Вы находитесь на странице: 1из 21

Z Friedens und Konflforsch

https://doi.org/10.1007/s42597-019-00009-w

AUFSATZ

Violence reduction or relocation?


Effects of United Nations troops presence on local levels of
violence

Laura Peitz · Gregor Reisch

© Arbeitsgemeinschaft für Friedens- und Konfliktforschung e.V. (AFK) und die Autoren 2019

Abstract The article contributes to the research on local effects of United Nations
peacekeeping operations (UNPKOs) on conflict dynamics. Focusing on spatial ef-
fects we distinguish between effects of peacekeeping presence on conflict intensity
in a given area (violence reduction) and effects on the levels of violence in sur-
rounding locations (violence relocation). We expect that the effects of peacekeeping
presence differ with force projection capabilities of troop deployments. Our analyses
are based on data covering the two most recent UNPKOs and violent events in the
Democratic Republic of the Congo from 2000 to 2014, allowing for a geographi-
cally and temporally disaggregated approach. The results show that PKOs reduce the
local level of violence where they are stationed, particularly when better equipped.
However, we find that PKO presence can also lead to shifts in the areas affected by
conflict.

Keywords UN peacekeeping · Conflict dynamics · Spatio-temporal


disaggregation · Spatial modeling

Electronic supplementary material The online version of this article (https://doi.org/10.1007/


s42597-019-00009-w) contains supplementary material, which is available to authorized users.
L. Peitz ()
Berlin Graduate School for Global and Transregional Studies, Hertie School of Governance, Berlin,
Germany
E-Mail: l.peitz@transnationalstudies.eu
G. Reisch
Collaborative Research Center (SFB) 700, Freie Universität Berlin, Berlin, Germany
E-Mail: gregor.reisch@fu-berlin.de

K
L. Peitz, G. Reisch

1 Introduction

Since the first United Nations (UN) peacekeeping operation (PKO)1 was launched in
1948, the world has seen more than 70 such missions (United Nations Peacekeeping
2018a), with considerable changes in their number and character since the 1990s
(Howard and Dayal 2018, p. 75–76). By the end of the year 2018, the UN deployed
over 100,000 peacekeeping personnel in 15 PKOs all around the globe (United
Nations Peacekeeping 2018a). The continuous deployment of UNPKOs is based on
the wide-spread expectation of them being “one of the most effective tools available
to the UN to assist host countries navigate the difficult path from conflict to peace”
(United Nations Peacekeeping 2018e).
The reduction or elimination of violence and the containment of armed conflict
are central goals shared by almost all UN missions (Diehl and Druckman 2010,
p. 29–32). Furthermore, the UN grants special attention to the protection of civilians,
with more than 95% of peacekeepers today having a corresponding mandate (United
Nations Peacekeeping 2018d). Despite these being the most fundamental goals of
UNPKOs, there still exists only limited knowledge about whether UNPKOs actually
achieve the reduction and containment of fighting as well as violence against civilians
on the ground.
Academics today largely agree on UNPKOs’ ability to mitigate intrastate conflict
and to foster peace (Fortna 2008, 2004; Gilligan and Sergenti 2008; Hultman et al.
2016, 2013, 2014; Kathman and Wood 2016; Doyle and Sambanis 2006). Most
studies measure the presence of UNPKOs on a yearly and state-level basis. However,
as pointed out by Beardsley and Gleditsch (2015, p. 68), PKOs existing in space
plausibly affects locations of conflict. State-level analyses hence neglect potential
variation in the presence and effectiveness of peacekeeping in different areas of
countries (Autesserre 2014; Beardsley and Gleditsch 2015; Costalli 2014; Diehl
2014; Fjelde et al. 2019), especially in cases of geographically isolated conflicts
where conflict area and country-level characteristics differ considerably (Benson
2018, p. 3). Accordingly, the above-mentioned authors have rightly demanded further
geographic disaggregation in conflict and peacekeeping studies.
The goal of this article is to assess the spatial effects of UN peacekeeping pres-
ence on a sub-national level. We argue that in order to adequately analyze effects of
UN peacekeeping on armed conflict dynamics, it is necessary to grasp the within-
case variation of civil conflict as well as of PKO deployment. Therefore, we seek
to answer the question of whether UN peacekeeping presence affects local levels of
violence. We test hypotheses about micro-level dynamics of UN peacekeeping using
novel geographically disaggregated datasets on the presence of different types of
peacekeepers and violent events in the Democratic Republic of the Congo (DRC).
We differentiate effects of PKO presence on conflict intensity in the area of troop
deployment (violence reduction) and on conflict intensity in neighboring areas (vi-
olence relocation).

1 In this paper, we apply a broad definition of peacekeeping including diverse types of UN missions such

as peacebuilding and enforcement missions.

K
Violence reduction or relocation?

Results indicate that peacekeeping presence, particularly with increasing ability


to project force, reduces violence—with the clearest effects on fighting. We further
show the occurrence of relocation, by which heavily armed peacekeeping presence
in one location increases violence in neighboring areas. This article is part of the
relatively new literature on micro-level effects of UN peacekeeping. It re-assesses
previous hypotheses with new conflict data, adds new insights on different violence-
related effects of varying types of peacekeepers, and emphasizes potential relocation
effects. Thereby, it contributes to the scientific evaluation of whether peacekeeping
really is “one of the most effective tools available to the UN” (United Nations
Peacekeeping 2018e).

2 Previous research on effects of UN peacekeeping

Mission success is multifaceted and as such difficult to pinpoint. As a consequence,


we observe a great variety of dependent variables in studies on peacekeeping effec-
tiveness (see Fortna and Howard 2008; Di Salvatore and Ruggeri 2017 for literature
reviews).2 Several publications focus on the effect of UNPKOs on the duration of
post-conflict peace (Gilligan and Sergenti 2008; Hultman et al. 2016; Doyle and
Sambanis 2000; Fortna 2004), while others analyze their effect on conflict inten-
sity—distinguished in either levels of battlefield violence (Hultman et al. 2014)
or violence against civilians (Hultman 2010, 2014; Hultman et al. 2013; Kathman
and Wood 2016; Melander 2009)—or a combination thereof (Hegre et al. 2019).
Another strand of research analyzes more specific consequences of UNPKOs, like
sexual abuse (Aoi et al. 2007; Karim and Beardsley 2017), conflict contagion (Beard-
sley 2011), security (Brosig and Sempijja 2018), or democratization (Steinert and
Grimm 2014). Many of these cross-national studies on peacekeeping effects have
contributed to today’s consensus of peacekeeping being quite effective (Fortna and
Howard 2008).
Most of these studies on the effects of UNPKOs are based on rationalist assump-
tions expecting UNPKOs to alter the incentives of conflict parties by increasing the
cost of fighting through mechanisms such as credible commitment, deterrence, and
information (Di Salvatore and Ruggeri 2017, p. 8). Despite the inherently spatial
assumptions of such arguments (e.g. physical separation, patrolling and monitoring
activities), they have so far almost exclusively been tested on the national level. Only
recently have scholars started to zoom in and assess the presumed spatial dynamics
on the local level. One of the initial studies is Costalli (2014), in which he tests
the propositions that PKOs reduce violence by decreasing the possibility of surprise
attacks and by raising the costs of aggression for the case of the Bosnian civil war.
He finds that neither the presence nor active initiatives of peacekeepers in a given
municipality reduce the intensity of violence there. Another early study on spatial
dynamics of UN peacekeeping is Beardsley and Gleditsch (2015). They assess the
ability of peacekeepers to reduce the geographic scope of local conflict. The authors
posit that peacekeepers limit the movement of armed actors and thereby geograph-

