governance, like elections. All of these functions are designed to ensure that opportunistic belligerents do not take advantage of the fragile situation that exists when implementation begins. The literature on peacekeeping has expanded to examine various aspects of peacekeeping functions and mission characteristics. Some disaggregate missions and examine more granular forms of data to assess factors that influence mission outcomes (e.g. Collier, Hoeffler & Söderbom, 2008; Hultman, Kathman & Shannon, 2013, 2014), uncovering important insights on troop deployments and timing.

Peacekeeping, like other forms of humanitarian aid, can also have unintended consequences that either fail to prevent renewed conflict or inadvertently encourage it by concentrating vulnerable populations, making it easier for enemies to attack or failing to respond to violence beyond designated areas, for example (Luttwak, 1999; Hultman, 2010).² Peacekeeping missions often work to provide aid to civilians, but they may also inadvertently help sustain fighters who might otherwise agree to peace without supplies from aid.

Peacekeeping and consent

While recent research leverages greater detail about peacekeeping missions and the post-conflict environment to explain peace outcomes, few previous studies consider the demand for peacekeeping from the civil war combatants.³ Fortna & Martin’s (2009) study is the first formal analysis of peacekeeping consent as a signal of peaceful intentions in civil conflict. Their analysis emphasizes the separating equilibria to examine the conditions under which belligerents will prefer peace with peacekeepers to war and to bilateral settlements. Because peacekeeping is costly to belligerents, there are conditions in which only reliable belligerents will accept peacekeeping, thus separating themselves from unreliable ones. Their analysis also identifies conditions in which semi-separating behavior occurs – some insincere belligerents behave in the same way as sincere ones, accepting peacekeeping with nefarious motives; thus, belligerents may be bluffing when they agree to peacekeeping. An important insight from this work is that none of the belligerents are going to agree to peacekeeping that they consider too costly relative to the potential benefits of a peace process, which is one explanation for why we do not always observe consent for peacekeeping. Another is that the belligerents do not even consider peacekeeping because they know it will not be forthcoming from the international community, an explanation I control for in the analysis below. The innovation I offer is that belligerents have the opportunity to influence the costs of peacekeeping, which may increase the likelihood of peacekeeping but may paradoxically be associated with an increased risk of renewed fighting.

Allowing a peacekeeping mission can guarantee a break in the fighting since missions are usually sent where ceasefires exist or are negotiated before and deploy after a signed ceasefire. Belligerents who want only a short-term peace to regroup may need peacekeepers to get their opponent to agree to a ceasefire in the first place. Without peacekeepers, the opponent would be too suspicious of a bilateral settlement offer to agree to stop fighting. Thus, peacekeepers may make a regrouping strategy possible where none would exist without them. The temptation to pursue a regrouping strategy may be particularly acute in civil war cases in which the struggle is for as much control of the new regime as possible. Because the stakes are so high, it is reasonable to expect that high-risk strategies may be more palatable.⁴ According to reports from Secretary-General Ban Ki-moon, this is what occurred with the first observer mission (UNSMIS) sent to Syria in 2012. The UN negotiated a ceasefire agreement that included government reforms and consent from the Assad regime and the opposition groups; however, both sides renewed the violence after sourcing additional arms from external parties and reorganizing their forces (United Nations Secretary-General, 2012a,b; Gordon, 2012).

Peacekeeping also generates incentives for belligerents who do want peace to impose restrictions on peacekeeping because they lack trust in the peace process. Some groups may fear that their opponents will benefit from the peace in ways not regulated by the peacekeeping mission, like reorganizing or consolidating their combatants to fortify strategically valuable or vulnerable locations or receiving illicit support. Second, some groups may harbor distrust of the institution conducting the peacekeeping mission. One common concern is that the institution breaks its promise of security guarantees (Walter, 2002). Walter focuses specifically on when the institution fails to follow through with commitments, sending reduced missions, like the United Nations Assistance Mission for Rwanda (UNAMIR), which received less than half of the forces that had been agreed upon. Additionally, the members of the Rwandan Patriotic Front did not trust the UN as the peacekeeping provider and so sought limitations on the mission for their own sakes to protect themselves against uneven implementation (Dallaire, 2009). If trust in the peace process is lacking, then belligerents who genuinely want peace on existing settlement terms may still try to alter the costs imposed by peacekeepers not because they want to disrupt the process, but because they want to protect themselves.⁵

Unfortunately, the sincere and insincere motivations to limit peacekeeping are difficult to separate empirically. Signaling models of peacekeeping crucially rely on being able to identify whether costs and beliefs fall in the range where credible signals exist between pooling incentives in which everyone accepts peacekeeping or no one accepts peacekeeping. Measuring anticipated peacekeeping costs is difficult even ignoring strategic manipulation of costs. Fortna & Martin (2009) note these difficulties and attempt to capture costs and beliefs with proxies, but they do not explore whether those signals turned out to be credible signals or bluffing. I posit that we can learn whether some belligerents have weaker commitments to the peace process by examining negotiations surrounding peacekeeping missions.

Negotiation and information

Innovative models of negotiating behavior have led to alternative analyses of information revelation, emphasizing that the costliness of a signal can be endogenous to the expectations about what will follow.⁶ Since negotiations occur before any mission goes to support a ceasefire, it is reasonable to expect belligerents to negotiate based on expectations about the peace process. Whether belligerents fear being cheated or plan to cheat themselves, they can influence the costliness of peacekeeping with the leverage consent gives them.

