Keeping the peace?

(1)

Robin Sällström Bachelor Thesis

Peace and Conflict Studies C

Department of Peace and Conflict Studies Uppsala University

January 2021

Supervisor: Corinne Bara Word Count: 15 851 words

Keeping the peace?

(2)

2

Content§

1. Introduction: ... 4

2. Theory: ... 5

2.1 Previous research on peacekeeping: ... 5

2.2 Research Gaps: ... 10

2.3 Theoretical Framework: ... 11

2.3.1 UN and NATO security guarantees: ... 11

2.3.2 UN and NATO persuasion: ... 13

2.4 Summary of theoretical argument and hypothesis: ... 15

3. Research design: ... 16

3.1 Research Design and Method: ... 16

3.2 Case Selection:... 18

3.3 Operationalization: ... 22

3.3.1 Independent variable: ... 22

3.3.2 Dependent variable: ... 23

3.3.3 Mechanisms: ... 25

3.4 Source criticism and data collection: ... 30

4. Results and analysis: ... 30

4.1 KFOR June 1999 – December 2005: ... 31

4.1.2 Dependent Variable: ... 32

4.2 UNPROFOR April 1993 – November 1995: ... 39

4.2.2 Dependent Variable: ... 40

4.3 Discussion: ... 46

4.4 Limitations and alternative explanations: ... 48

5. Conclusion ... 49

5.1 Avenues for future research:... 50

(3)

3 Figures:

Figure 1: Theoretical argument Security Guarantees Figure 2: Theoretical argument Persuasion

Tables:

Table 1: KFOR dependent variable Table 2: KFOR Security Guarantees Table 3: KFOR Persuasion

(4)

4 1. Introduction:

The North Atlantic Treaty Organisation (NATO) was founded in 1949 as a defensive alliance for states in North America and Europe (NATO, n.d.). The primary goal of NATO was to facilitate trans-Atlantic cooperation on matters of defence and security, and to create a framework for mutual protection amongst its member states. Since 1949 the organisation has been expanded to include not only states such as the US, UK, and France, but also former soviet-bloc countries such as Poland, Romania and Hungary (NATO, n.d.). However, while collective defence is still NATO’s primary objective, it´s mission has expanded to incorporate the promotion of democratic values and conflict prevention beyond the geographic boundaries of its member states. Furthermore, since the end of the cold war, NATO has been involved in military conflicts and has participated in peace operations in e.g., Bosnia, Kosovo, and Afghanistan (NATO, 2020a). Such crisis management initiatives have become an important component of the Strategic Concept that was adopted by the heads of state of the organisation’s member states in 2010 (NATO, 2010, pp. 19-22). According to the strategic concept, NATO has an incentive to alleviate and prevent crises and armed conflicts beyond the borders of its member states, through e.g., military intervention or peace support operations (PSOs) (NATO, 2010, p. 19).

(5)

5 operations to determine if NATO peace support operations mitigate or exacerbate conflict. Moreover, If NATO peace support operation are found to mitigate conflicts, it might strengthen NATO’s legitimacy and credibility as a “peacekeeper”. Therefore, I aim to contribute to the research field of peacekeeping and peace operations by filling this research gap by studying and examining the mechanisms that affect NATO peace support operations and UN peacekeeping operations abilities to mitigate war intensity. A small-n qualitative comparative case study using Mill’s method of difference, and elements of structured focused comparison and process-tracing will be used to analyse such operations.

Since I aim to compare the effectiveness of NATO PSOs and UN PKOs the research question will be:

How effective are NATO peace support operations in mitigating war intensity compared to UN peacekeeping operations?

2. Theory:

2.1 Previous research on peacekeeping:

This section of the paper will discuss and review previous research on peacekeeping and its effect on war intensity. Since this paper aims to study the effectiveness of NATO and UN peace operations1_{, the review will primarily focus on research that discusses peacekeeping} performance and peacekeeping outcomes. Furthermore, this paper will also review literature that compares the effectiveness of UN PKOs and NATO PSOs. However, since there seems to be relatively little disaggregated research on how different peacekeeping-conducting organisations perform compared to each other, especially how NATO performs compared to

(6)

6 e.g., UN or ECOWAS, this paper will also review literature that examines the conduct and performance of NATO military operations.

To begin with, Lisa Hultman, Jacob Kathman and Megan Shannon have discussed how PKOs can reduce the number of battlefield casualties during armed conflicts in their article, Beyond Keeping Peace: United Nations Effectiveness in the Midst of Fighting. They argue that peacekeeping operations mitigate battlefield casualties through two mechanisms (Hultman, Kathman and Shannon, 2014, p. 741).

The first of these mechanisms is that the presence of peacekeepers during an ongoing conflict can mitigate commitment problems by providing security guarantees for warring parties. Such security guarantees can make the conflict parties more willing to disarm and demobilize, since peacekeepers should guarantee their safety if any other party decides to continue fighting (Hultman, Kathman and Shannon, 2014, pp. 741-742). The second mechanism is that the presence of peacekeepers also increases the cost of continued fighting by limiting opportunities for battlefield engagement through separation of warring parties, the creation of buffer zones and monitoring of such buffer zones (Hultman, Kathman and Shannon, 2014, p. 742). If the warring parties have fewer opportunities to engage in armed conflict, their ability to inflict casualties on each other should be reduced (Hultman, Kathman and Shannon, 2014, p. 742). Furthermore, the authors also find that large deployments of peacekeepers are more likely to mitigate commitment problems and increase the cost of continued armed conflict than small deployments of peacekeepers (Hultman, Kathman and Shannon, 2014, pp 745-748).

(7)

7 However, Walter argues that third party guarantees, and peacekeeping missions might be able to alleviate these problems by enforcing peace agreements and ensuring that none of the conflict parties are able exploit their opponents’ weaknesses (Walter, 1997, pp. 340-341). By doing so, third party actors can help the warring parties overcome the commitment problems associated with civil wars. She claims that third party peacekeepers can achieve this if they have an interest in ensuring a peaceful resolution to the conflict and are committed to upholding eventual civil war settlements. Furthermore, peacekeepers are more likely to be successful if they are willing to use military force to discourage rule-breaking and opportunistic power-grabs. Lastly, Walter argues peacekeepers should station large forces within the conflict area and/or garrison strategically important locations to deter aggression and reduce the risk of conflict resumption (Walter, 1997, pp. 340-341).

However, some research on peacekeeping finds that coercive measures are not the only mechanisms that affect peacekeepers abilities to mitigate conflict. In her book, Power in Peacekeeping, Lisa Howard argues that persuasion and inducement also play an important role in effective conflict mitigation. In addition to using coercive measures such as deterrence, defence, surveillance, compellence and arrests, Howard argues that peacekeepers can also use persuasive measures such as mediation, shaming, public outreach and public information campaigns, symbolic displays, and educational programs to mitigate armed conflict (Howard, 2019, pp. 36-42, 147). Furthermore, she also argues that peacekeepers can use inducement to mitigate armed conflict through quick impact projects, humanitarian aid, development aid, trust funds, market restrictions, economic sanctions, restoration of institutions and employment opportunities (Howard, 2019, pp. 83-91).

(8)

8 been successful and that peacekeeping generally reduces the risk of civil war resumption. Furthermore, they argue that the presence of peacekeeping operations after civil wars has a large effect on the duration of peace (Fortna and Howard, 2008, pp. 289-290). Fortna and Howard also claim that there is evidence that the presence of peacekeepers during civil wars generally increases the likelihood of the conflicts ending peacefully (Fortna and Howard, 2008, pp. 290-291).