2 In the following section, we focus on quantitative, large-N studies on UN peacekeeping effectiveness.

K
L. Peitz, G. Reisch

ically contain conflict. They test their propositions on data for UNPKOs in Africa
from 1990 to 2010 and find that missions with higher troop levels do have a conflict
containment effect.
In another geographically disaggregated study on eight African countries Ruggeri
et al. (2017) analyze how the local presence of peacekeepers and the size of deploy-
ments affect conflict onset and duration. They expect peacekeepers to reduce the
risk of armed conflict onset locally through monitoring and reporting activities on
the ground. Patrolling and deterrence are argued to be most effective with sufficient
numbers of peacekeepers. In addition, they expect peacekeepers—especially larger
troop deployments—to reduce the likelihood of conflict continuation through me-
diation and enforcement activities. Analyzing data on all UNPKOs in sub-Saharan
Africa from 1989 to 2006, they find evidence that UN peacekeepers are able to re-
duce the risk of local conflict duration, more effectively so with increasing numbers
of troops as expected. Yet, they do not clearly find that peacekeepers actually reduce
the risk of local conflict onset.
Most recently, Fjelde et al. (2019) argue that UNPKOs can be effective in the
protection of civilians by increasing the military and political costs for civilian
targeting through the use of force and monitoring. Local peacekeeping presence
is argued to be key, as peacekeepers can increase costs only in their immediate
surroundings. The effectiveness in reducing the risk of violence against civilians is
expected to increase with troop size. The study differentiates between government
and rebel violence, whereby only the latter is proposed to be reducible by UN
peacekeepers due to the requirement of host state consent. Fjelde et al. (2019)
analyze geo-referenced data for all UN missions in Africa from 2000 to 2011.
While they find no evidence for UN troops being able to reduce local one-sided
violence per se, they show that peacekeepers can—as expected—decrease the risk
of civilian targeting by rebels.
The present study speaks directly to this novel literature on the spatial dynamics
of UN peacekeeping. By testing on the sub-national level inherently spatial assump-
tions that have been made in previous, cross-national studies we address a very
relevant issue. We do so by using novel data for both the independent and depen-
dent variables that enable us to inspect peacekeepers’ effects on local violence. The
novelty is threefold: Despite being collected in a similar manner as the PKO data of
Fjelde et al. (2019) and Ruggeri et al. (2017), our data on UN peacekeeping pres-
ence in the DRC goes beyond these datasets as it allows—first—for the distinction
of different types of peacekeeping units according to their force projection capa-
bilities. Second, we deviate from the predominant usage of Uppsala Conflict Data
Program – Georeferenced Event Dataset (UCDP-GED) data and hence allow for the
re-assessment of previous results with different violence data. Furthermore, we use
the Event Data Project on Conflict and Security (EDACS) data to separately analyze
peacekeepers’ effects on violence resulting from fighting and violence against civil-
ians. Third, we explicitly address both the reduction of conflict by peacekeepers in
a given area as well as potential shifts of violence to other areas. By disaggregating
PKO units, types of violence, and peacekeeping effects, by making theoretically
grounded methodological decisions, by reflecting our results in light of previous
findings of quantitative as well as qualitative studies, and by using spatially and

K
Violence reduction or relocation?

temporally disaggregated data we hope to contribute to important current debates in


peace and conflict studies.

3 UN peacekeeping and local conflict dynamics

As peacekeepers are increasingly deployed to ongoing conflicts, “violence remains


a priority for peace missions” (Di Salvatore and Ruggeri 2017, p. 1; see also Hultman
et al. 2014, p. 737). We therefore focus on potential effects of UNPKO presence on
levels of violence, which we systematize along two dimensions: changes of conflict
intensity within a certain area (conflict reduction) and shifts in conflict intensity in
areas neighboring PKO presence (conflict relocation).

3.1 Peacekeeping and the reduction of violence

A central and visible task of peacekeepers is to reduce the level of violence on the
ground (Costalli 2014, p. 362; Hultman et al. 2014, p. 738). Fortna (2008) has iden-
tified four causal mechanisms of how peacekeepers affect peace: first, the change of
incentives for war; second, the prevention and control of accidents and involuntary
defection; third, the reduction of uncertainty and fear; and fourth, the prevention of
political abuse (Fortna 2008, p. 76–103). The first two of these have direct spatial
implications: Changing the incentives of conflict parties is about raising the cost
of war. Enemy factions carrying out military offensives or violence against civil-
ians in the presence of UN peacekeepers risk local PKO units becoming additional
combatants. Furthermore, PKOs make surprise attacks and civilian targeting more
difficult through patrolling and monitoring of troop movements as well as through
the demobilization of conflict parties. The prevention and control of accidents and
involuntary defection can be achieved by creating buffer zones or otherwise sepa-
rating armed actors geographically to impede direct contact between combatants. In
addition, PKOs can prevent low-level incidents and accidents from escalating via
on-the-spot low-level mediation and arbitration.
Applied to the sub-national level these assumptions imply that UN troops need
to be present in the same areas where armed actors operate. Otherwise they cannot
separate combatants physically, limit the opportunities for military advancements
and surprise attacks, or reduce the likelihood of accidental engagements. Armed
actors are aware of their own ability to assert control over territory or to project
power, a function which is declining with distance from their (home) base (Buhaug
2010). The same is true for PKOs and it is safe to assume that other (non-state)
armed actors acknowledge this in their decisions when and where to use force.
Thus, the presence and activities of UN troops close to areas where armed actors
operate are expected to change the incentives for fighting by raising the costs of
continued fighting and providing credible security guarantees. From this reasoning
it can be expected that peacekeepers through certain activities, such as separating and
disarming conflict parties, monitoring, and deterrence, can reduce the willingness of
armed actors to exercise violence. This leads to our first hypothesis:

K
L. Peitz, G. Reisch

H1: The presence of UN peacekeepers leads to a reduction of violence.

Like conventional military units, UN peacekeeping units differ with regard to their
equipment and capabilities. While some units consist of unarmed military observers
only, others have helicopters, airplanes, or vessels at their disposal. Several authors
have argued that the effect of UN peacekeepers on local levels of violence depends
on the force at their disposal and their credibility to use it (Costalli 2014, p. 363;
Hultman et al. 2014, p. 743; Fortna 2008, p. 87–89; Kreps and Wallace 2009). It is
therefore important to account for the composition of PKO deployments. Military
troops serve at the frontlines and are directly tasked with the separation, inter-
position, and disarmament of combatants, which are central to reducing violence.
Unarmed observers, on the contrary, are less fit for such tasks. Accordingly, authors
have argued that unarmed observers are not able to deter hostilities (Costalli 2014;
Hultman et al. 2014).
But there are also differences among military units with regard to their equipment.
Mobility, firepower, and the ability to protect one’s own force are crucial factors to
credibly deter other actors from the use of force. Well-trained and heavily armed
troop deployments are better equipped to forcibly address commitment problems.
They increase the costs of using violence for state and non-state actors alike and
credibly reassure factions that opportunistic violations of the peace will be punished
(Gilligan and Sergenti 2008, p. 92; Kreps and Wallace 2009, p. 12; Walter 1997,
p. 340). Furthermore, if deployed to areas where neither state nor non-state actors
are able to assert territorial control, a robust presence of peacekeeping troops can
better fill the power vacuum and intervene locally to stop armed actors from fighting
(Ruggeri et al. 2017, p. 168–169).
This means that in order to deter or stop combatants from fighting or conducting
violence against civilians in a certain area it makes a difference whether deployed
troops are just composed of infantry troops or if they have armored vehicles at their
disposal. In the latter case, distances can be traversed more quickly, and troops are
better protected. Regarding mobility, reconnaissance, and firepower the deployment
of helicopters or aircrafts should have an even greater impact on local UNPKO
capabilities. In challenging terrain, helicopters can sometimes be the only way to
provide logistical support. Furthermore, attack helicopters can provide close-air sup-
port for ground units, but they do not require forward observers, so effectiveness of
troop campaigns can be enhanced and sorties without troop deployment are possi-
ble, which can shorten response times significantly (Nadin et al. 2015, p. 97–99).
Therefore, our second hypothesis reads:

H2: The higher the level of force projection capabilities of UNPKO troops the
greater the reduction of violence.