Mission mandates, troop levels, and deployment locations vary widely. Troop contributing nations have also diversified. While the quantitative peacekeeping literature tends to attribute this variability to the politics within the institution, qualitative studies refine this assumption, exploring the interactive process among belligerents and institutions (e.g. Gilligan & Stedman, 2003; Gilligan & Sergenti, 2008; Fortna, 2008; Benson & Kathman, 2014; Prunier, 2008). Complex peacekeeping requires substantial negotiations between the institution and the belligerents, providing opportunities for belligerents to shape the characteristics of the mission. For example, negotiations around a UN mission to Darfur involved several high-level meetings including representatives from regional members and the five permanent members to produce the Tripoli Consensus, though the Sudanese government still harbored concerns that they wanted addressed. Next, the Department of Peacekeeping Operations negotiated with potential troop contributing nations.⁷ Negotiations around peacekeeping missions vary, but each involves rounds of negotiations with host governments and rebel factions, regional states, and, in the case of the UN, the permanent members of the Security Council, as well as nations contributing both financial and manpower resources. While belligerents can object to certain mission characteristics, the institution can still try to impose its most preferred features, but it risks critical belligerents refusing any form of consent for the mission. This occurred with Darfur, in which the Sudanese government repeatedly refused a UN mission but relented when the USA and China offered a hybrid African Union/UN mission as a compromise.

How belligerents interpret the costs of peacekeeping is salient to how it might function as a signal of commitment to peace. Peacekeeping costs might take the form of disarmament, verification, troop withdrawals, ceding territory, and/or political obligations and rules that may be onerous. A belligerent planning to use the ceasefire to regroup would seek to limit these costs or apply them unevenly to create a new advantage, for example, by limiting the ability of the mission to verify compliance. If this is the goal of one of the belligerents, then we should expect limits on peacekeeping missions to increase the chances that the peace process fails.

Belligerents may also view peacekeeping as costly because it alters the existing conditions on the ground in ways that might incentivize an opponent to cheat. Belligerents who do not trust their opponent may try to limit the costs of peacekeeping to protect themselves against cheating. The higher the costs of peacekeeping, the more likely sincere groups will try to avoid those costs by fulfilling their obligations. But fulfilling their obligations makes them vulnerable if the fighting begins again, so limiting peacekeeping reduces the costs to defend sincere belligerents in case of cheating. Both types of belligerents, those who plan renewed war and those who fear it, have an incentive to limit their vulnerability by reducing peacekeeping costs. Placing limits on the peacekeeping mission implies that one or more belligerents is not fully committed to the peace process, either from opportunistic motives or mistrust.

Hypothesis 1 (Regroup/Distrust): Restrictions on peacekeeping missions increase the likelihood that the peace process fails compared to ceasefires without consent for peacekeeping.

Two other types of consent are worth considering as well. Some missions go without all belligerents giving consent. In these partial consent cases, at least some groups have refused peacekeeping, so we should expect those situations to be more likely to fail. Refusing peacekeeping suggests these groups expect a better outcome if the war continues; thus we should expect them to be disruptive to the peace process.⁸

Hypothesis 2 (Partial consent): Partial consent on peacekeeping missions increases the likelihood that the peace process fails compared to ceasefires without consent for peacekeeping.

Second, cases in which none of the belligerents chose to impose restrictions suggests that they saw no need to alter the expected costs of peacekeeping. This scenario reduces to a standard, exogenous costs signaling model because the belligerents are not actively imposing their own constraints on the mission. The separating equilibrium from the standard signaling model suggests that consent should be associated with a reduced likelihood of failure, reflected in the hypothesis below.⁹

Hypothesis 3 (No restrictions): Unrestricted consent for peacekeeping decreases the likelihood that the peace process fails compared to ceasefires without consent for peacekeeping.

Below, I describe several empirical tests across different types of consent, all of which take into account the endogeneity between the choice of consent and the expected outcome of the peace process.

Research design

To determine whether the negotiating process surrounding peacekeeping missions can be informative of belligerent expectations, I collected information on requests for peacekeeping in civil conflicts since 1947. While nothing about the signaling dynamics necessarily excludes interstate conflicts, I focused on civil conflicts since, as Walter (2002) suggests, the security dilemma is heightened in these situations. Peacekeeping missions sent to interstate conflicts have a decidedly interpositional character, lacking the security and enforcement dynamics that are central to missions sent to civil war cases.

Building on Fortna’s (2008) ceasefires data, cases include ceasefires that last at least one month in civil conflicts and follow-on peacekeeping missions since the beginning of the UN system through 2011.¹⁰ The total number of cases currently coded is over 200; however, the total number of cases used across three analyses is 172, due in part to missing data as well as ceasefires that are coded but are ineligible as attempts to reach peace.¹¹ Fortna’s analysis already excludes very short-lived cases in the data, but the dataset also includes a few ceasefires that were negotiated for reasons other than peace negotiations. If we include the ineligible ceasefires in the dataset, then their ‘failures’ (renewed fighting) inflate the data with failed cases in which war resumes because the purpose of the ceasefire had been met.¹² I also added ceasefires in Somalia since some of these events were missing in the source data.