However, the authors note that most of the literature on peacekeeping has focused on UN PKOs, and that there is comparatively little research on peacekeeping efforts conducted by non-UN actors. They argue that there is an avenue for more disaggregated research about peacekeeping missions conducted by actors such as NATO, The Economic Community of West African States (ECOWAS), and other regional and sub-regional actors (Fortna and Howard, 2008, p. 291).

While research on the effect of non-UN peacekeeping operations might indeed be limited, there have been some authors who have written about why non-UN PKOs might not be as successful as UN PKOs. In the article Regional peacekeeping is not the way, Walter Dorn has discussed some of the problems that can hamper PKOs conducted by regional organisations or other groups such as NATO (Dorn, 1998, p. 1). Dorn argues that there are several reasons for why such PKOs might face problems when trying to mitigate casualties and keep peace. For example, regional actors often have ulterior motives and interests that might hamper their ability to be impartial (Dorn, 1998, p. 1). Furthermore, regional peacekeeping efforts might suffer from low legitimacy if they are dominated by ambitious regional actors (Dorn, 1998, pp. 1, 3). Lastly, regional PKOs might also lack the authority and the resources of UN PKOs, which can reduce their ability to fulfill their mandates (Dorn, 1998, pp. 1, 3).

(9)

9 peacekeepers are less effective at preventing rebel one-sided violence than UN peacekeepers (Bara and Hultman, 2020, pp. 17-20). They argue that the capabilities of different peacekeepers might affect their ability to deter violence which aligns with Dorn’s argument regarding the importance of peacekeeper’s capacities (Bara and Hultman, 2020, p. 22; Dorn, 1998, pp. 1, 3). Furthermore, they argue that previous reports on regional non-UN peacekeeping missions suggest that such actors often lack the training, equipment, and resources of UN missions. However, they also argue that some non-UN actors such as NATO might have better capabilities than other non-UN- and regional actors (Bara and Hultman, 2020, p. 22).

While there seems to be some evidence that non-UN PKOs are less effective than UN PKOs, it is uncertain to what degree legitimacy affects the efficacy of such PKOs and peace operations (Sambanis and Shulhofer-Wohl, 2005, pp. 13-17). Nicholas Sambanis and Jonah Schulhofer-Wohl argue that it is hard to control for impartiality and determine how perceived legitimacy impacts the efficacy of non-UN peacekeeping operations (Sambanis and Shulhofer-Wohl, 2005, p. 18). Rather, they argue that non-UN PKOs often lack the combination of limited enforcement and civilian administration that is often present during UN operations. Such deficiencies can help explain why non-UN operations seem to be less effective than UN operations (Sambanis and Schulhofer-Whol, 2005, p. 16). They also add that differences in mandates of UN and non-UN missions might explain why UN PKOs seem to be more effective (Sambanis and Schulhofer-Whol, 2005, pp. 16-17).

(10)

10 to face challenges concerning implementation (Sendmeyer, 2010, p. 27). He adds that the NATO-led ISAF mission to Afghanistan suffered from a lack of continuity during command rotations and a lack of cooperation with non-ISAF forces present in the country (Sendmeyer, 2010, pp. 29-32). Even though the 2010 Strategic Concept has tried to address such issues, NATO still faces challenges regarding interoperability, cohesion, and operational flexibility (Sendmeyer, 2010, pp. 36-41, 42-44).

Based on previous research, it appears that peacekeeping operations can mitigate war intensity by separating combatants, increasing the cost of armed violence, providing third party security guarantees, compellence, persuasion and inducement (Walter, 1997; Hultman, Kathman and Shannon, 2014; Howard, 2019). Furthermore, the presence of peacekeepers in conflict area seems to increase the chance of peaceful conflict resolution, increase peace duration, and reduce the risk of civil war resumption (Fortna and Howard, 2008). However there seems to be little disaggregated research comparing the performance of different peacekeeping organisations, especially NATO (Howard, 2008). Even though some authors such as Dorn have theorised that non-UN PKOs should perform worse than UN PKOs, and some authors have found evidence that regional peacekeeping operations perform worse than UN PKOs, there is little research that supports his theory (Dorn, 1998). Furthermore, even though there is some research that examines the performance of NATO PSOs, that research does not compare NATO and UN PKOs, nor does it investigate if the conflict-mitigating causal mechanisms associated with UN PKOs are present during NATO PSOs (Sendmeyer, 2010).

2.2 Research Gaps:

(11)

11 seems to be a need for more disaggregated research on such peacekeeping efforts (Dorn, 1998, pp. 1, 3; Fortna and Howard, 2008, p. 291). Furthermore, more recent research also agrees that there is a lack of disaggregated research on the effectiveness of non-UN PKOs (Hegre, Hultman and Mokleiv Nygård, 2019, p. 231). Therefore, it should still be fruitful to conduct more in-depth studies of how PKOs conducted by actors such as NATO perform in comparison to UN PKOs. Moreover, NATO seems like an especially important actor to study, since the NATO has conducted PSOs such as the peacekeeping missions to Bosnia, Kosovo, Northern Macedonia, and Afghanistan since the end of the Cold War, and the NATO strategic concept emphasises the importance of crisis management outside its member states (NATO, 2010, p. 19; NATO, 2020a).

2.3 Theoretical Framework:

This section of my paper will describe the relevant concepts used in my thesis and my theoretical arguments regarding the causal relationship between NATO peace support operations and war intensity. The theoretical argument will discuss how UN PKOs and NATO PSOs are assumed to affect war intensity, and the theoretical argument will be followed by a hypothesis that will be examined later in this thesis. The previous research on peacekeeping indicates that there is a positive relationship between peacekeeping and reduced war intensity. Moreover, there are several reasons for why some peacekeeping missions seem to be more effective at reducing war intensity than others, and two important mechanisms seem to be conflict mitigation through security guarantees and persuasion.

2.3.1 UN and NATO security guarantees:

(12)

12 other should diminish. Additionally, when peacekeepers separate armed combatants and if such measures are successful, they increase the material and political cost of continued fighting, since the combatants should have to engage the peacekeepers first to attack their adversaries (Hultman, Kathman and Shannon, 2014, p. 742). Furthermore, peacekeepers can also have the ability to mitigate commitment problems, i.e., when parties are unable to cease fighting due to mutual distrust, by disarming combatants and by imposing sanctions on actors who violate ceasefires and peace agreements (Fearon, 1995, p. 381; Hultman, Kathman and Shannon, 2014, p. 742; Walter, 1997, pp. 340-341). However, such security guarantees could theoretically be weakened if peacekeepers appear to act in manner that is not impartial, by e.g., targeting specific parties or regions. Such behaviour could result in a loss of trust for peacekeepers among aggrieved parties and might cause them to intensify military operations to increase their security, thus creating security dilemmas (Glaser, 2016, pp. 23-24).

While both NATO and the UN should be able to create similar security guarantees during their PSOs and PKOs, I expect that UN PKOs will generally implement stronger security guarantees than NATO, since there are some key differences between UN PKOs and NATO PSOs. While both UN and NATO missions can use separation of combatants to reduce their abilities to engage each other and increase the cost of continued fighting, it is uncertain if NATO security guarantees would be as robust as UN security guarantees, due to questionable NATO impartiality. Even though NATO PSOs are supposed to be impartial in nature, the organisation argues that there is a need for operational flexibility, and it is willing to use force against actors that persist in using violence even if such actions appears to violate the principle of impartiality (NATO, 2001, p. 3-1).