3.2 Peacekeeping and the relocation of violence

UNPKOs are becoming increasingly offensive in nature and the DRC missions are
prime examples of this development (Cruz et al. 2017; Berdal and Ucko 2015).
Peacekeepers can hence be active conflict parties, potentially posing an existential

K
Violence reduction or relocation?

threat to other combatants.3 Mobility then can become a matter of survival to armed
actors (Beardsley and Gleditsch 2015, p. 72): Combatants stand a high risk of
being defeated or demobilized and disarmed by peacekeepers if they try to fight
them repeatedly in the same areas, since this allows peacekeepers to play out all
advantages of superior force. Furthermore, mobility can become an asset if actors
can maintain an informational advantage through varying the locations where they
engage in fighting or where they conduct surprise attacks. In addition, armed actors
can signal resolve by engaging the UN’s forces in varying locations and can thereby
potentially advertise their cause in order to secure a wider base of support. Being
an active party to an ongoing conflict might hence alter incentive structures of other
armed actors with the result of a relocation of violence.
The degree of violence relocation depends on the relative asymmetry of the
conflict parties, with weaker actors tending to be more mobile (Beardsley et al.
2015). PKO units with higher military strength pose a greater threat to other conflict
parties and alter the military asymmetry in their favor. We therefore expect the
relocation effect to increase with the military asymmetry between peacekeepers
and armed actors: The higher the force projection capabilities of peacekeepers, the
greater the importance of combatants’ mobility in order to evade defeat and maintain
informational advantages. This leads to the third hypothesis:

H3: UNPKO presence with higher levels of force projection capabilities leads to
greater relocation of violence.

4 Empirical analysis

To assess the posed hypotheses, we analyze geo-referenced data on the deployment


of UN peacekeepers and violent events in the eastern provinces of the DRC from
2000 to 2014. The (recent) history of armed conflict in the DRC is hard to capture
in its complexity. What follows, is a very brief, and thus necessarily incomplete,
overview for a better clarity of our cases (for more detailed and in-depth analyses
see Prunier 2009; Stearns 2011).

4.1 Brief history of armed conflict in the DRC

In 1998, a rebellion aimed at overthrowing the government of Laurent-Désiré Kabila


erupted in the Kivu provinces in Eastern Congo, starting the “Second Congo War”
(1998–2003) (de Vries 2015). The intervention of Rwanda and Uganda on the rebel
side and Angola, Chad, Namibia, Sudan, and Zimbabwe supporting the Kabila gov-
ernment added a regional dimension (Doss 2014a). Combined international efforts
led to the signature of the Lusaka Ceasefire Agreement and in November 1999, the
UN established a small observer mission of initially up to 500 military observers in

3 Here our argument differs from that of Beardsley and Gleditsch (2015), who expect PKOs to be con-
ducive to conflict containment, but claim this for PKOs taking up defensive positions.

K
L. Peitz, G. Reisch

the DRC: The United Nations Organization Mission in the Democratic Republic of
the Congo (MONUC).
While security in most of the country steadily improved since the early 2000s,
armed conflict concentrated in its eastern regions, with continuously high levels of
fighting between rebels and the Congolese army as well as frequent targeting of civil-
ians (Tull 2009). In reaction to the persisting instability, until mid-2010 MONUC
incrementally grew into a multidimensional PKO with a Chapter VII-mandate. In
July 2010, MONUC was renamed to United Nations Organization Stabilization Mis-
sion in the Democratic Republic of the Congo (MONUSCO) to reflect a new phase
and the will to engage more proactively in the protection of civilians (Karlsrud 2015,
p. 44).
Between 2010 and 2012, MONUSCO completed more than two dozen separate
operations against various armed groups (Doss 2014b). As a reaction to the capture
of the city of Goma by the M23 in 2013 the UN created a specialized military
unit with an unprecedented offensive mandate—the so-called Force Intervention
Brigade (FIB). The responsibility of this military formation was to strengthen the
mission and to make space for stabilization activities by neutralizing armed groups
via targeted offensive operations (United Nations Peacekeeping 2018b). The FIB
launched its first operations in July 2013 in active support of the Congolese army
(Tull 2017). By the end of 2018 MONUSCO had an authorized number of personnel
of over 18,000, including over 16,000 military personnel and 660 military observers
(United Nations Peacekeeping 2018c).

4.2 Units of analysis

The empirical analysis is restricted to the eastern provinces of North and South
Kivu, Ituri, Haut-Uele, Maniema, and Tanganyika, which together constitute an
area roughly the size of Sweden. The case of peacekeeping in the DRC with the
consecutive missions of MONUC and MONUSCO was selected, because taken
together they constitute the longest and most expensive peacekeeping mission in
UN history (Koko 2011, p. 29; Verweijen 2017, p. 370). They are hence politically
very relevant. Moreover, in relative terms to the size of the DRC or its population
MONUC/MONUSCO belong to the smallest missions deployed by the UN (Tull
2009, p. 223). This and the complexity of the DRC conflict make it a challenging
example to test the effects of peacekeeping presence.
The eastern DRC provides a rich case for the questions at hand allowing a differ-
entiated analysis over an extended period of time as well as across various conflict
and mission phases. Due to the vast size of even just the eastern provinces a ubiqui-
tous and continuous coverage by peacekeepers is not possible. As a consequence, we
can observe considerable variance in violence and peacekeeping presence over time
and space, especially since we included the less violent provinces of Maniema and
Tanganyika.4 In addition, rather than being able to focus on one or few conflict areas,
peacekeepers in the eastern DRC have over time been faced with “a multiplicity of
overlapping, often highly localized, conflicts” (Berdal 2009, p. 113) involving over

4 Yet, results were insensitive to the exclusion of these two provinces.

K
Violence reduction or relocation?

40 armed groups (Vogel 2013). The eastern DRC with its numerous complex, in-
tractable conflicts, its often challenging terrain, and thinly spread peacekeeping force
hence presents a tough test case to our hypotheses (Nadin et al. 2015, p. 116–118).
If UN peacekeepers are found to reduce local levels of violence here—as they are
expected by policy-makers and the public—there might be a chance that certain
results could travel to peacekeeping missions in other contexts where the conditions
for peacekeeping are less challenging.
The restricted study area of the eastern provinces has been chosen in light of
certain peculiarities of the DRC. First, the majority of violence is concentrated in
these areas, while the rest of the country has largely seen low levels of violence
between 2000 and 2014. Peacekeepers therefore are requested in several mandates
to deploy to “the Eastern part of the [DRC]” (United Nations Security Council 2004)
with the consequence of around 95% of military forces being sent to those areas
(Doss 2010). Analyzing only the eastern areas takes into account these particularities.
Moreover, it allows for fairly small units of analysis, hence increasing disaggregation
and spatial accuracy, without the cost of immense zero inflation and unfeasibly high
demands in computational power that similarly small units of analysis would entail
for the entire DRC.
To construct the units of analysis, a raster layer was laid over the study area,
dividing it into square grid cells with a resolution of about 30 km for cell sides,
resulting in a total number of 574 grid cells. Grid cells were chosen as the units of
analysis rather than administrative entities since rebel groups cannot be assumed to
take notice of sub-national boundaries, but rather cross them regularly. In addition,
sub-national entities in the DRC usually surpass 30 km grid cells in size so that an
aggregation to that level of analysis would result in a loss of detail. Furthermore,
the grid cell approach facilitates comparison of results as sub-national entities are
not constant over time and space (see also Schutte and Weidmann 2011, p. 147).
Temporal units are calendar quarters from 2000 to 2014. This procedure results in
a strongly balanced panel with 34,440 units of analysis for the entire study period
(574 grid cells × 60 quarters).