Coding consent

The primary concept of interest is consent and the limits belligerents choose to put on peacekeeping missions. I measured consent by collecting details of what type of consent was given – specifically, the types of restrictions put on missions by the belligerents. The institution sending a mission may also impose its own restrictions, but these are not included in the codings.¹³ To determine whether the belligerents were responsible for limitations on a mission, I consulted agreement texts, if they existed, Secretary-General reports, and conflict histories. I coded whether belligerents limited the tasks the mission could perform (i.e. restrictions on the mission’s mandate). I also coded whether the belligerents allowed peacekeepers full access across their territory or confined their operations to specific areas. Finally, I coded whether the belligerents negotiated requirements on the number or source of troops. Cases in which only some of the belligerents gave consent are coded as partial consent since at least one major group did not want peacekeeping at all. I used these details to create the Consent type variable.¹⁴

Consent type was coded 1 if the mission did not have consent from all parties deemed crucial to the peace process according to the institution sending the mission. This value captures cases of ‘partial’ consent. Consent type was coded 2 if all parties gave consent but some or all imposed some kind of restrictions on the mission, as described above. Finally, Consent type was coded 3 if the mission received full, unrestricted (permissive) consent from all parties. Ceasefires with no mention of peacekeeping or refusal by all parties to accept offers of peacekeeping are coded as 0, capturing intraconflict peace processes in which no belligerent pursued peacekeeping.¹⁵

Coding success and failure

There are many ways to define peace success, several of which focus on the outcomes of third-party attempts to maintain the peace. In the UN’s case, missions that are sent tend to be the focus, and the Secretariat has at times defined success as having carried out the mandate they were given, even if the war restarted.¹⁶ Others have considered not just the absence of war (negative peace) but the existence of stable government or improved social welfare (positive peace). I adopt the standard of whether violence resumes as a definition of success, a very basic standard that is somewhat easier to define and measure than various forms of positive peace. Because I consider whether conflicts without peacekeeping are similar to or different from those with peacekeeping, the standard cannot focus on a mission mandate since these are not observed if there is no mission. Fighting is an indicator that at least one side attempted to ‘renegotiate’ the peace arrangements.

There is a temporal dimension to coding ceasefire outcomes as well. Doyle & Sambanis (2006) use a measure of whether peace persisted until two years after the ‘peace stimulus’, meaning after the mission left the conflict, as their primary dependent variable.^17,18

I use two different measures to capture whether the mission was successful or not. The first dependent variable follows Doyle & Sambanis’s (2006) coding rule (hereafter: D&S rule) and records a peace failure if fighting resumed within two years after the ceasefire was signed absent peacekeeping or two years after the mission ended if peacekeeping took place.¹⁹ One criticism of the two-year time limit measure is that it sets an arbitrary date at which peace is hailed as a success, when many peace processes may fall apart three or more years after the mission leaves.²⁰ To address the possibility that violence may resume beyond the two-year limit set by Doyle & Sambanis’s (2006) coding rules, I also code whether belligerents renew the war ten years beyond the ceasefire or mission. The variable is coded 1 if violence among the same groups begins again. The level of violence must be severe enough to threaten the government or include a major military offensive that restarts the fighting. There must be multiple fatalities and a reaction by the interim or established government that reflects some threat to its authority.²¹ These two variables together have the advantage of providing a strict definition of violence after the peace attempt and a broader notion of peace success that captures failures that happen during the mission and long after.

Additional variables²²

Previous research identifies many factors that are associated with renewed violence after breaks in the fighting. I include several variables that capture general conditions that make settlement more or less difficult. One of the most important predictors of renewed violence is the condition of the war when the ceasefire occurred. There may be a clear winner, a negotiated settlement or a truce, in which nothing is settled. Prior theories have suggested that belligerents caught in a costly stalemate will see peace as being more beneficial than returning to a war that neither can win (Licklider, 1993). Others suggest that victory for one side means that the losing side cannot challenge the winner, making peace much more likely to endure (Fortna, 2008). Walter (2002) demonstrates that stalemate is likely to bring parties to the negotiating table, but it is indeterminate whether belligerents will choose to implement the agreement because it depends on whether the settlement addresses the commitment problem that demobilization creates. The discussion is further complicated by claims that wars end when the belligerents have learned enough about their opponent for their expectations to converge on a settlement that both prefer to fighting.²³ Each outcome is coded as a dichotomous variable using Fortna’s (2008) coding rules, and captures whether the war ended in victory for one side, a negotiated settlement or a truce – cases in which there was neither victory nor negotiated settlement.²⁴ Victory implies complete information and thus minimizes the odds of renewed war, all else equal. A negotiated settlement suggests that the belligerents learned enough to reach agreement, but it remains possible for one or more belligerents to challenge the settlement in the future. Truces have not settled the issues and reflect merely a willingness to stop the fighting. Thus, compared to truces, we should expect wars that end in victory for one side or negotiated settlement to be less likely to experience renewed violence.

I use the natural log of total deaths (battle-deaths and civilian casualties) to measure war costs. I updated Fortna’s data using UN High Commissioner for Refugees (UNHCR) reports. More deaths may discourage renewed war because the war has grown so costly that peace is a more desirable option. Conversely, more deaths may create resentment that is likely to lengthen the war and make peace more fragile, reflecting a positive relationship. It is possible that costlier war and greater resentment push against each other, which would result in no discernible effect of casualties. I also use war duration as a proxy measure for the costliness of the war. Longer wars are more costly, but they may not generate the same resentment or mistrust that large numbers of casualties may create (Doyle & Sambanis, 2006; Fortna, 2008; Gilligan & Sergenti, 2008; Gilligan & Stedman, 2003).

Previous analyses show that peacekeeping missions are less likely to go to places where the government army is strong (Fortna, 2008; Gilligan & Sergenti, 2008; Gilligan & Stedman, 2003). I include this measure as an important factor that influences where peacekeepers are sent. Stronger governments may prefer to keep third parties out and manage the conflict themselves, and through this dynamic we might expect stronger government armies to be associated with peace. I used the Correlates of War (COW, 2010) National Material Capabilities data to code the size of the government’s military (measured in thousands of troops) and the SIPRI Yearbooks (1972–2009) for data not covered by COW.²⁵ Any other missing cases were coded from news sources.