(13)

13 creating security dilemmas, i.e., situations where parties targeted by NATO escalate the conflict to become more secure (Glaser, 2016, pp. 23-24). Since, the UN does not have the same history seemingly partial behaviour, I expect that UN PKOs are less likely to create security dilemmas than NATO PSOs.

Thus, the security guarantees provided by UN PKOs are expected to increase the effectiveness of UN PKOs since they should reduce combatants' abilities to engage each other, increase the cost of continued fighting, and mitigate commitment problems. On the other hand, NATO PSO are expected to be less effective since NATO operations are expected to be perceived as more biased than UN PSO and are therefore more likely to create security dilemmas. If these expectations are correct, NATO PSOs should be less effective at mitigating war intensity than UN PKOs.

Figure 1: Theoretical argument Security Guarantees

2.3.2 UN and NATO persuasion:

Peacekeepers' abilities to use persuasion to change the behaviour of political factions and warring parties could also affect war intensity. Lisa Howard argues that peacekeepers' abilities to mitigate conflict generally derive from their abilities to persuade rather than violent sanctions (Howard, 2019, pp. 32-33). According to Howard, there are five mechanisms of persuasion that can help peacekeepers mitigate armed conflicts. These mechanisms are civilian and military

NATO PSO -> less impartial -> higher risk of security dilemmas -> weaker security guarantess – less effective conflict mitigation

(14)

14 mediation, shaming of violent actors, public outreach and public information campaigns, symbolic displays, and educational programs (Howard, 2019, pp. 36-42). She adds that UN peacekeepers have successfully carried out peacekeeping operations such as the UNTAG mission to Namibia by using measures such as mediation, shaming, outreach, symbolic displays, and educational programs to mitigate conflict (Howard, 2019, pp. 36-42).

Furthermore, for these mechanisms to be successful, peacekeepers should try to respect and understand the people they are supposed to help, ensure that their messaging is clear and coherent, and ensure that their behaviour does not contradict their messaging (Howard, 2019, pp. 36-42). While the UN might not necessarily be more likely to respect or understand the people and societies they are supposed to protect during PKOs than NATO, or ensure that their meaning is clear and coherent, I expect that the UN will have greater abilities to ensure that their messaging aligns with the behaviour of their peacekeepers. The main reason for this is that one of the core principles of UN peacekeeping operations is impartiality (UN, 2008, pp. 33-34). While NATO also regards impartiality as an important principle of their PSOs, their history of primarily targeting certain actors during their operations indicates that NATO sometimes engages in behaviour that might be perceived as partial and will thus probably lower their credibility as an impartial actor. Since the UN does not have the same history of seemingly partial behaviour during peacekeeping missions, there should be smaller risk that UN peacekeepers act in a manner that contradicts their principles of impartiality. Assuming that UN PKOs are more able to ensure that their peacekeepers align their actions with their messaging than NATO PSOs, their abilities to use persuasion effectively should also be greater (Howard, 2019, pp. 44-45).

(15)

15 which should hamper NATO’s abilities to mitigate conflict intensity. Furthermore, NATO’s military activities during PSOs might not only lower their credibility as keepers of peace, but also their credibility as protectors of civilians, especially if their military activities result in collateral damage. Such actions might antagonise civilians belonging to the targeted parties and thus reduce civilians’ pressure on the targeted parties to engage in peaceful conflict resolution.

Figure 2: Theoretical argument Persuasion

2.4 Summary of theoretical argument and hypothesis:

Both security guarantees and persuasion seem to be important conflict mitigating mechanisms during peace operations. Based on previous research and available UN and NATO documents, these mechanisms should be stronger during UN PKOs than NATO PSOs. While both NATO and the UN should be able to create similar security guarantees during their operations, I expect that NATOs security guarantees should be weaker since NATO’s questionable impartiality might create security dilemmas, and the military operations that often precede or coincide with their PSOs might cause resentment and increase polarization. Furthermore, I also expect that NATO PSO will not be able to use persuasion as effectively as the UN due to NATO’s history of seemingly partial behaviour. Such behaviour could appear to contradict NATO PSO messaging about keeping the peace and protecting civilians, thus lowering NATO’s credibility, and it might also make warring parties that have been targeted by NATO military operations less willing to negotiate and engage in peaceful conflict resolution. Since questionable impartiality could hamper NATOs ability to implement robust security guarantees and their

NATO PSO -> incoherent messaging and behaviour -> less effective persuasion -> less effective conflict mitigation

(16)

16 ability to use persuasion effectively, I expect that NATO PSOs will mitigate war intensity to a lesser extent than UN PKOs. Therefore, my hypothesis is:

H1: NATO peace support operations will mitigate war intensity to a lesser extent than UN peacekeeping operations.

3. Research design:

In this section of the paper, I will discuss the research method I am going to use to study the effect of NATO PSOs on war intensity and how I am going to select my cases. Moreover, I will also discuss how I am going to operationalise the different variables and the validity and reliability of the variables. Furthermore, I will also explain how I am going to gather data for the study and discuss what sources I am going to use.

3.1 Research Design and Method:

(17)

17 conduct a small-n case study, the results might not be as generalizable as the results of a large-n study.

Moreover, the research design used in this paper will be Mill’s method of difference, i.e., a most-similar design. In a most similar research design, the selected cases should ideally have similar scores on as many plausible explanatory variables as possible other than the variables of interest, i.e., the independent variable and/or the dependent variable (Gerring, 2007, p 131, Brancati, 2016, pp. 200-201). The goal of selecting cases that are similar on variables other than the independent and the dependent variable is to rule out other possible explanations for variation in the dependent variable other than the independent variable. Since I have already presented a hypothesis in the theory section of this thesis, the cases should be selected based on their scores on the independent variable (NATO PSO/UN PKO) rather than the dependent variable (war intensity) (Gerring, 2007, p. 131).

However, it can be hard to establish causality in a most-similar research design and since my independent variable and dependent variables are focused on the presence of NATO/UN peace operations and war intensity respectively, they do not address the proposed causal mechanisms, i.e., security guarantees and persuasion (Brancati, 2016, pp. 201-202). Therefore, I will also incorporate aspects of process-tracing in my case study to examine if the proposed causal mechanisms, are present and working as expected in the selected cases. According to Beach and Pedersen, the aim of process-tracing is to trace causal mechanisms and has generally been used in single case studies (Beach and Pedersen, 2013, pp. 2-3). Since I aim to examine if the mechanisms are present and working as expected I am going to rely on theory testing process tracing rather than theory building- or explaining outcome process-tracing (Beach and Pedersen, 2013, pp. 11-12). The structure of these process-tracing elements will be discussed in more detail in section “3.3.3 Mechanisms” of the research design.

(18)

18 acquire comparable data from several cases by asking several standardised and general questions of each case (George, 1979. p. 213). These questions should focus only on the area of interest in the case study and should be applicable to all the cases in the case study (George, 1979. p. 213). In this thesis I will conduct a structured focused comparison by asking theory-driven questions about the dependent variable and the proposed causal mechanism.

To find support for the hypothesis i.e., NATO peace support operations will mitigate war intensity to a lesser extent than UN peacekeeping operations, there should be a higher war intensity in the selected NATO case than in the selected UN case. However, it can be challenging to establish if the independent variable is the only factor that affects the dependent variable in two-case most similar design, and it might be hard to establish if there is causal relationship between peacekeeper organisation and war intensity even if the results support the hypothesis (Brancati, 2016, pp. 201-202). Therefore, to confirm the hypothesis, the proposed causal mechanism should also be present and working as expected. However, if both cases have a similar level of war intensity or if the selected NATO case has lower war intensity than the selected UN case, and the proposed causal mechanism is not present or not working as expected, the hypothesis should be rejected.