4.3 Data

Data for the dependent variable (DV)—local levels of violence—is based on the
Event Data on Armed Conflict and Security (EDACS), which compiles spatially and
temporarily disaggregated data on the use of violence in the context of civil wars
in Sub-Sahara Africa (Chojnacki et al. 2012b). EDACS defines violent events as
“violent incidents with at least one fatality resulting from the direct use of armed
force” (Chojnacki et al. 2012b, p. 385). The reason for the choice of this dataset
is that it provides a more consistent, comprehensive and detailed picture than com-
parable datasets. Events for the entire data collection are coded separately by two
coders based on a constant, pre-defined set of reliable international news sources to
ensure reliability, accuracy, and data consistency (Chojnacki et al. 2012b, p. 387).
In contrast to other comparable datasets that base data collection on diverse, fluc-
tuating sources—such as UCDP-GED (Sundberg and Melander 2013)—EDACS is
thus less prone to overrepresentation of single events caused by the availability of

K
L. Peitz, G. Reisch

occasional reports (Chojnacki and Reisch 2012, p. 41).5 What makes EDACS addi-
tionally suited for this analysis is the inclusion of events with unknown or unnamed
actor participation and of less well-observed conflict dyads. Despite “uncertainty
[being] a key characteristic to conflict” (Eck 2012, p. 135), UCDP-GED, on the
contrary, excludes events with unknown actors.
The local level of violence is measured using two different operationalizations:
a) the total number of violent events and b) the total number of fatalities, both
occurring in a given grid cell during a given quarter. In addition, data allows for
a further disaggregation of violence into fighting6 (number of fighting events and
fatalities resulting from fighting) and violence against civilians (events involving
one-sided violence against civilians and number of civilian fatalities resulting from
violence against civilians).7
Data for the independent variable (IV), UN peacekeeping presence, is based on
the deployment maps attached to the reports of the Secretary General to the Secu-
rity Council on MONUC and MONUSCO, usually produced on a quarterly basis.
The maps indicate locations of UN peacekeeping bases, including type, rough size,
and troop contributing country of the individual units deployed. These deployment
maps are an unbiased and exhaustive source of UN deployment data. To increase
inter-coder reliability of the data, maps were coded separately by two coders and
differences in coding were intensively checked.
According to our hypotheses, the IVs are operationalized to reflect PKO presence
in a given grid cell as well as possible. We use the information on PKO base size
included in the maps to construct “mobility buffers” around each site. This follows
the fact that the range of peacekeepers’ movement depends on the size of a PKO
base. Very small bases do not have the capacity to swarm out, as a certain number
of peacekeepers always have to rest or stay behind to protect the base. Troops in
larger bases can split up and thereby can regularly patrol larger areas around their
bases. With increasing troop size UNPKOs are thus better able to patrol and protect
critical infrastructure in the wider proximity of their bases and to establish distinct
areas of control.
Given the poor state of infrastructure in the DRC, especially the road networks in
the eastern part, and limited UN airlift capacities, military units have a very limited
reach beyond their bases (Kjeksrud and Ravndal 2011, p. 12). In 2012, after four
years of road works by international agencies, the road network in North Kivu al-
lowed an average travel speed of usually considerably less than 20 km/hour for a light
vehicle on large parts of the major roads. The trip from Goma to Kanyabayonga
(distance ca. 150 km) at that time took a hard day’s drive in a good light vehicle and
was usually impossible in an armored vehicle (Quick 2015, p. 90–92). To provide
another example: In 2011 a UN convoy drove from Bukavu, the capital of South

5 See Figure A-1 in the online appendix for a comparison of EDACS and UCDP-GED data illustrating

this point.
6 We define as fighting “armed interaction between two or more organized groups” (Chojnacki et al. 2012a,

p. 4), hence including fighting between non-state armed actors as well as fighting between these and gov-
ernment forces.
7 Please consult the online appendix for descriptive statistics on DVs and IVs.

K
Violence reduction or relocation?

Fig. 1 Operationalization of independent variables (left: Mobility buffers, right: Grid cell share covered,
second quarter of 2014). Source: Authors

Kivu, into Shabunda territory and managed to travel 22 km, 34 km and 44 km on


consecutive days (Quick 2015, p. 94–95).
In order to take geographical und infrastructural challenges as well as limited
capacities into account, buffers for UN bases with a size of less than 46 peacekeepers
(about one platoon) are set to a radius of 2 km; for bases with 46 to 255 peacekeepers
(one company) to 10 km; for those with 256 to 751 troops (two companies or one
battalion) to 20 km; and for bases larger than 752 troops to 30 km. We then measure
the share of a grid cell covered by peacekeepers. Where multiple peacekeeping
buffers overlap, we count the area of overlap only once. With this approach, we
construct a conservative measure of peacekeeping presence, posing a rather tough
test for our hypotheses.8 The left side of Fig. 1 shows peacekeeping bases in the
second quarter of 2014 with surrounding mobility buffers. Its right part shows the
gridded study area for the same point in time displaying the share of grid cells
covered by peacekeeping presence.
For the second and third hypotheses we disaggregate peacekeeper presence ac-
cording to the respective force projection capabilities of PKO bases. To this end,
we distinguish between bases without armed troops (i.e. military observers and re-
gional headquarters without military units), those with infantry only, bases with

8 To match deployment maps with the temporal units of calendar quarters in cases of irregular release

dates, we took the mean of variables over the months of a given quarter.

K
L. Peitz, G. Reisch

mechanized infantry, and bases with helicopters, airplanes, or marine units (in the
remainder called “aviated” troops). The operationalization of both DVs and IVs en-
ables a straightforward interpretation of results, as the units of the DVs are either
number of events or number of fatalities; and the units of IVs are the share of a grid
cell’s area covered by peacekeepers (the latter enters models with a one-quarter lag
to address potential problems of endogeneity or reverse causality).

4.4 Methods

For the statistical testing of our hypotheses, we rely on spatial panel data mod-
els. Such models are ideal, since they can accommodate spatial spillovers. Loosely
defined, these take place when “changes occurring in one region exert impacts on
other regions” (LeSage and Pace 2014b, p. 1536). As it is likely that local levels of
violence and peacekeeping presence in one unit are related to those of neighboring
areas, relying on conventional regression models would lead to an underestimation
of the real variance. As a result, the null hypothesis would potentially be unjustifi-
ably rejected (Ward and Gleditsch 2008, p. 8). Spatial econometric models, on the
contrary, are able to capture both temporal and spatial dependencies across units.
To test the proposed hypotheses, a number of Spatial Durbin Models (SDMs)
were estimated. SDMs allow spatial dependence not only for dependent, but also
separately for independent variables, which is convincing on theoretical grounds
for the case at hand: The level of violence in grid cell i depends not only on the
presence of peacekeepers in i and the level of violence in neighboring cells j, but
also on the presence of peacekeepers in j. Likelihood ratio tests as well as the Akaike
Information Criterion (AIC) all point towards SDM being the best model for the
data at hand. The SDMs take the following form:

yit D %Wyit C ˇXit C W Xi C ai C  t C uit

W is the spatial matrix for the spatially lagged dependent and independent vari-
ables, ρ is the endogenous interaction effect among the dependent variable, β the
effect of the independent variables and θ of the spatially lagged independent vari-
ables, ai is the individual fixed effect, γt is the time fixed effect and uit is a normally
distributed error term. The spatial weights matrix W, which identifies the neighbors
of each grid cell, is constructed as a row-normalized 1st-order queen contiguity ma-
trix, where all cells sharing at least one common point have a weight greater than
zero.9
Models were estimated with two-way fixed effects. The advantage of this ap-
proach is that it helps increasing the validity of our results despite the limited
availability of reliable spatially disaggregated data on conflict determinants in the

9 Despite some having warned against the practice of row-standardized contiguity matrices (Elhorst 2014,

p. 12–13; for a contrary view, see LeSage and Pace (2014a)), this approach is theoretically plausible in
our case: The effect of peacekeeping presence and violent events in neighboring grid cells can be assumed
to be stronger in border areas, where there are only few neighboring cells, since peacekeepers and rebel
groups have less options of movement here than in central areas.