Some conflicts may be more difficult to solve merely because there are more groups to satisfy. As the number of groups bargaining increases, the information and commitment problems grow, making it much more difficult to locate a deal that all groups would prefer to fighting. With more factions, it is also more likely that an important group is excluded from the negotiating process because they refuse to participate. As Doyle & Sambanis (2006) argue, spoilers can disrupt the process with a few acts of violence that generate mistrust among the principal groups trying to negotiate a deal. Thus, the greater the number of factions, the more likely the conflict is to return to violence. The number of factions is measured dichotomously using Doyle & Sambanis’s (2000) definition, with 0 capturing conflicts with only two groups and 1 capturing conflicts with three or more groups. I also use a measure of whether a neighbor intervenes in the conflict. Near neighbors often involve themselves in next-door conflicts either because of spillover effects and refugee flows or because conflicts present an opportunity to influence a nearby regime (Fortna, 2008). Neighbors supporting one side in the fighting may encourage renewed conflict if they are a source of additional war-fighting support.

I consider the relationships between a civil war nation and great powers, represented by the permanent members of the Security Council. Using the Bailey, Strezhnev & Voeten (2017) data, I calculate the differences between a civil war nation’s ideal point and the closest P5 member and farthest P5 member. I also use an existing measure which captures whether a major power was involved in the civil war. These dynamics control for whether backing from outside actors influenced decisions to give consent and to maintain peace.

Finally, I also include a dichotomous measure of peacekeeping to capture whether the institution actually sent peacekeepers. If our understanding of what peacekeeping does is correct, it should decrease the chances that peace fails.

Methodology

Belligerents deciding whether to request or consent to peacekeeping have expectations about what the peace process will be like based on what has occurred. The decision is endogenous, affected by the war-fighting process and expectations about the peace. Thus, the empirical model must take into account the possibility of endogeneity bias in the estimates. For these analyses, I use bivariate probit analysis. This estimation procedure requires a single dichotomous, endogenous variable and dichotomous outcome variable. Because the type of consent is categorical, I break up the categories into dichotomous codings that compare each type of ‘granted’ consent to the ‘no-consent’ category.²⁶ In other words, I estimate a bivariate probit on three separate subsets of the data: partial and no consent cases, restricted and no consent cases, and unrestricted and no consent cases.²⁷ I also bootstrap the standard errors for the analyses using the Doyle & Sambanis (2006) failure rule purely for producing predicted probabilities to use in the substantive discussion below.²⁸

The key difficulty in addressing endogeneity with an instrumental variable procedure is finding suitable instruments that are related to the endogenous variables but are not systematically related to the outcome variable, except through the endogenous variables.²⁹ At this stage, I focus on a factor that is unrelated to the civil conflict but is likely to be an important factor for getting a peacekeeping mission approved; specifically, I looked to the intra-institutional politics. A robust literature on the UN Security Council shows that the relationships between and among the Security Council members affects whether these institutions will reach agreement to take multilateral action for any given issue (Beardsley & Schmidt, 2012; Allen & Yuen, 2014). I carry these insights over to explain when consent for peacekeeping is even possible. Using ideal point measures from Bailey, Strezhnev & Voeten (2017), I measure the dispersion among the UN Security Council permanent members using the coefficient of variation among their ideal points. This measure captures the environment in which peacekeeping missions (and therefore the possibility of granting consent) may be agreed to, but it should not be systematically related to the belligerents’ decisions to renew fighting except through the consent variables. The more the P5 cluster together (smaller coefficient of variation), the more likely they will be able to reach agreement on a peacekeeping resolution, which means consent of some sort may be more likely because peacekeeping is possible.

Results

I present the results of the bivariate probit analyses below and conduct substantive analyses of the relationships between consent, peacekeeping, and peace outcomes. Using two measures of ceasefire outcomes, I compare different time horizons. Implications follow the results.

Negotiated consent and peace failure

Tables I and II present results for peace outcomes following the D&S rule and the ten-year rule, respectively. Models testing the effects of partial consent and restricted consent fail the test of exogeneity in all but the unrestricted case for the D&S rule, implying that there is endogeneity that needs to be accounted for, just as the theory predicts. The evidence suggests that the belligerents are deciding on their willingness to admit peacekeepers based on their expectations about the peace process. Further, the result on the instrumental variable is significant and in the expected direction in almost all models.³⁰

Using the estimates from Table I, I estimated predicted probabilities of peace failure for ceasefires that

Table I.

Bivariate probit estimations for each type of consent (D&S rule)

	(1)		(2)		(3)
	Partial		Restricted		Unrestricted
	Coeff.	SE	Coeff.	SE	Coeff.	SE
Consent decision
Victory	1.234*	0.493	–1.164***	0.322	–0.244	0.518
Treaty	1.472***	0.446	0.867**	0.264	0.821*	0.335
ln(Deaths)	0.129	0.107	0.245**	0.0848	–0.037	0.0886
ln(Gov’t army strength)	–0.121	0.185	–0.262*	0.122	–0.369*	0.145
Factions	0.448	0.367	0.180	0.232	0.255	0.216
Neighbor	0.854**	0.321	0.251	0.307	0.991***	0.282
War duration	–0.0795*	0.0394	–0.0678**	0.025	0.0338	0.0289
IP distance to nearest P5	–0.00437	0.405	0.0154	0.588	0.519	0.415
IP distance to farthest P5	–0.495	0.445	–0.678*	0.290	–1.067*	0.464
Major power involvement	0.595^†	0.357	0.402	0.430	–0.369	0.422
Coeff. of variation IP	–9.690^†	4.961	–15.31***	3.570	–11.76**	3.635
Constant	1.124	2.933	4.391**	1.687	7.015**	2.310
Peace failure
Partial consent	2.378***	0.701	–	–	–	–
Restricted consent	–	–	3.476***	0.952	–	–
Unrestricted consent	–	–	–	–	2.670^†	1.446
Peacekeeping (dichot.)	–0.0258	0.538	–1.481^†	0.849	–2.544***	0.660
Victory	–1.705**	0.596	–1.317^†	0.680	–1.671***	0.495
Treaty	–1.660**	0.609	–1.446*	0.703	–1.294*	0.526
ln(Deaths)	0.0407	0.106	0.123	0.119	0.0971	0.0998
ln(Gov’t army strength)	–0.183	0.127	0.00822	0.157	0.00933	0.166
Factions	0.207	0.416	0.259	0.395	0.361	0.330
Neighbor	0.119	0.423	0.179	0.453	0.705	0.438
War duration	–0.00303	0.029	–0.0204	0.0292	–0.0285	0.0284
IP distance to nearest P5	–0.0529	0.492	–0.368	0.491	–0.667	0.527
IP distance to farthest P5	0.357	0.348	0.499	0.363	0.422	0.349
Major power involvement	–0.403	0.314	–0.197	0.406	–0.536	0.383
Constant	–0.376	1.640	–2.833	2.053	–1.880	1.869
ρ	–1	0	–1	6.45x10¹⁰	–0.116	12.540
N	106		119		125
χ²	133.1		185.1		123.9
p	1.73x10⁻¹⁷		2.68x10⁻²⁷		8.10x10⁻¹⁶
LR Test of ρ = 0:
χ²(1)	7.41945		7.90732		0.054291
p > χ²	0.0065		0.0049		0.8158
Replications	1,000		955		1,000