3.2 Case Selection:

(19)

19 The universe of cases for NATO PKOs and UN PSOs are NATO and UN operations that are carried out under any of the chapters VI, VII or VIII of the UN charter, are authorised by the UN security council, and occur in conflict where NATO and the UN are not warring parties. Operations that occur in conflict where either of these actors is a warring party will not be considered for the case selection, since they would violate the principles of consent and principled impartiality (NATO, 2001, p. 3-1; UN 2008, pp. 13-14, 31-35). However, operations that occur after these actors have ceased to be a warring party operating will be considered for the case selection since they would not violate the principle of impartiality and consent even though they might appear to be biased. Furthermore, operations that operate outside chapters VI, VII or VIII of the UN Charter will not be considered since they would not operate within the international framework for peacekeeping and peace enforcement (UN 2008, pp. 13-14). Moreover, operations that are carried out without authorization from the UN security council will not be considered PSOs since they would effectively violate international law.

To ensure that the cases are comparable I will also try to select cases of peace operations that had similar numbers of military personnel operated in the same region and operated in conflicts where the death toll exceeded the UCDP threshold for war (1000 deaths per calendar year) before the selected missions were deployed (PCR, n.d.) and operated in conflicts where a peace treaty or a ceasefire was in place before the mission started.

(20)

20 contingent that exceeded 10 00 troops if the troop contingent of the selected NATO PKO exceeded 10 000 troops.

The conflict intensity is also an important confounder to account for since, different peacekeeping organisation might not necessarily prioritise equally intense conflicts. For example, the UN has been found to prioritise intense conflicts which could have an impact on their peacekeepers ability to mitigate violence (Hegre, Hultman, Nygård, 2019, p. 222). Thus, to ensure that the selected peace operations operate in conflicts of similar intensity, both of the selected peace operations should operate in conflicts where the total number of battle-related deaths per calendar year exceeded 1000 before the start of operations.

Lastly, it is also important to consider whether a peace treaty/ceasefire was in place at the start of operations since the presence or absence of such treaties might have an impact on the peacekeepers abilities to mitigate conflicts. Grieg and Diehl have found that the presence of peacekeeping operations in ongoing conflict seem to decrease the chance of negotiations succeeding (Grieg and Diehl, 2005, p. 636-639). However, they also argue that the presence of peacekeepers might enhance the durability of existing treaties by providing security guarantees, encouraging compliance with existing treaties and discouraging rule breaking (Grieg and Diehl, 2005, p. 643). Since the presence of such agreements might have an impact on peacekeeping efficacy, the results of the analysis might be skewed if one of the selected peace missions operate in a conflict without any cease fire agreement or peace treaty and the other occurs in a conflict with a cease fire or peace treaty. Thus, there should already be a cease fire agreement or peace treaty in place in both the selected cases to ensure that the presence or absence of such agreement will not skew the result of the analysis.

(21)

21 campaign against Yugoslavian forces but had ceased to be a warring party when the Yugoslavian government agreed to a cessation of hostilities in June 1999 (NATO, 2020b; USIP, 1999). At its height, KFOR had around 43 000-50 000 military troops and the PSO operated under chapter VII of the UN charter, and was therefore tasked with restoring and enforcing peace (UN, 2008, pp. 13-14; NATO, 2015; NATO, 2020b).

There are several reasons for why I have not selected any of the other NATO PSOs, such as the PSOs in Bosnia (IFOR/SFOR), Afghanistan (ISAF) or Northern Macedonia (Operation Essential Harvest, Amber Fox and Allied Harmony), instead of KFOR. For example, I have not selected IFOR/SFOR since the operation succeeded the UNPROFOR PKO in Bosnia, which could have impact on results. Furthermore, I have not selected ISAF, since the PSO has operated more like a counterinsurgency operation rather than a PSO, which arguably violates NATO principles regarding principled impartiality and had such large troop contingent that it is hard to find comparable UN PKOs (NATO, 2001, p. 3-1). Additionally, the NATO PSO in Northern Macedonia has operated in a conflict where the number of yearly battle-related deaths (72 at its height) is far below the UCDP threshold for war and is therefore not as suitable when examining the effect of NATO PSOs on war intensity (Nato, 2015; UCDP, 2019b).

(22)

22 According to Bara and Hultman, UNPROFOR had around 11 000 troops in Bosnia and Herzegovina during 1993 and the size of the troop contingent increased to more than 30 000 in 1995 (Bara and Hultman, 2020). Furthermore, Bosnia and Herzegovina were engaged in an armed conflict between Bosnian Serbian, Bosnian and Croatian forces that resulted in more than 7629 battle-related deaths in 1992, exceeding the UCDP threshold for war (UCDP, 2019c). Thus, the UNPROFOR operations in Bosnia and Herzegovina seems like a suitable case since it is comparable to KFOR, regarding the regions it was active and the type conflict it was sent to, and it had a comparatively large troop contingent for a UN PKO. While some of the newer multidimensional stabilisation missions might also seem suitable since they have more robust peace enforcement mandates than UNPROFOR, they tend to be much smaller than NATO PSO.

Since I will only study UNPROFOR operations in Bosnia between April 1993 and November 1995, I will only study the early periods of KFOR operations to limit the risk of temporal effect skewing the results. In the KFOR case the studied period will cover the start of operations in June 1999 to December 2005. December 2005 seems like suitable cut-off since negotiations regarding the final status of Kosovo started in February 2006, which indicates that the situation in Kosovo was perceived to be relatively stable (SCR, 2020a). While the unilateral declaration of independence by Kosovo in 2008 might also be a suitable cut-off point, it is possible that 7-year difference in length might skew the results to much, and therefore the 2005 cut-off point seems to be preferable.

3.3 Operationalization:

3.3.1 Independent variable:

(23)

23 relatively easy to replicate (only NATO PSOs and UN PKOs can be included in this variable), it should have high validity and reliability (Sundberg and Harbom, 2011, pp 97-100).

However, since the independent variable seems to be relatively simple, I am going to discuss the time period during which the selected peace operations where active in the analysis and I will also discuss if there were other international peace operations active in same conflicts as the selected cases.

3.3.2 Dependent variable:

Since the presence of NATO PSO or UN PKO is expected to cause variation in war intensity, the dependent variable will be war intensity. Furthermore, war intensity will be operationalized as not only battlefield deaths, but also violence against civilians, and the number of refugees and internally displaced persons to avoid oversimplification and to capture more dimensions of the dependent variable. All these components of the dependent variable will be measured on a ratio scale, i.e. the scores can be ranked, have equal units distance, and have an absolute zero (0 battlefield deaths = no battlefield deaths) (Sundberg and Harbom, 2011, pp 100-101). For all these components I will compare the results at the end of studied period with the results for the year before the studied period, i.e., 1992 for UNPROFOR and 1998 for KFOR.

(24)

24 (Sundberg and Harbom, 2011, pp 97-98). Furthermore, this definition should also have relatively high validity since other researchers should be able to replicate the results by using this definition, although the exact number of battlefield deaths might vary depending on the sources (Sundberg and Harbom, 2011, pp 99-100).

Additionally, I will use the UCDP definition for one-sided violence, i.e. “The deliberate use of armed force by the government of a state or by a formally organised group against civilians”, to measure violence against civilians (PCR, n.d.). To measure violence against civilians I will also use the sources such as reports and conflict databases, and these sources will be supplemented with reports from NGO: s such as Amnesty International and Human rights watch. Like the definition for battlefield deaths, the definition for violence against civilians should also have relatively high validity since only direct acts of violence against civilians falls under the definition, and the definition should also be quite reliable and easy to replicate since it does not have many components (Sundberg and Harbom, 2011, pp 97-100).