K
Violence reduction or relocation?

eastern DRC. Instead of relying on such data, we opted for controlling for additional,
potentially relevant independent variables via fixed effects. Naturally, this pertains
only to space-specific, time-invariant variables that could affect the local level of
violence, such as topography, vegetation, infrastructure, proximity to international
borders and to the capital. Time-invariant third variables are therefore not controlled
for in our models, for there is no reliable data available on potential time-varying
determinants on the local levels of violence in the DRC. Datasets such as the PRIO
Grid (Tollefsen et al. 2012) either pose the problem of insufficient temporal dis-
aggregation of the data (such as population being measured only every five years),
the very high numbers of missing data (for example data on diamonds and gold),
or a lack in substantial change in variables over time (such as in the measure for
the number of excluded groups). For these reasons, the models do not control for
third variables. Instead, we control for potential omitted variable bias via fixed ef-
fects. In addition, as is shown by LeSage and Pace (2009, p. 67–68), spatial models,
especially SDMs, are affected less than Ordinary Least Squares (OLS) models by
omitted variable bias. As a consequence, even though omitted variables problems
are present, coefficients are still likely to be consistent.10
To check the robustness of our results, several additional models were estimated,
each time changing another aspect of the specification (among these non-dynamic
models, different grid sizes, UCDP-GED data, smaller study area). The results we
present below are remarkably insensitive to these changes and thus appear to be very
robust.11 An important aspect to regard is the non-random deployment of UNPKO
troops. It is very plausibly related to violence, as a consequence of which our re-
sults could be biased. If, for example, peacekeepers were systematically assigned to
areas of low violence, results might show a correlation of peacekeeping presence as
expected in our hypotheses. However, such results would then be driven by deploy-
ment patterns rather than actual peacekeeping effects. Yet, previous research (both
on the national and sub-national level) has repeatedly shown that peacekeepers are
actually sent to “hard” cases (among others: Fortna 2008; Gilligan and Stedman
2003; Melander 2009; Ruggeri et al. 2017; Fjelde et al. 2019). We are hence confi-
dent that—if anything—violence-reducing effects are underestimated in our models.
Nevertheless, we address the issue by keeping constant the “difficulty” of observa-
tions via the inclusion of the temporally lagged DV (main models), of the moving
average of the DV over the past four calendar quarters to account for potential fluc-
tuations, and by estimating peacekeeping effects separately for observations without
violence, “easy” cases (with fatality counts of 1 to 100) and “hard” cases (fatality
counts of over 100). Results of additional models indicate that the findings presented
below hold for all levels of preceding violence.

10 Statistical modeling is implemented via Quasi-Maximum Likelihood Estimation using Stata’s xsmle

command (Belotti et al. 2017).


11 See online appendix for results of robustness checks.

K
L. Peitz, G. Reisch

Table 1 Direct effects of


Violent events Fatalities
peacekeeping presence on local
levels of violence UNPKO presence –0.17 *** –6.87 ***
Rho 0.17 *** 0.11 ***
N 33,866 33,866
* p < 0.1; ** p < 0.05; *** p < 0.01

5 Results

Spatial regression models produce a number of coefficients, which are partly not
easily interpreted. Across all models estimated, ρ and θ are mostly statistically
significant at 5% or 1%. This indicates that spatial spillovers occur via the DVs and
IVs. As a consequence of the spatial spillover and feedback effects in the DV (ρ) and
thus the non-linear relationship between the DV and IV, it is not possible to interpret
regression coefficients like linear regression slope estimates. Rather, we need to
interpret effects using marginal effects reflecting partial derivatives (LeSage and
Pace 2014b, p. 1542). In accordance to the summary measures proposed by LeSage
and Pace (2009), we therefore rely for the interpretation of effects of peacekeeping
presence not simply on β, but rather on the so called direct effects (DEs) and
indirect effects (IEs, also called spatial spillover effects). Thereby, the DEs show
how changes in the IV in grid cell i impact the DV in i. IEs, on the contrary,
represent the impact on the DV in cell j from a change in an IV in grid cell i. Since
the impact of changes in an IV differs for each unit, DEs and IEs are reported as
scalar summary measures of the partial derivatives, i.e. average DE and IE (see also
Elhorst 2014, p. 21). We thus estimated average DEs and IEs in all our models
and focus on DEs in the interpretation of the results for H1 and H2, while IEs are
interpreted for H3.
Table 1 reports the DEs of SDMs with two-way fixed effects for the first hypoth-
esis. Turning to the second column, which reports results for H1 when measuring
the level of violence via the number of violent events per grid cell, the DE of the
presence of PKO troops is negative and statistically significant at the 1% level. This
indicates that higher peacekeeping presence in a given area on average reduces vi-
olence in that same area, just as expected by H1. The third column reports results
using the number of fatalities as the DV. The coefficient in the model on fatalities
largely exceeds the one in the model on the number of events. This is plausible,
since it is generally speaking “easier” for peacekeepers to prevent one fatality than
one entire violent event. Changing a grid cell without any peacekeeping presence
to being fully covered by peacekeepers (which is equal to a one-unit increase) on
average reduces the number of fights by 0.17 and the number of fighting fatalities
by more than six.
Hypothesis 2 expects units with higher force projection capabilities to have
a larger violence decreasing effect. Results are presented in Table 2 and are largely
as expected: the violence decreasing effect (measured as both events and fatalities)
differs with troop capabilities, in that only mechanized and aviated troops have

K
Violence reduction or relocation?

Table 2 Direct effects of


Violent events Fatalities
different UNPKO units on local
levels of violence Unarmed 0.31 14.51 **
Infantry –0.06 –1.12 *
Mech. Inf –0.10 *** –1.23 **
Aviated –0.39 *** –14.11 ***
Rho 0.17 *** 0.11 ***
N 33,866 33,866
* p < 0.1; ** p < 0.05; *** p < 0.01

clearly statistically significant (negative) DEs12 and coefficients grow in magnitude