Bootstrapped standard errors. ^† p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

accompany treaties and truces in Tables III and IV, respectively. The likelihood of peace failure when partial consent is given with a treaty but without peacekeeping is 83%. The same conditions without consent yields a chance of failure that is much lower, around 12%. When a peacekeeping mission goes to the conflict, the chance of failure given partial consent is 82%. The lesson to be drawn from these results is that partial consent is not enough agreement to give peacekeeping much better odds of succeeding, supporting Hypothesis 2. Thus, the UN’s first principle to secure broad consent is an important one, given the particularly precarious nature of ceasefires that only get partial consent and the fact that a peacekeeping mission does not offer much improvement over failure odds. Partial consent that comes with truces is even more precarious, yielding chance of failure at 99% with some consent versus only 69% without consent.

Table II.

Bivariate probit estimations for each type of consent (ten-year rule)

	(1)		(2)		(3)
	Partial		Restricted		Unrestricted
	Coeff.	SE	Coeff.	SE	Coeff.	SE
Consent decision
Victory	0.929*	0.470	–1.053*	0.507	0.132	0.431
Treaty	1.326**	0.502	0.990*	0.414	1.050**	0.401
ln(Deaths)	0.117	0.0854	0.110	0.0942	–0.0584	0.0734
ln(Gov’t army strength)	–0.0376	0.113	–0.186	0.114	–0.203	0.114
Factions	0.302	0.346	0.222	0.327	0.0809	0.312
Neighbor	0.875*	0.411	0.680^†	0.394	1.289***	0.341
War duration	–0.0949**	0.036	–0.0598*	0.029	0.0103	0.0263
IP distance to nearest P5	–0.219	0.468	–0.197	0.628	0.328	0.432
IP distance to farthest P5	–0.698*	0.306	–0.603^†	0.363	–0.841*	0.363
Major power involvement	0.586*	0.278	0.452	0.435	–0.188	0.430
Coeff. of variation IP	–10.70***	3.024	–13.92***	3.708	–12.26***	3.178
Constant	2.488	1.700	4.693*	1.926	6.079**	1.927
Peace failure
Partial consent	2.653***	0.437	–	–	–	–
Restricted consent	–	–	2.223***	0.570	–	–
Unrestricted consent	–	–	–	–	2.875***	0.528
Peacekeeping (dichot.)	–1.263**	0.482	–1.023	0.727	–1.553**	0.577
Victory	–1.438***	0.386	–1.086**	0.381	–0.817*	0.341
Treaty	–1.044*	0.417	–0.879*	0.384	–0.796*	0.345
ln(Deaths)	0.080	0.0682	0.127^†	0.071	0.0825	0.0636
ln(Gov’t army strength)	–0.0545	0.0905	0.0402^†	0.100	0.106	0.0871
Factions	0.171	0.281	0.251	0.287	0.270	0.259
Neighbor	–0.0506	0.301	0.305	0.362	0.00088	0.299
War duration	–0.0224	0.0195	–0.0238	0.020	–0.0259	0.0182
IP distance to nearest P5	–0.25	0.354	–0.359^†	0.396	–0.618	0.344
IP distance to farthest P5	0.044	0.220	0.0999	0.238	0.0749	0.238
Major power involvement	–0.188	0.264	–0.231	0.310	–0.451	0.288
Constant	0.363	1.061	–1.127	1.216	–0.847	1.097
ρ	–1	2.45x10⁻¹²	–1	8.24x10⁻⁹	–1	3.03x10⁻¹⁰
N	105		118		122
χ²	87.38		75.45		97.14
p	1.95x10⁻⁹		1.70x10⁻⁷		4.37x10⁻¹¹
LR Test of ρ = 0:
χ²(1)	9.37519		11.137		3.93676
p > χ²	0.0022		0.0008		0.0472

^† p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

Table III.

Predicted probability of peace failure and type of consent with a treaty

	Partial consent model	Restricted consent model	Unrestricted consent model
Consent not given	12%	9%	37%
Consent without peacekeeping	83%	92%	94%
Consent with peacekeeping	82%	67%	32%

Once again, peacekeeping does not change the chances of peace failure noticeably when a truce is the only thing accompanying the ceasefire. If the war ends in a decisive victory, the difference in risk is similar to treaties.³¹

Table IV.