(25)

25 I will base the structured focused comparison of the dependent variable on these 3 definitions and I will ask the following questions of both cases:

1. Has the number of battle-related deaths per year increased or decreased since the operation started?

2. Has violence against civilians per year increased or decreased since the operation started?

3. Has the number of internally displaced persons (IDPs) and refugees increased or decreased since the operation started?

3.3.3 Mechanisms:

As previously mentioned, elements of theory testing process-tracing will be used to examine the mechanisms of security guarantees and persuasion. To determine if the mechanisms are present and working as expected I will investigate if events have occurred that indicate that the proposed causal mechanisms are present.

3.3.3.1 Security Guarantees:

Regarding the security guarantee mechanism, I will examine if the UN and NATO have acted in a seemingly partial manner towards one or more conflict parties prior to or during the PKO/PSO. The reason for this is that partial behaviour is expected to hamper peacekeepers abilities to uphold security guarantees by creating security dilemmas and resentment, which should reduce peacekeepers' abilities to mitigate war intensity.

(26)

26 one or more conflicts before the peace operations started. One indicator for partial behaviour would be military attacks on a conflict party by NATO or UN forces before the operations started. While military attacks on conflict parties might not necessarily indicate that the peacekeepers are biased, it would probably still result in lowered trust for peacekeepers amongst the targeted parties. When investigating this mechanisms, military attacks will be defined as violent acts conducted by NATO or UN military forces that either target military units affiliated with a conflict party, or territory controlled by a conflict party. Since peacekeepers could act in a seemingly biased manner during operations, I will also investigate if the peacekeepers have carried out military attacks against forces affiliated with one or more parties during operation, using the same definition for military attacks. If the peacekeepers have not carried out such military attacks against one or more conflict parties before or during the operations, this step of the proposed mechanism will not be present.

I will also investigate if perceived peacekeeper bias has caused resentment amongst parties targeted by peacekeepers. In this case resentment will be defined as verbal condemnation of violent acts carried out by peacekeepers, by parties targeted by peacekeepers. For such condemnations to be included they must be carried out by national, regional, or military leadership figures. While resentment might also manifest on a more localised level, e.g., through expressions of outrage and anger amongst individual military units or sections of the civilian population, such outburst is not necessarily an indicator of wider resentment amongst the targeted parties.

(27)

27 the causal mechanism should be strengthened. However, if the number of violent attacks has not intensified, the proposed causal mechanism will be rejected.

Since I aim to treat the causal steps like variables, I will formulate the following questions for both cases:

1. Have the peacekeeping organisations acted in a seemingly biased manner before or during operations?

2. Has perceived peacekeeper bias caused resentment? 3. Has this resentment created security dilemmas?

If the answer to these questions is “yes”, there should be an observable causal process that looks like this:

Biased behaviour before or during operation -> Resentment amongst targeted parties -> Security Dilemmas.

(28)

28 3.3.3.2 Persuasion:

Regarding the persuasion mechanism, I will examine if the UN and NATO have acted in a manner that contradicts their messaging regarding peacekeeping, peace restoration and protection of civilians. The reason for this is that I expect that behaviour which contradicts peace operations messaging will lower their credibility and hamper their abilities to use persuasion effectively and will hamper their effort to mediate, reach out to the public, implement education programs etc.

To begin with, I will try to investigate if the peacekeepers have claimed to be impartial, that they are protecting civilians and/or that they are trying to restore peace before or during their mission. Instances of the peacekeepers claiming to be impartial, claiming to protect civilians or claiming to be trying to restore peace in publicly available documents, mission mandates and press releases will be regarded as evidence for the presence of this causal step. However, if such claims are only present in e.g., classified documents, this causal step will be rejected since classified information would probably not be publicly available and should not have an impact on peacekeepers ability to use persuasion.

(29)

29 Lastly, I will also try investigating if the peacekeepers efforts to use persuasion have been ineffective. When investigating this causal step, I will focus on three indicators: protests against peacekeepers, condemnations of peacekeeper behaviour and mediation failures. Condemnations of peacekeepers will be defined as verbal condemnation of peacekeepers by national, regional, and military figures2_{, and protests against peacekeepers will be defined as} written or verbal statements from civilians and local civil society organisations that express displeasure with peacekeeper policies. Furthermore, mediation failures will be defined as negotiations between conflict parties that are mediated by the peacekeeping organisations, that break down and/or result in no agreement. The presence of two or more of these indicators will be regarded as evidence of ineffective persuasion and presence of this “step”.

Since I also aim to treat the causal steps for this mechanism like variables, I will formulate the following questions for both cases:

1. Have the peacekeepers claimed that they are impartial, that they are protecting civilians and/or that they are trying to restore peace?

2. Have the peacekeeping organisations acted in a seemingly biased manner before or during operations that contradicts their messaging?

3. Have the peacekeepers efforts to use persuasion been ineffective?

If the answer for these questions is “yes”, there should be an observable causal process that looks like this:

Messaging about impartiality, protection of civilians and restorations of peace -> Biased behaviour before or during operation that contradicts messaging -> Persuasion less effective

2_{In the case of Kosovo, I will include Yugoslavian and Serbian national leaders since Kosovo did not declare}

(30)

30 If any of these causal steps are missing, the causal mechanism will be rejected.

3.4 Source criticism and data collection:

As mentioned earlier when discussing the measurements for the dependent variable, I will primarily rely on data from UCDP, journal articles, UNHCR reports and reports from other organisations. While sources such as UCDP, UNHCR, reports from organisations such as the UN Security Council should be relatively reliable and accurate, there is a risk that other sources such as newspaper articles could be less reliable. For example, it is possible that newspaper reports will or omit important data due to time and resource constraints, or due to political bias. Therefore, when I have relied on news reports I have tries to corroborate these reports by either cross-checking with other newspapers, press releases or the available UN Security Council event chronologies (SCR, 2020) to see if these sources can confirm such reports.

Furthermore, while NATO has a dedicated website for the KFOR mission, there is a risk that data provided by NATO could be distorted. Therefore, I have primarily used NATO sources to gather information about KFOR’s tasks and mandates, and I have tried to avoid relying too much on NATO when gathering data on controversial subjects such as battlefield deaths or violence against civilians. While similar problems might apply for the UN and UNPROFOR, I have generally relied on data from journal articles or the oxford peacekeeping handbook when analysing the UNPROFOR case, since UNPROFOR’s archived website is quite outdated, and is not considered an “…official document of the United Nations” (UNPROFOR, 1996).

4. Results and analysis:

(31)

31 summarises the result and these tables will be followed by more detailed analysis of the cases. I will also discuss potential limitations of the study and alternative explanations for the results

4.1 KFOR June 1999 – December 2005:

The NATO KFOR PSO was started on June 12, 1999, after Yugoslavian forces had started to withdraw from Kosovo in accordance with terms of the Military-Technical agreement signed by the Federal Republic of Yugoslavia, and NATO (NATO, 2020b). The operation was authorised by UN Security Council Resolution 1244, operated under chapter VII of the UN Charter, and was tasked with facilitating safe return of displaced people, demilitarizing the Kosovo liberation army, training the Kosovo Security Force (KSF), upholding public order, and deterring armed conflict between Yugoslavian and Kosovo Albanian forces (KFOR, n.d.; NATO, 2020b). KFOR is still an ongoing operation, but since the start of operations the size of its troop contingent has gradually been reduced from around 43 000-50 000 in 1999 to 3 500 in 2020 (NATO, 2020b).