with increasing capabilities. The effect of aviated units is about six times that of
basic infantry units for events and twelve times for fatalities. Interestingly, in the
model with fatalities as the DV unarmed military observers have large positive DEs,
pointing towards a violence increasing effect of such units.
UNPKOs are expected to affect local levels of violence via two mechanisms: by
stopping armed actors from violently engaging each other and by protecting civilians
from being harmed. Both forms of violence follow different tactical considerations
and therefore require varying peacekeeping approaches. In contrast to fighting, which
can be only used to contest, defeat, or weaken an armed adversary, violence against
civilians as a tactic employed by state and non-state armed actors alike can serve
a larger variety of purposes (cf. Hultman 2014 for an overview of the literature): E.g.
to indirectly enhance power relative to the adversary, to secure or violently extract
resources, to control the civilian population, or to cleanse an area of civilians in order
to establish territorial control. As a consequence of varying tactical considerations
the local effects of peacekeepers are likely to differ between fighting and violence
against civilians (Hultman et al. 2014, 2013; Kathman and Wood 2016).
We therefore re-estimated the above models separately for fighting and violence
against civilians. As can be read off Table 3, results are quite unambiguous for the
case of local levels of fighting, regardless of the operationalization of the DV. Aviated
troops seem to have a considerably larger decreasing effect on levels of fighting
than infantry (whereby it does not seem to make a difference whether infantry is
mechanized or not). Unarmed military observers, contrarily, appear to increase the
levels of fighting substantially. Just as in the models above, violence decreasing
effects are stronger in terms of fatalities than in terms of events. The picture is less
clear when turning to violence against civilians. Aviated troops seem to be able to
reduce the levels of violence against civilians in their areas of operation. Military
observers might be able to exert a similar effect (negative, statistically significant
effect only in the case of DV measured as events). In contrast, infantry does not
appear to have any effect on violence against civilians.
We now turn to relocation effects of different types of peacekeepers on neigh-
boring areas (H3), for which we inspect IEs. Average IEs can be interpreted “as the
impact of changing a particular element of an exogenous variable on the dependent
variable of all other units” (Elhorst 2014, p. 21), with spatial spillover effects be-

12 The DE for infantry is “only” statistically significant on the 10% level in the model on fatalities.

K
L. Peitz, G. Reisch

Table 3 Direct effects of different UNPKO units on local levels of fighting and violence against civilians
Fighting Violence against civilians
Violent events Fatalities Violent events Fatalities
Unarmed 1.21 *** 12.97 *** –1.19 *** –6.67
Infantry –0.08 *** –1.28 *** 0.01 0.02
Mech. Inf –0.06 *** –0.83 *** –0.02 * –0.34
Aviated –0.16 *** –7.08 *** –0.07 *** –1.25 ***
Rho 0.12 *** 0.08 *** 0.19 *** 0.02 *
N 33,866 33,866 33,866 33,866
* p < 0.1; ** p < 0.05; *** p < 0.01

Table 4 Indirect effects of


Violent events Fatalities
different UNPKO units on local
levels of violence Unarmed (spatial lag) 0.89 –0.17
Infantry (spatial lag) 0.23 *** 1.02
Mech. Inf. (spatial lag) –0.03 –0.41
Aviated (spatial lag) 0.44 *** 9.39 ***
Rho 0.17 *** 0.11 ***
N 33,866 33,866
* p < 0.1; ** p < 0.05; *** p < 0.01

Table 5 Indirect effects of different UNPKO units on local levels of fighting and violence against civilians
Fighting Violence against civilians
Violent events Fatalities Violent events Fatalities
Unarmed (spatial lag) 1.94 *** 12.84 –1.81 *** –24.43 **
Infantry (spatial lag) 0.12 *** 0.65 0.09 ** 0.10
Mech. Inf. (spatial lag) –0.02 0.76 0.01 –0.55
Aviated (spatial lag) 0.21 *** 5.07 *** 0.12 *** 1.31 **
Rho 0.12 *** 0.08 *** 0.19 *** 0.02 *
N 33,866 33,866 33,866 33,866
* p < 0.1; ** p < 0.05; *** p < 0.01

ing strongest on first-order neighbors and decaying towards higher-order neighbors


(LeSage and Pace 2009, p. 40). Results in Table 4 report IEs on the level of violence
in general. Interestingly, aviated troops appear to have a strong relocation bias (with
positive, statistically significant IEs in both operationalizations). The findings for
infantry are somewhat ambiguous as relocation effects are statistically significant
only for the event operationalization. Military observers and mechanized infantry
do not show any statistically significant relocation effects.
Differentiating violence into fighting and violence against civilians shows largely
similar results. Aviated troops seem to relocate fighting as well as violence against
civilians to neighboring areas, which is potentially also true for infantry. However,
results for military observers differ between the two types of violence. Results
indicate that military observers might increase fighting in neighboring areas (IEs

K
Violence reduction or relocation?

for event model statistically significant). In contrast, they seem to decrease violence
against civilians in neighboring areas (Table 5).
Our results show that peacekeeping presence can indeed lead to a reduction
in local levels of violence (both in terms of number of events and fatalities), as
expected by H1. With regard to H2, this violence reducing effect seems to intensify
with increasing force projection capabilities. The results for relocation effects for
general levels of violence (H3) indicate effects clearly only for aviated troops. Yet,
our data allow for the examination of the hypotheses separately for fighting and one-
sided violence against civilians. These additional analyses yield interesting results
that underscore the value of a differentiated treatment of the two concepts.
Our analysis paints a relatively clear picture regarding the reducing effects on
fighting: The presence of UN troops leads to a reduction in local levels of fighting
events as well as in the number of fatalities from fighting. Higher force projection
capabilities in the form of aviated troops seem to have a greater decreasing effect
than (mechanized) infantry. These findings are in line with results of Hultman et al.
(2014) regarding the UN’s general ability to reduce state-level battlefield violence
in ongoing civil wars in Africa. Somewhat worrisome, we find a reversed effect for
unarmed peacekeepers. Their presence leads to a considerable increase in the local
number of fighting events and fatalities from fighting. This effect is similar to the
one reported by Hultman et al. (2014, p. 747), who find an increasing number of
UN observers to correlate with more battlefield violence.
The results for violence against civilians, on the other side, are somewhat more
ambiguous. It seems to be effectively addressed only with aviated troops, while
(mechanized) infantry has no reduction effect. Contrasting the results for fighting,
violence against civilians could potentially be reduced with military observers in the
area (with the event operationalization being negative and statistically significant at
the 1% level). The findings are in line with those of Fjelde et al. (2019), who find
no discernable effect on one-sided violence against civilians in general, but only
for rebel violence. The ambiguity in their and our findings indicates that address-
ing violence against civilians could be less straightforward for peacekeepers than
countering fighting.
One could conclude that aviated peacekeepers are most successful in halting
violence. However, our research shows that particularly the presence of such forces
is associated with relocation effects of fighting and violence against civilians alike.
Results thus suggest that the shown ability of aviated units to deter armed actors
seems to relatively quickly decrease with the distance to UN bases with armed actors
being able to shift their areas of operations. Interestingly, unarmed peacekeepers
seem to reduce the level of violence against civilians not only in areas where they
are based, but also in surrounding areas. Yet, results hint at a relocation of fighting
by unarmed military observers. This contrasts the findings of no relocation effect by
Fjelde et al. (2019). Their study, however, does not differentiate between types of
peacekeepers. Our result that relocation effects are not homogenous across types of
peacekeepers might hence explain missing relocation effects in Fjelde et al. (2019).