Predicted probability of peace failure and type of consent with a truce

	Partial consent model	Restricted consent model	Unrestricted consent model
Consent not given	69%	59%	78%
Consent without peacekeeping	99%	99%	99%
Consent with peacekeeping	99%	69%	74%

For restricted consent with a treaty and without peacekeeping, there is a 92% chance of peace failing when consent is given compared to 9% when it is not. When peacekeeping does happen, the chances are significantly reduced, down to 67%. For a truce without peacekeeping, the chances of failure are 99% with restricted consent compared to 59% without consent. Peacekeeping reduces the chances of failure only 2%, however. Across both partial consent and restricted consent, truces are at much greater risk of failure, so much so that peacekeeping does not materially change the risk of failure. These results suggest support for Hypothesis 1.

The results for unrestricted consent are less clear and have a sign on the coefficient opposite to what I expected in Hypothesis 3. I cannot reject the null hypothesis that ρ = 0, so the evidence does not support the suspicion of endogeneity in these cases. Further, the coefficient does not meet standard levels of significance, but it does meet a slightly lower standard of 0.10. There is, however, a strong effect in the short term for peacekeeping missions. While the risk of conflict recurrence is moderate given a treaty without consent or peacekeeping (37%), when there is consent but no peacekeeping, the chance of failure is 94%, but peacekeeping reduces that chance to 32%.³² Once again, truces are more precarious. When belligerents give unrestricted consent for peacekeeping, it weakly suggests a precarious peace, contrary to expectation, but the effect of peacekeeping seems to be substantial in these cases, nonetheless.

Comparing the results from Table I and Table II, the effect of peacekeeping wanes somewhat in conditions that produce restricted or unrestricted consent. Peacekeeping appears much more strongly in the case of partial consent in the longer term; however, these effects may reflect changes in the small number of cases with partial consent. Also notable are the longer-term effects of the war outcomes. Both decisive victory and treaty outcomes are more likely to remain peaceful than truces, suggesting support for the argument that peace happens when expectations about continuing the war compared to the peace converge. The usual factors that we often think affect belligerents’ desire for peace, such as how many deaths have occurred, seem to play a minor role at best. Further, there is evidence that some factors only carry a short-term effect, such as the strength of the government’s army, and then the effect is relegated to the consent decision, not the decision to renew violence. Finally, lengthy wars are less likely to result in consent for peacekeeping, but they appear to have little effect on the decision to renew the war in both the short and long term. Together, these relationships suggest that ‘sunk cost’ factors, like war duration and deaths, may affect the decision to grant consent, but they offer little to guide expectations about renewed violence and so have no systematic effect on that decision across any of the models.

Implications

I have argued above that conceptualizing consent as a negotiable cost for the belligerents makes consent somewhat informative as a flag for acutely precarious ceasefire situations. We cannot measure belligerents’ level of commitment to peace directly, but the constraints put on a peacekeeping mission may capture some element of commitment. The empirical models show that the informational value of consent is mostly consistent with those expectations – limits on consent suggest failure is likely. A noteworthy result, however, is that full, unrestricted consent is no guarantee of peace any more than situations without offers or requests for peacekeeping and is somewhat indicative of a precarious peace process as well, particularly in the long run. While it is still true that more permissive consent with peacekeeping is associated with better chances for peace compared to more restrictive consent situations, none of these peacekeeping scenarios performs much better than ceasefires with no consent. Instead, the effects of consent and peacekeeping bring the prospects for peace to the same level as ceasefires with no consent. This demonstrates the informational value of consent to identify more precarious ceasefires and the value of peacekeeping to support the peace processes. When belligerents give consent for peacekeeping but do not get the support, the prospects for peace are dismal.

This result highlights why criticism of UN and other peacekeeping outcomes may be misplaced regarding success. This analysis does not show that peacekeeping performs better than ceasefires without consent (and therefore without peacekeeping). But if we compare ceasefires where belligerents gave full consent (no restrictions) and got peacekeeping to ceasefires with full consent where there was no peacekeeping, we see a positive effect on the likelihood of peace, unlike what occurs with restricted forms of consent and peacekeeping.³³ With full consent, peacekeeping makes the chances for peace in an otherwise precarious case about the same as, or slightly better than, a ceasefire that offered no consent at all. Given that the effects of peacekeeping are negligible for restricted forms of consent, the takeaway is that consent can tell us a great deal about what to expect in the peace process.

These results also support the argument that peacekeeping goes to the difficult cases. Prior explanations for the difficulty of peacekeeping have focused solely on the conflict characteristics or the mission features, but these results suggest we should consider the strategic incentives of the belligerents as well. Part of the problem may be that peacekeepers are sometimes sent despite the fact that some belligerents may actually be signaling that they do not want peace on existing terms by refusing consent or placing limits on missions. Those groups placing restrictions on missions may want to appear cooperative to get embargoes or sanctions relaxed, but they do not want to bear the full costs of a large, invasive peacekeeping mission so they negotiate limits with the institution to protect their assets. Finally, the belligerents may simply be unable to trust each other or the peacekeepers, and as protection against implementing a disadvantageous peace (or worse a renewed war at a serious disadvantage), they limit aspects of the mission to reduce their own vulnerability. Thus, peacekeeping is indeed difficult, as others have argued, but not just because the civil wars were longer or more people died, but because the nature of peacekeeping itself may create an opportunity to regroup in a better fighting position for one or more of the belligerents or at least create the fear of this possibility.