(32)

32

4.1.2 Dependent Variable:

War Intensity: 1998: 2005:

Battlefield deaths: 1404 0

Fatalities from violence against civilians:

143+3 ₀

Refugees and IDPs: 200 000 IDPs 200 000 Refugees

246 000 IDPs 0 Refugees

Table 1: KFOR dependent variable

In 1998, the year before KFOR started operating in Kosovo, the UCDP estimated that conflict resulted in around 1235 battle related deaths (UCDP, 2019a). These number would rise to around 1404 battle-related deaths in 1999. However, since the majority of these battle deaths occurred before KFOR entered Kosovo, these deaths provide little information about the performance of KFOR. Furthermore, after the federal republic of Yugoslavia signed the military-technical agreement in June 1999, the number of battle-related deaths seems to have dropped and KFOR seems to have been able to avoid armed clashes during its operations and only engaged in one notable battle with armed combatants in Prizren in June 1999, which allegedly resulted in the death of 2 Yugoslavian snipers (O’Kane and Watt, 1999: SCR, 2020). Additionally, since UCDP records no battle-related deaths between 2000-2005, the number of

(33)

33 battle-related deaths seems to have decreased since KFOR started operating in Kosovo (UCDP, 2019a). Moreover, sources such as ACLED record no battles during 2000-2005, and thus seems to corroborate the UCDP statistics.

2. Has violence against civilians per year increased or decreased since the operation started? Regarding violence against civilians, it is harder to find statistics since neither UCDP nor ACLED record any case of violence against civilians during the 1998-2005 period. UCDP adds that many civilians were killed during the Kosovo war and that it is hard to distinguish between collateral damage and deliberate killings of civilians (UCDP, 2019a). However, since UCDP argues that some of the cases of state-based violence during the war in 1998-1999 could be cases of one-sided violence it seems plausible that violence against civilians was higher before KFOR entered Kosovo, since no battle-related deaths were recorded during 2000-2005 at all (UCDP, 2019a). Furthermore, sources such as Human rights watch have documented at least 143 civilian fatalities due to Kosovo Albanian and Yugoslavian attacks4_{(HRW, 1999). Initially,} civilian casualties seem to have increased in 1999 due to the NATO bombing campaign, but after KFOR entered Kosovo in June 1999 civilian casualties seems to have decreased. Moreover, neither UCDP nor ACLED record any cases of one-sided violence, and Amnesty International only record one case of possible one-sided violence in 2000, when a Kosovo Albanian civilian was allegedly killed by KFOR troops (Amnesty, 2000). Thus, violence against civilians seems to have decreased since KFOR started operating.

(34)

34 Kosovo at the end of October 1998 (COE, 1999). This number would rise to around 600 000 IDPs and 850 000 refugees during March 1999 – June 1999 (Trako, 2018, p. 6). Most of these refugees would return to Kosovo after the start of KFOR operations, but ethnic violence directed against non-Albanian minorities in Kosovo would result in new waves of displacement during the last months of 1999 and March 2004 (Ambroso, 2006. p 7). It is also difficult to find any estimates estimate of the total number of refugees at the end of 2005, but UNCHR reports that there were around 224 000 IDPs from Kosovo in Serbia and Montenegro, and around 22 000 IDPs in Kosovo proper (UNHCR, 2006, p. 314). Thus, the total number of refugees and IDPs seems to have been smaller at the end of 2005 compared to number of IDPs and refugees during the first half of 1999, but the number of IDPs in 2005 seems to have been slightly higher than at the end of 1998.

4.1.3 Mechanisms:

Security Guarantees:

Causal step: 1. 2. 3. Accepted/Rejected

Present? Yes Yes No Rejected

Table 2: KFOR Security Guarantees

(35)

35 Yugoslavia, and attacked Yugoslavian units, their air-defence systems, and their military infrastructure (Lambeth, 2001; NATO, 2016). Thus, NATO did conduct attacks against one of the conflict parties before the KFOR operations started. However, Allied Force concluded 2 days before KFOR entered Kosovo, and thereafter there are no reports of fighting between NATO and Yugoslavian forces or Kosovo Albanian forces other than the incident in Prizren in July 1999. However, this firefight seems to have been initiated by Yugoslavian snipers and should not be considered a NATO attack on Yugoslavian forces (O’Kane and Watt, 1999). Thus, while NATO appears to have acted in a seemingly biased manner before the start of KFOR operations, KFOR itself does appear to have acted in a biased manner during operations.

2. Has perceived peacekeeper bias caused resentment?

(36)

36 3. Has this resentment created security dilemmas?

(37)

37 Persuasion:

Causal step: 1. 2. 3. Accepted/Rejected:

Present? Yes Yes No Rejected

Table 3: KFOR Persuasion

1. Have the peacekeepers claimed that they are impartial, that they are protecting civilians and/or that they are trying to restore peace?

(38)

38 creating a secure environment for the citizens of Kosovo indicates that KFOR was expected to protect civilians in the region.

2. Have the peacekeeping organisations acted in a manner that contradicts their messaging before or during operations?

Even though KFOR itself does not seem to have acted in a biased manner against any of the warring parties after the start of operations, NATO’s bombing campaign against Yugoslavian forces in Kosovo and Yugoslavia proper conducted between March 24 1999 – June 10 1999 implies as degree of bias against the Yugoslavian government (NATO, 2016). Thus, NATO’s messaging in Kosovo could have been regarded as contradictory despite KFOR’s relatively unbiased behaviour. Moreover, despite relatively low levels of civilian fatalities in Kosovo after June 1999, there were instances when KFOR were unable to prevent violence against civilians. For example, KFOR was unable to prevent the displacement of minorities caused by communal violence between the Albanian majority and perceived pro-Serbian minorities in late 1999 (Ambroso, 2006. p 7). Furthermore, according to HRW, KFOR was unable to prevent the ethnic riots in March 2004, that resulted in the deaths of at least 30 civilians, that primarily belonged to the Serbian minority (HRW, 2004). Thus, NATO seems to have acted in manner that contradicted their messaging about impartiality and protection of civilians (NATO, 1999b; NATO. 1999b; Nato, 2020b).

3. Have the peacekeepers efforts to use persuasion been ineffective?

(39)

39 that negotiations on the final status of Kosovo did not start until early 2006 (SCR, 2020a). However, other negotiations such as the talks with the Kosovo Liberation Army (KLA) in September 1999 regarding demobilisation of the KLA seems to have been relatively successful (SCR, 2020a). Moreover, between June 1999 and December 2005, most civilian protest seems to have been directed towards different political and ethnic groups in Kosovo, or UNMIK, rather than KFOR (ICG, 2004). Thus, since there seems to have been no major condemnations of KFOR by political leaders, nor any notable mediation failure or anti-KFOR protest during the studied period, the NATO persuasion efforts are regarded to have been effective.

4.2 UNPROFOR April 1993 – November 1995:

The UNPROFOR PKO in Bosnia and Herzegovina was initially an extension of the UNPROFOR operation that had been active in Croatia from February 1992 (Tardy, 2014a, p. 2; Tardy, 2014b). The UNPROFOR PSO was authorised by the UN security council and was supposed to operate under chapter VII of the UN charter (Tardy, 2014a, p. 2). However, while UNPROFOR started operating in Bosnia in July 1992, the PKO was only tasked with protecting humanitarian convoys and securing the Sarajevo airport and did not participate in other peacekeeping task such as protecting civilians and separating combatants until its mandate was expanded in April 1993 when it was also tasked with protecting certain several so called “safe areas” (Tardy, 2014a., p. 2). This mandate expanded in May and June 1993 and continued to operate under these mandates until the end of operations in November 1995. While this PKO was smaller than KFOR, it had around 11 000 troops in 1993 and the size of UNPRFOR’s military contingent reached its height in august 1995, when the mission had around 30 000 troops stationed in Bosnia (Bara and Hultman, 2020).