K
L. Peitz, G. Reisch

6 Conclusion

In this paper we assessed the local effects of UNPKO presence on levels of violence.
In sum, our results indicate that UNPKO troops are able to reduce violence in areas
where they are active. We differentiate between types of UN troops as well as
two types of violence. This approach proves useful in that we can observe varying
violence reduction and relocation effects for different types of peacekeepers. While
aviated troops have the strongest violence reducing effects, in contrast to other troops
they also provoke violence relocation. In addition, effects differ between fighting
and violence against civilians. Military observers appear to have a rather particular
role in that they quite substantially increase and might relocate fighting, but decrease
the level of violence against civilians where they are based and potentially even in
surrounding areas.
We are aware that since we concentrate on a constellation of just two consecutive
UNPKOs in the context of only one country we need to be cautious when interpreting
results. Hence, we do not try any generalizations based on our findings, but rather
see them as an additional step in the research on local effects of UNPKOs. Future
research should be conducted for additional—potentially less offensive—UNPKOs
or in comparative research settings focusing for example on variations of the political
contexts of different UNPKOs. In addition, future studies should include additional
potential explanatory factors, such as the local deployments of host country troops or
presence of internally displaced people, and might account for the dependent variable
being count data, for example by relying on spatial models for count data (Poisson
or negative binomial models), which are, however, up to this point not yet frequently
applied (Glaser 2017). Moreover, future research could pay particular attention to
the number and type of (non-state) armed actors. As Benson (2018, p. 14) shows,
the effect of targeted interventions on the level of civilian victimization depends
on characteristics and capacities of armed groups. Hence, additional factors like
the number and characteristics of non-state armed actors should be considered in
future research, in order to account for their strategic and tactical behavior and the
consequences of their respective choices for the civilian population.
Autesserre (2014, p. 496; 2017, p. 117) remarks that the obstacles to peacekeeping
are such that the most puzzling question is why international efforts sometimes
succeed, rather than why they fail. We are confident that applying the approach taken
in this paper to other contexts or different research questions would be a fruitful
endeavor in this regard. It could be used to identify cases, contexts, and constellations
of locally successful UNPKO deployments and thus help to gain much needed
insights how to enhance peacekeeping successes (Autesserre 2017, p. 116). All in
all, our research adds layers and complements insights to existing research on the
effects of UNPKOs on micro-level conflict dynamics and underlines the added-value
of a spatially and temporally disaggregated approach.
Our analysis indicates that UN peacekeepers can contribute to local conflict re-
duction, especially when adequately equipped. But results also emphasize the need
for peacekeeping approaches that are sensitive to local requirements, conflict dynam-
ics, and actors involved. Troop deployment needs to be aware of potential violence
relocating effects, particularly in the case of aviated troops. In addition, the UN

K
Violence reduction or relocation?

should be sensitive to the type of violence occurring or likely to occur in a given


area. Based on conflict sensitive context analyses, unarmed peacekeepers rather than
armed troops could be deployed in order to address violence against civilians. We
do indeed find evidence for PKOs potentially being the “effective tool” they aspire
to be.
Acknowledgements We would like to thank the German Research Foundation for its financial support, our
colleagues Lisa Pech and Paul Zeidler for their valuable support in geo-coding the data, and our reviewers
and editors for their helpful comments and suggestions.

References

Aoi, Chiyuki, Cedric de Coning, and Ramesh Thakur (eds.). 2007. Unintended consequences of peace-
keeping operations. Tokyo, New York, Paris: United Nations University Press. ISBN 978-92-808-
1142-1
Autesserre, Séverine. 2014. Going micro. Emerging and future peacekeeping research. International
Peacekeeping 21(4):492–500.
Autesserre, Séverine. 2017. International peacebuilding and local success. Assumptions and effectiveness.
International Studies Review 19(1):114–132.
Beardsley, Kyle. 2011. Peacekeeping and the contagion of armed conflict. The Journal of Politics
73(4):1051–1064.
Beardsley, Kyle, and Kristian Skrede Gleditsch. 2015. Peacekeeping as conflict containment. International
Studies Review 17(1):67–89.
Beardsley, Kyle, Kristian Skrede Gleditsch, and Nigel Lo. 2015. Roving bandits? The geographical evolu-
tion of African armed conflicts. International Studies Quarterly 59(3):503–516.
Belotti, Federico, Gordon Hughes, and Andrea Piano Mortari. 2017. Spatial panel-data models using Stata.
The Stata Journal: Promoting communications on statistics and Stata 17(1):139–180.
Benson, Jay. 2018. The geography of violence against civilians. Implications for peace enforcement. OEF
Research https://doi.org/10.18289/OEF.2018.024
Berdal, Mats R. 2009. Building peace after war. London: Routledge.
Berdal, Mats, and David H. Ucko. 2015. The use of force in UN peacekeeping operations. The RUSI
Journal 160(1):6–12.
Brosig, Malte, and Norman Sempijja. 2018. Does peacekeeping reduce violence? Assessing comprehen-
sive security of contemporary peace operations in Africa. Stability: International Journal of Security
and Development 7(1):133.
Buhaug, Halvard. 2010. Dude, where’s my conflict? LSG, relative strength, and the location of civil war.
Conflict Management and Peace Science 27(2):107–128.
Chojnacki, Sven, and Gregor Reisch. 2012. Perspectives on war: Generation eventdata. Was zeigen
Ereignisdaten, worin unterscheiden sie sich und was können wir lernen? Sicherheit und Frieden
30(1):36–43.
Chojnacki, Sven, Christian Ickler, Katharina Schoenes, Michael Spies, and Tim Wildemann. 2012a.
“EDACS codebook version 3.7”, The Event Data on Armed Conflict and Security (EDACS) Project.
Berlin: Freie Universität Berlin.
Chojnacki, Sven, Christian Ickler, Michael Spies, and John Wiesel. 2012b. Event data on armed con-
flict and security. New perspectives, old challenges, and some solutions. International Interactions
38(4):382–401.
Costalli, Stefano. 2014. Does peacekeeping work? A disaggregated analysis of deployment and violence
reduction in the Bosnian war. British Journal of Political Science 44(02):357–380.
Cruz, Carlos Alberto Santos, William R. Phillips, and Salvator Cusimano. 2017. Improving security
of United Nations peacekeepers. We need to change the way we are doing business. Avail-
able online: https://peacekeeping.un.org/sites/default/files/improving_security_of_united_nations_
peacekeepers_report.pdf.
Di Salvatore, Jessica, and Andrea Ruggeri. 2017. Effectiveness of peacekeeping operations. In The Oxford
Encyclopedia of Empirical International Relations Theories, ed. William R. Thompson. New York:
Oxford University Press.
Diehl, Paul F. 2014. Future directions for peacekeeping research. A series of commentaries. International
Peacekeeping 21(4):481–483.