The implication for institutions with peacekeeping functions is that consent should not be understood as the key to enduring peace. If the most expansive form of consent cannot guarantee peace, why pursue consent at all? Extreme arguments would suggest either invasion-type peacekeeping that ignores consent or no peacekeeping at all, but neither of these interpretations gets us closer to peace, and even these models reveal a positive effect for peacekeeping under some conditions. Instead, this evidence implies that institutions should pursue consent for informational reasons. Negotiators might use the information from peacekeeping negotiations to gauge whether a conflict is ready for a mission or would benefit from a costly delay. Using consent as an informational tool to identify who might be weakly committed to peace suggests that the UN’s recent de-emphasis on consent as a primary pillar for peace, justified by the R2P,³⁴ may increase the occurrence of peacekeeping failure because the UN is sending peacekeeping missions to conflicts in which some belligerents are refusing consent or are tying the hands of the mission up front.

Rather than intervene without agreement, institutions may find more success if they focus their efforts on altering other incentives to make sustainable peace more attractive. The UN has pursued this strategy with the Peacebuilding Support Office. These efforts might alter the dynamics so that the benefits of peace make genuine cooperation attractive and therefore more reliable. Further, consent may serve a wider purpose in legitimizing institutional actions for international audiences, which may have great bearing on material support for those actions; however, an analysis of these logics is beyond the scope of this article.³⁵

Two other points are worth highlighting. Many studies focus on the politics within the institution that affect the content of resolutions and missions, and indeed, these behaviors influence the nature of peacekeeping and its outcomes, but this also means space exists for belligerents to shape peacekeeping missions, especially given the disparate preferences of the Council members. Belligerents have some leverage against powerful actors even when, and often because, the institution forces powerful states to work in tandem. Second, the argument here suggests that belligerents may manipulate peacekeeping to protect themselves, a strategic factor that makes peace harder to maintain. It also demonstrates, however, that there may be more reliable clues to belligerent commitments to peace based on their negotiating behavior. As Slantchev (2003) argues, we can learn additional information beyond war-fighting by observing behavior at the negotiating table. A practical implication is that information gained at the bargaining table could be used to direct the finite resources of peacekeeping to missions where they have the best chance to prevent peace from failing, all else equal.

Conclusion

This analysis shows that consent for peacekeeping is informative, but not as we previously thought. Instead of identifying belligerents who are sincere in the pursuit of peace, consent can show us situations in which the peace is particularly precarious – either because there are real incentives to renege or there are fears that an opponent plans to do so. It further shows that without full consent, peacekeeping makes little difference to outcomes, but with full consent, peacekeeping can improve odds of peace to be on par with a natural, bilateral settlement. Further, consent at the bargaining table gives some information about belligerents’ expectations around the peace process, providing information that cannot be gained through fighting. These behaviors shed light on the prospects for peace in ways previously unexplored.

Footnotes

Replication data

The dataset and do-files for the empirical analysis in this article, along with the Online appendix which details the coding procedures, can be found at . Coding notes are available from the author.

Acknowledgements

Thank you to Susan Allen, Amanda Licht, Emily Ritter, Meg Shannon, Melissa Willard-Foster, Bert Johnson, Sarah Stroup, Pat Regan, John Western, and three anonymous reviewers for comments on earlier drafts.

Funding

This material is based upon work supported by the National Science Foundation under grant no. 1827373.

ORCID iD

Amy Yuen

Notes

References

Allen

Susan

Yuen

Amy

(2014) The politics of peacekeeping: UN Security Council oversight across peacekeeping missions. International Studies Quarterly 58(3): 621–632.

Anderson

Mary

(1999) Do No Harm: How Aid Can Support Peace or War. Boulder, CO: Lynne Rienner.

Bailey

Michael

Strezhnev

Anton

Voeten

Erik

(2017) Estimating dynamic state preferences from United Nations voting data. Journal of Conflict Resolution 61(2): 430–456.

Beardsley

Kyle

Schmidt

Holger

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

Benson

Michelle

Kathman

Jacob

(2014) United Nations bias and force commitments in civil conflicts. Journal of Politics 76(2): 350–363.

Chapman

Terrence

(2011) Securing Approval: Domestic Politics and Multilateral Authorization for War. Chicago, IL: University of Chicago Press.

Collier

Paul

Hoeffler

Anke

Söderbom

Måns

(2008) Post-conflict risks. Journal of Peace Research 45(4): 461–478.

Cooley

Alexander

Ron

James

(2002) The NGO scramble: Organizational insecurity and the political economy of transnational action. International Security 27(1): 5–39.

Correlates of War Project (2010) National material capabilities data, Version 4.0 (http://www.correlatesofwar.org/data-sets/national-material-capabilities).

10.

Corwin

Phillip

(1999) Dubious Mandate: A Memoir of the UN in Bosnia, Summer 1995. Durham, NC: Duke University Press.

11.

Dallaire

Romeo

(2009) Shake Hands with the Devil: The Failure of Humanity in Rwanda. Toronto: Random House of Canada.

12.

de Waal

Alex

(1998) Famine Crimes: Politics and the Disaster Relief Industry. Bloomington, IN: Indiana University Press.

13.

Diehl

Paul F

(1994) International Peacekeeping. Baltimore, MD: Johns Hopkins University Press.

14.

Downs

George

Rocke

David

(1994) Conflict, agency, and gambling for resurrection: The principal-agent problem goes to war. American Journal of Political Science 38(2): 362–380.

15.

Doyle

Michael W

Sambanis

Nicholas

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

16.

Doyle

Michael W

Sambanis

Nicholas

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

17.

Fearon

James D

(1995) Rationalist explanations for war. International Organization 49(3): 379–414.

18.

Fortna

Virginia Page

(2004) Does peacekeeping keep peace? International intervention and the duration of peace after civil war. International Studies Quarterly 48(2): 269–292.

19.

Fortna

Virginia Page

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

20.