(40)

40 and Operation Deliberate Force. Most of these operations other than Deliberate Force were tasked with enforcing weapons embargos and no-fly zones (NATO, 2020a). Furthermore, UNHCR also a conducted humanitarian assistance operation during the Bosnian war (Cutts, 1999, p. 1-2).

4.2.2 Dependent Variable:

War Intensity: 1992: 1995:

Battlefield deaths: 3 972 1 345

Fatalities from violence against civilians:

3 657 8 410

Refugees and IDPs: 810 000 IDPs 1 816 840 Refugees

1 300 000 IDPs 1 200 000 Refugees

Table 4: UNPROFOR Dependent Variable

(41)

41 Bosnia there had only been a 66% decrease in state-based fatalities compared to the almost 100% decrease in Kosovo (UCDP, 2019a; UCDP, 2019c).

2. Has violence against civilians per year increased or decreased since the operation started? In 1992 before the expansion of the UNPROFOR mandate, UCDP recorded around 3 657 fatalities due to on-sided violence, i.e., violence against civilians (UCDP, 2019b). Violence against civilians started to decrease after 1993, with UCDP recording around 517 fatalities from one-sided violence in 1993 and 309 fatalities in 1994. However, in 1995 the number of fatalities rose significantly and UCDP estimates that around one-sided violence resulted in around 8 410 fatalities (UCDP, 2019b). Most of these fatalities seems to have occurred during the capture of Srebrenica by Bosnian Serb forces in July 1995. Srebrenica was of the so-called safe areas that UNPROFOR was supposed to protect, and Human Rights Watch (HRW) estimates that around 7 000 – 8 000 civilians were killed by Bosnian Serb soldiers despite the presence of Dutch UN peacekeepers (HRW, 1995; HRW, 2005). Thus, violence against civilians seems to have increased after the UNPROFOR mandate was expanded in 1993 and reached its height during the last year of UNPROFOR operations in 1995.

(42)

42 the Bosnian war in December 1995 was around 1,3 million IDPs and around 1,2 million refugees according to UNHCR (UNHCR, 2000, p. 219). Since the total number seems to have shrunk by around 100 000 people between 1992 and 1995, the number of IDPs and refugees decreased slightly after the expansion of the UNPRFOR mandate in 1993.

4.2.3 Mechanisms:

Security Guarantees:

Present? No Yes No Rejected

Table 5: UNPROFOR Security Guarantees

(43)

43 This unwillingness to use forces resulted in the UNPROFOR peacekeepers being unable to protect the civilian population during the Srebrenica Massacre, despite being shelled and disarmed by Bosnian Serb Forces (HRW, 1995). Moreover, UNPROFOR commanders appear to have been unwilling to request air support from the active NATO air contingents in Bosnia, and when NATO forced Bosnian Serb forces to temporarily lift the siege of Sarajevo in 1994 the UNPROFOR peacekeeper did not secure the 20 km exclusion zone around the city, thus allowing Bosnian Serb forces to resume the siege (Tardy, 2014a, p. 6). Therefore, UNPROFOR did not act in a seemingly biased manner before or during the operations since they did not actively attack any of the warring parties in the Bosnian war.

2. Has perceived peacekeeper bias caused resentment?

While the UN did not act in a seemingly biased manner prior to or during UNPROFOR operations in Bosnia, the activities of UNPROFOR appear to have generated local resentment. For example, the Bosnian President during the war, Alija Izetbegovic, criticised UNPROFOR’s perceived inability to respond and react to the Bosnian Serb siege of Sarajevo (AP, 2015). Furthermore, the Croatian leadership also appears to have been frustrated with the UNPROFOR operations in the Balkans since UNPRFOFOR was perceived to help Serbian forces solidify their control over occupied territory in Croatia, and in early 1995 the Croatian President Franjo Tudjman resorted to threatening to expel UNPROFOR forces from Croatia (Tardy, 2014b, p. 6-7).

(44)

44 and the Bosnian Serbs and Yugoslavians being frustrated with composition of the UNPROFOR troop contingent.

3. Has this resentment created security dilemmas?

While all the major conflict parties expressed frustration with either UNPROFOR conduct or composition, it seems unlikely that the presence of UNPROFOR troops created security dilemmas. On the contrary, the intensity of violence between the warring parties seems to have dropped since UCDP only record 1345 fatalities due to state-based violence in 1995 compared to 3 972 fatalities in 1992 (UCDP, 2019c). This general decline in conflict intensity is corroborated by data from the Bosnian Book of the Dead (Tokača, 2012). While the Bosnian Book of the Dead does not record causes of death, it still estimates that around 9 634 soldiers affiliated with warring parties died in 1995, compared to around 23 304 soldiers in 1992 (Tokača, 2012, p. 113). Thus, both state-based violence general mortality rates amongst soldiers seems to have decreased during the conflict. While hundreds of UN peacekeepers were taken hostage during war with at least 200-400 peacekeepers being hostage in May 1995 alone, these number are not sufficient indicators of a general increase in conflict intensity during 1993-1995 when accounting for the number of overall fatalities (Br, 1995; Daalder, 1998). Thus, resentment against UNPROFOR in Bosnia does not appear to have resulted in security dilemmas.

Persuasion:

Present? Yes No Yes Rejected

Table 6: UNPROFOR Persuasion

(45)

45 As mentioned earlier, the expansion of the UNPRFOR mandate in April and May 1993 created so called safe areas in the Bosnian cities of Srebrenica, Bihac, Gorazde, Sarajevo, Tuzla, and Zepa (Tardy, 2014a, p. 2). Furthermore, UNSC resolution 836 explicitly tasked UNPROFOR with ensuring the safety of these safe areas and authorised them to use force to deter attacks against these areas (UNSCR, 1993c). Furthermore, in UNSC 819, 824 and 836, the UN Security Council authorised UNPROFOR in Bosnia to operate under Chapter VII of the UN Charter (UNSCR, 1993a; UNSCR, 1993b; UNSCR, 1993c). Chapter VII of the UN Charter explicitly states that any measures taken by the UNSC or UN peacekeepers to ensure the compliance of warring parties must be “…without prejudice to the rights, claims, or position of the parties concerned.”, i.e., such measures must be unbiased (UN, 2016). Thus, the UNPROFOR mandate from April 1993 tasked UNPROFOR with protecting civilians in the designated safe areas and the PKO was also expected to act in an impartial manner. However, even though UNPROFOR operated under chapter VII of the UN Charter and thus had the authority to restore peace in Bosnia, the UNPROFOR mandate did not order the PKO to enforce peace between the warring parties.

2. Have the peacekeeping organisations acted in a seemingly biased manner before or during operations that contradicts their messaging?

(46)

46 3. Have the peacekeepers efforts to use persuasion been ineffective?