K
L. Peitz, G. Reisch

Diehl, Paul F., and Daniel Druckman. 2010. Evaluating peace operations. Boulder: Lynne Rienner Pub-
lishers.
Doss, Alan. 2010. Keeping the peace, protecting the people. 7th Folke Bernadotte Memorial Lecture,
London, 20 May 2010.
Doss, Alan. 2014a. United Nations Organization Mission in the Democratic Republic of the Congo
(MONUC). In The Oxford Handbook of United Nations Peacekeeping Operations, ed. Joachim A.
Koops, Thierry Tardy, Norrie MacQueen, and Paul D. Williams. Oxford: Oxford University Press.
Doss, Alan. 2014b. United Nations Organization Stabilization Mission in the Democratic Republic of
the Congo (MONUSCO). In The Oxford Handbook of United Nations Peacekeeping Operations,
ed. Joachim A. Koops, Thierry Tardy, Norrie MacQueen, and Paul D. Williams. Oxford: Oxford
University Press.
Doyle, Michael W., and Nicholas Sambanis. 2000. International Peacebuilding. A theoretical and quanti-
tative analysis. American Political Science Review 94(04):779–801.
Doyle, Michael W., and Nicholas Sambanis. 2006. Making war and building peace. United Nations peace
operations. Princeton: Princeton University Press.
Eck, Kristine. 2012. In data we trust? A comparison of UCDP GED and ACLED conflict events datasets.
Cooperation and Conflict 47(1):124–141.
Elhorst, Jean Paul. 2014. Spatial Econometrics. From cross-sectional data to spatial panels. Springer briefs
in regional science. Berlin: Springer.
Fjelde, Hanne, Lisa Hultman, and Desirée Nilsson. 2019. Protection through presence. UN peacekeeping
and the costs of targeting civilians. International Organization 73(01):103–131.
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.
Fortna, Virginia Page. 2008. Does peacekeeping work? Shaping belligerents’ choices after civil war.
Princeton: Princeton University Press.
Fortna, Virginia Page, and Lise Morjé Howard. 2008. Pitfalls and prospects in the peacekeeping literature.
Annual Review of Political Science 11(1):283–301.
Gilligan, Michael, and Stephen John Stedman. 2003. Where do the peacekeepers Ggo? International Stud-
ies Review 5(4):37–54.
Gilligan, Michael J., and Ernest J. Sergenti. 2008. Do UN interventions cause peace? Using matching to
improve causal inference. Quarterly Journal of Political Science 3(2):89–122.
Glaser, Stephanie. 2017. A review of spatial econometric models for count data. Stuttgart: University of
Hohenheim.
Hegre, Håvard, Lisa Hultman, and Håvard Mokleiv Nygård. 2019. Evaluating the conflict-reducing effect
of UN peacekeeping operations. The Journal of Politics 81(1):215–232.
Howard, Lise Morjé, and Anjali Kaushlesh Dayal. 2018. The use of force in UN peacekeeping. Interna-
tional Organization 72(1):71–103.
Hultman, Lisa. 2010. Keeping peace or spurring violence? Unintended effects of peace operations on
violence against civilians. Civil Wars 12(1–2):29–46.
Hultman, Lisa. 2014. Violence against civilians. In Routledge Handbook of Civil Wars, ed. Edward New-
man, Karl R. De Rouen, 289–299. London: Routledge.
Hultman, Lisa, Jacob D. Kathman, and Megan Shannon. 2013. United Nations peacekeeping and civilian
protection in civil war. American Journal of Political Science 57(4):875–891.
Hultman, Lisa, Jacob D. Kathman, and Megan Shannon. 2014. Beyond keeping peace. United Nations
effectiveness in the midst of fighting. American Political Science Review 108(4):737–753.
Hultman, Lisa, Jacob D. Kathman, and Megan Shannon. 2016. United Nations peacekeeping dynamics and
the duration of post-civil conflict peace. Conflict Management and Peace Science 33(3):231–249.
Karim, Sabrina, and Kyle Beardsley. 2017. Equal opportunity peacekeeping. Women, peace, and secu-
rity in post-conflict states. Oxford studies in gender and international relations. New York: Oxford
University Press.
Karlsrud, John. 2015. The UN at war. Examining the consequences of peace-enforcement mandates for the
UN peacekeeping operations in the CAR, the DRC and Mali. Third World Quarterly 36(1):40–54.
Kathman, Jacob D., and Reed M. Wood. 2016. Stopping the killing during the “peace”. Peacekeeping and
the severity of postconflict civilian victimization. Foreign Policy Analysis 12(2):149–169.
Kjeksrud, Stian, and Jacob Aasland Ravndal. 2011. Emerging lessons from the United Nations mission
in the Democratic Republic of Congo. Military contributions to the protection of civilians. African
Security Review 20(2):3–16.
Koko, Sadiki. 2011. MONUC and the quest for peace in the Democratic Republic of Congo: Assessment
of a peacekeeping mission. African Security Review 20(2):29–41.

K
Violence reduction or relocation?

Kreps, Sarah E., and Geoffrey L. Wallace. 2009. Just how humanitarian are interventions? Peacekeeping
and the prevention of civilian killings during and after civil wars. In APSA 2009 Toronto Meeting
Paper.
LeSage, James P., and R. Kelley Pace. 2009. Introduction to Spatial Econometrics. Boca Raton: Chapman
& Hall/CRC.
LeSage, James P., and R. Kelley Pace. 2014a. The biggest myth in Spatial Econometrics. Econometrics
2(4):217–249.
LeSage, James P., and R. Kelley Pace. 2014b. Interpreting Spatial Econometric models. In Handbook of
Regional Science, ed. Manfred M. Fischer, Peter Nijkamp, 1535–1552. Berlin, Heidelberg: Springer.
Melander, Erik. 2009. Selected to go where murderers lurk? Conflict Management and Peace Science
26(4):389–406.
Nadin, Peter, Patrick Cammaert, and Vesselin Popovski. 2015. Spoiler groups and UN peacekeeping.
Abingdon, New York: Routledge.
Prunier, Gérard. 2009. Africa’s World War. Congo, the Rwandan genocide, and the making of a continental
catastrophe. New York: Oxford University Press.
Quick, Ian D. 2015. Follies in fragile states. How international stabilisation failed in the Congo. London:
Double Loop.
Ruggeri, Andrea, Han Dorussen, and Theodora-Ismene Gizelis. 2017. Winning the peace locally: UN
peacekeeping and local conflict. International Organization 71(1):163–185.
Schutte, Sebastian, and Nils B. Weidmann. 2011. Diffusion patterns of violence in civil wars. Political
Geography 30(3):143–152.
Stearns, Jason K. 2011. Dancing in the glory of monsters. The collapse of the Congo and the Great War of
Africa. New York: Public Affairs.
Steinert, Janina Isabel, and Sonja Grimm. 2014. Too good to be true? United Nations peacebuilding and
the democratization of war-torn states. Conflict Management and Peace Science 32(5):513–535.
Sundberg, Ralph, and Erik Melander. 2013. Introducing the UCDP Georeferenced Event Dataset. Journal
of Peace Research 50(4):523–532.
Tollefsen, Andreas Forø, Håvard Strand, and Halvard Buhaug. 2012. PRIO-GRID. A unified spatial data
structure. Journal of Peace Research 49(2):363–374.
Tull, Denis M. 2009. Peacekeeping in the Democratic Republic of Congo: Waging peace and fighting war.
International Peacekeeping 16(2):215–230.
Tull, Denis M. 2017. The limits and unintended consequences of UN peace enforcement: The force inter-
vention brigade in the DR Congo. International Peacekeeping 25(2):167–190.
United Nations Peacekeeping. 2018a. Data. https://peacekeeping.un.org/en/data. Accessed 4 Apr 2018.
United Nations Peacekeeping. 2018b. MONUSCO. Background. https://monusco.unmissions.org/en/
background. Accessed 22 May 2018.
United Nations Peacekeeping. 2018c. MONUSCO fact sheet. https://peacekeeping.un.org/en/mission/
monusco. Accessed 22 May 2018.
United Nations Peacekeeping. 2018d. Protecting civilians. https://peacekeeping.un.org/en/protecting-
civilians. Accessed 4 Apr 2018.
United Nations Peacekeeping. 2018e. What is peacekeeping. https://peacekeeping.un.org/en/what-is-
peacekeeping. Accessed 4 Apr 2018.
United Nations Security Council. 2004. Resolution 1565. New York: United Nations Security Council.
Verweijen, Judith. 2017. Strange battlefield fellows. The diagonal interoperability between Blue Helmets
and the Congolese army. International Peacekeeping 24(3):363–387.
Vogel, Christoph. 2013. Congo. Why UN peacekeepers have a credibility problem. The Guardian, 30 Aug
2013.
Vries, Hugo de. 2015. Going around in circles. The challenges of peacekeeping and stabilization in the
Democratic Republic of the Congo. The Hague: Netherlands Institute of International Relations Clin-
gendael.
Walter, Barbara F. 1997. The critical barrier to civil war settlement. International Organization 51(3):
335–364.
Ward, Michael Don, and Kristian Skrede Gleditsch. 2008. Spatial Regression Models. Quantitative appli-
cations in the social sciences, Vol. 155. Thousand Oaks: SAGE.

Вам также может понравиться