Fortna

Virginia Page

Martin

Lisa

(2009) Peacekeepers as signals: The demand for international peacekeeping in civil wars. In: Milner

Helen V

Moravcsik

Andrew

(eds) Power, Interdependence, and Non-State Actors in World Politics: Research Frontiers. Princeton, NJ: Princeton University Press, 87–107.

21.

Gilligan

Michael

Sergenti

Ernest

(2008) Do UN interventions cause peace? Using matching to improve causal inference. Quarterly Journal of Political Science 3(2): 89–122.

22.

Gilligan

Michael

Stedman

Stephen John

(2003) Where do the peacekeepers go? International Studies Review 5(4): 37–54.

23.

Goemans

Hein

(2000) War and Punishment. Princeton, NJ: Princeton University Press.

24.

Gordon

Michael R

(2012) Iran supplying Syrian military via Iraqi airspace. New York Times 4 September.

25.

Hultman

Lisa

(2010) Keeping peace or spurring violence? Unintended effects of peace operations on violence against civilians. Civil Wars 12(1–2): 29–46.

26.

Hultman

Lisa

Kathman

Jacob

Shannon

Megan

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

27.

Hultman

Lisa

Kathman

Jacob

Shannon

Megan

(2014) Beyond keeping the peace: United Nations effectiveness in the midst of fighting. American Political Science Review 108(4): 737–753.

28.

King

Gary

Nielsen

Richard

(2018) Why propensity scores should not be used for matching. Working paper (https://gking.harvard.edu/files/gking/files/psnot.pdf).

29.

King

Gary

Tomz

Michael

Wittenberg

Jason

(2000) Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science 44(2): 347–361.

30.

Kydd

Andrew

(2000) Trust, reassurance, and cooperation. International Organization 54(2): 325–357.

31.

Kydd

Andrew

Walter

Barbara

(2002) Sabotaging the peace: The politics of extremist violence. International Organization 56(2): 263–296.

32.

Licklider

Roy

(1993) Stopping the Killing: How Civil Wars End. New York: New York University Press.

33.

Lischer

Sarah

(2003) Collateral damage: Humanitarian assistance as a cause of conflict. International Security 28(1): 79–109.

34.

Lischer

Sarah

(2005) Dangerous Sanctuaries: Refugee Camps, Civil War, and the Dilemmas of Humanitarian Aid. Ithaca, NY: Cornell University Press.

35.

Luttwak

Edward N

(1999) Give war a chance. Foreign Affairs 78(4): 36–44.

36.

Peña

Javier Márquez

(2014) moreClarify: Stata module to estimate quantities of interest through simulation and resampling methods. Statistical Software Components S457851, Boston College Department of Economics (http://javier-marquez.org/software/moreclarify/).

37.

Prunier

Gérard

(2008) Darfur: A 21st Century Genocide, 3rd edition. Ithaca, NY: Cornell University Press.

38.

Roodman

David

(2011) Fitting fully observed recursive mixed-process models with cmp. Stata Journal 11(2): 159–206.

39.

Slantchev

Branislav

(2003) The principle of convergence in wartime negotiations. American Political Science Review 97(4): 621–632.

40.

Stockholm International Peace Research Institute (1972–2009) SIPRI Yearbook: World Armaments and Disarmament. Stockholm: Almquist & Wiksell.

41.

Terry

Fiona

(2002) Condemned to Repeat? The Paradox of Humanitarian Action. Ithaca, NY: Cornell University Press.

42.

United Nations Department of Peacekeeping Operations (2008) United Nations peacekeeping operations: Principles and guidelines (https://peacekeeping.un.org/sites/default/files/peacekeeping/en/capstone_eng.pdf).

43.

United Nations Department of Peacekeeping Operations (2010) Second generation disarmament, demobilization and reintegration (DDR) practices in peace operations (https://peacekeeping.un.org/sites/default/files/2gddr_eng_with_cover.pdf).

44.

United Nations General Assembly (2005) 2005 World Summit Outcome. A/60/L.1.

45.

United Nations Secretary-General (1992) An agenda for peace: Preventive diplomacy, peacemaking and peace-keeping. A/47/277-S/24111.

46.

United Nations Secretary-General (1995) Supplement to ‘An agenda for peace’: Position paper of the Secretary-General on the occasion of the fiftieth anniversary of the United Nations. A/50/60-S/1995/1.

47.

United Nations Secretary-General (2012a) Letter date 19 April 2012 from the Secretary-General addressed to the President of the Security Council. S/2012/238.

48.

United Nations Secretary-General (2012b) Report of the Secretary-General on the implementation of Security Council Resolution 2043 (2012). S/2012/523.

49.

United Nations Secretary-General (2015) Report of the Secretary-General on the protection of civilians in armed conflict. S/2015/453.

50.

Voeten

Erik

(2005) The political origins of the UN Security Council’s ability to legitimize the use of force. International Organization 59(3): 527–557.

51.

Walter

Barbara

(1997) The critical barriers to civil war settlement. International Organization 51(3): 335–364.

52.

Walter

Barbara

(2002) Committing to Peace: The Successful Settlement of Civil Wars. Princeton, NJ: Princeton University Press.

53.

Werner

Suzanne

Yuen

Amy

(2005) Making and keeping peace. International Organization 59(2): 261–292.

Supplementary Material

Please find the following supplemental material available below.

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

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

0.00 MB

Negotiating peacekeeping consent: Information and peace outcomes

Abstract

Keywords

What role do peacekeepers play?

Peacekeeping and consent

Negotiation and information

Research design

Coding consent

Coding success and failure

Additional variables 22

Methodology

Results

Negotiated consent and peace failure

Implications

Conclusion

Footnotes

Replication data

Acknowledgements

Funding

ORCID iD

Notes

References

Supplementary Material

Additional variables²²