As mentioned earlier, during the Bosnian war, UNPROFOR was criticised by the Bosnian President, the Croatian President, the Yugoslavian government, and the Bosnian Serb leadership (Williams, 1995; Gray, 1996, p. 250-251; AP, 2015). Furthermore, all the instances of verbal condemnations occurred after the expansion of the UNPROFOR mandate in 1993 and could therefore plausibly be explained by perceived contradictory UNPROFOR behaviour and messaging. Additionally, there were several instances of failed negotiations led by the UN after the expansion of the UNPROFOR mandate in 1993. For example, the Vance-Owen plan for the partition of Bosnia was rejected by the Bosnian Serb national assembly in May 1993, and the Owen-Stoltenberg plan was rejected in late 1993 (Greenberg and McGuinnes, 2000, p. 37). Moreover, there seems to have been instances of civilians protesting against UNPROFOR policies, and Bosnian Muslim civilians in particular, seem to have protested perceived UNPROFOR inaction (GAO, 1995, p. 24). Thus, since there were several anti-UNPROFOR civilians protest during 1993-1995, at least two major mediation failures and UNPROFOR was at various points criticised by political leaders from all major warring parties, the UN persuasion efforts are considered ineffective.

4.3 Discussion:

(47)

47 of war intensity seems to contradict the proposed hypothesis, i.e., that NATO peace support operations will mitigate war intensity to a lesser extent than UN peacekeeping operations and implies that NATO PSOs mitigate war intensity to a greater extent than UN PKOs. However, it is also necessary to the discuss the mechanisms of NATO and UN peace operations to examine whether there is support for the proposed theory.

The analysis of the two proposed causal mechanisms seem to contradict the proposed theory of this paper. In the theory section of this paper, I argued that NATO’s history of seemingly biased behaviour would probably damage their credibility as impartial actors, and thus hamper NATO’s abilities to mitigate war intensity during their peace support operations. However, even though NATO appears to have acted in a seemingly biased manner in the months leading up to the start of KFOR operations, this behaviour does not seem to have hampered NATO’s abilities to mitigate war intensity. On the contrary, NATO seems to have been able to work productively with both the Kosovo Albanians, the Kosovo Serbs, and the Yugoslavian/Serbian government, despite the NATO bombing campaign against Yugoslavian forces during March 1999 – June 1999. Furthermore, even though the proposed causal mechanisms for inefficient security guarantees is also rejected in the UNPROFOR case since the UN did not act in a seemingly biased manner, the UNPROFOR appears to have been distrusted and criticised by all the active conflict parties. Additionally, UN persuasion seems to have been less effective than NATO persuasion, despite Operation Allied Force in Kosovo.

(48)

48 4.4 Limitations and alternative explanations:

Since, neither the hypothesis nor the theory seems to be supported by the results of this case study, it is important to discuss potential alternative explanations. One plausible alternative explanation for the results is the military strength of the peacekeeping operations. As mentioned earlier, Hultman, Kathman and Shannon have found that the size of peacekeepers military contingents has an impact on war intensity and that peacekeeping operations with military contingents of more than 10 000 troops are expected to reduce battlefield deaths by at least 73% (Hultman, Kathman and Shannon, 2014, p. 742). Even at its height in 1995, UNPROFOR had a smaller troop contingent than KFOR, and this might have hampered their ability to effectively deter violence to the same degree as KFOR.

Another possible explanation for the results is the fact that KFOR entered Kosovo after the Yugoslavian government had agreed to withdraw its forces from Kosovo. While there seems to be have been fragile ceasefire in place when the UNPROFOR mandate was expanded, none of the warring parties in Bosnia had withdrawn or demobilized their forces. This could plausibly have hampered UNPROFOR’s abilities to fulfil their mandate, since the warring parties would have been more prepared and better equipped to resist UNPROFOR efforts to deter violence. However, without further research it is hard to determine if one or both of these factors could explain the results of this paper.

(49)

49 troop contingents in NATO PSOs would probably be to conduct a case study that compares different NATO PSOs, but this is unfortunately beyond the scope of this paper. Moreover, since this paper only studies two cases, another limitation would be generalizability. Thus, while the results of this case study imply that NATO PSOs are more effective at mitigating war intensity than UN PKOs, it is hard to determine if other studies examining other cases would produce the same results.

5. Conclusion

The purpose of this paper was to conduct a comparative case study analysing the effect of NATO PSOs and UN PKOs on war intensity, to help fill the existing gap concerning disaggregated research on non-UN peacekeeping operations. To do so, this paper focused on comparing NATO peace support operations and UN peacekeeping operations abilities to mitigate war intensity through a structured focused comparison and aspects of theory-testing process tracing. However, results of the analysis did not confirm the proposed theory that UN PKOs would mitigate war intensity to a greater degree than NATO PSOs, but rather seem to imply that some NATO PSO might reduce war intensity to a greater extent than UN PKOs. Furthermore, the results also imply that NATOs seemingly biased behaviour does not necessarily hamper their abilities to provide security guarantees, nor does it necessarily hinder NATO from using persuasion effectively.

(50)

50 5.1 Avenues for future research:

(51)

51 6. Reference list:

AP. 2015. Bosnia - Izetbegovic Attacks UNPROFOR's Role. [video online] Available at:

https://www.youtube.com/watch?v=7KAHs3A-Q5Q [Accessed 5 January 2021].

Ambroso Guido. 2006. The Balkans at a crossroads: Progress and challenges in finding durable solutions for refugees and displaced persons from the wars in the former Yugoslavia [pdf] Geneva: Policy Development and Evaluation Service, United Nations High

Commissioner for Refugees Available at: https://www.unhcr.org/4552f2182.pdf [Accessed 2 January 2021].

Amnesty. 2000. Mitrovicarapport. [online] London: Amnesty International. Available at:

https://amnesty.no/mitrovicarapport [Accessed 2 January 2021].

Bara, Corinne, and Lisa Hultman. 2020. Just Different Hats? Comparing UN and Non-UN Peacekeeping, International Peacekeeping (27:3): 341-368.

Beach, Derek, and Rasmus Brun Pedersen. 2013. Process-Tracing Methods: Foundations and Guidelines. Ann Arbor: The University of Michigan Press.

Br, Joel. 1995. BOSNIAN SERBS SEIZE MORE U.N. TROOPS. Washington Post [online] 29 May. Available at: https://www.washingtonpost.com/archive/politics/1995/05/29/bosnian-serbs-seize-more-un-troops/991628ef-8469-436d-8759-470fe4ab11d4/ [Accessed 6 January 2021].

(52)

52 Burns, John F. 1993. Cease-Fire in Bosnia Holds for a 2d Day, And Sarajevo Basks. New York Times [online] 30 March. Available at:

https://www.nytimes.com/1993/03/30/world/cease-fire-in-bosnia-holds-for-a-2d-day-and-sarajevo-basks.html?auth=linked-google [Accessed 28 December 2020].

COE. 1999. Humanitarian situation of the Kosovo refugees and displaced persons. [online] Strasbourg: Parliamentary Assembly of the Council of Europe. Available at:

http://assembly.coe.int/nw/xml/XRef/X2H-Xref-ViewHTML.asp?FileID=8705&lang=en

[Accessed 2 January 2021].

Cutts, Mark. 1999. The humanitarian operation in Bosnia, 1992-95: dilemmas of negotiating humanitarian access. [pdf] Geneva: United Nations High Commissioner for Refugees, Policy Research Unit. Available at: https://www.unhcr.org/3ae6a0c58.pdf [Accessed 8 January 2021].

Daalder, Ivo H. 1998. Decision to Intervene: How the War in Bosnia Ended. Brookings [online] 1 December. Available at: https://www.brookings.edu/articles/decision-to-intervene-how-the-war-in-bosnia-ended/ [Accessed 6 January 2021].

Dorn, W. 1998. Regional peacekeeping is not the way. Peacekeeping & International Relations 27(4): 1-1, 3+.