Politics of Transitional Justice
Examining Arrests of Former Wartime Leaders as An Electoral Manipulation Strategy in Post-Conflict Countries
Triveni Chand Master's Thesis
Spring 2020
Department of Peace and Conflict Research, Uppsala University
Supervisor: Anders Themnér
ABSTRACT
The systematic variation in arrests of former wartime leaders (including political/military leaders and those with commanding positions from both sides of conflict among other high-level wartime actors) in post-conflict countries have rarely been recognized and studied. Building on past literature that interlinks transitional justice with domestic politics, this study argues that the variation in arrests of former wartime leaders can be explained by elections and electoral manipulation theory. Amid the costs and opportunities associated with elections in general, I argue that incumbents also opt for arrests of former wartime leaders as an electoral manipulation strategy to eliminate political opponents and consolidate power in the guise of justice and, at the same time, minimize the costs associated with electoral manipulation tools. Hence, I hypothesize the arrest of former wartime leaders likely to be during the election period (the pre-election period, election day, and immediate post-election period). All else equal, the statistical test does not support the hypothesis while the complementary evidence from post-conflict Nepal and Sri Lanka suggest that presence (or absence) of justice in post-conflict countries is largely shaped by domestic politics. Similarly, few arrests in Sri Lanka and Nepal offer mild support to the theoretical expectations while few other arrests in Sri Lanka suggest that some arrests during the hypothesized election period are coincidental. This further questions the explanatory power of the suggested theory and findings.
Key Words: Post-conflict Countries, Transitional Justice, Criminal Prosecution, Former
Wartime Leaders, Arrests of Former Wartime Leaders, Election Manipulation
ACKNOWLEDGMENTS
I want to express my gratitude toward my supervisor Anders Themnér for being a great
supervisor and generously providing me with the unpublished dataset for this study. Thank
you, Anders, for providing helpful comments and suggestions and addressing my queries on
the dataset. I also want to extend my deep graduate to Henrik Persson who took out the time to
address my questions concerning the dataset and providing me with additional documents. A
great thank you to Lisa Hultman for providing me with the peacekeeping dataset and feedback
on quantitative parts. Heartfelt thank you to my family for all the love and support.
TABLE OF CONTENTS
List of Abbreviations ... 6
List of Figures, Tables, and Annexes……...7
1. INTRODUCTION...8
2. THEORETICAL FRAMEWORK………...…….. 12
2.1 Previous Research ...12
2.1.1 Issues of Stability vs. Justice...12
2.1.2 Relative Power of Outgoing and Incoming Elites ...13
2.1.3 International Support and Pressure ...14
2.1.4 “Politics of Present” ...14
2.2. Election and Electoral Manipulation...15
2.2.1 Defining Elections...15
2.2.2 Election in Post-conflict Countries ...15
2.2.3 Defining Electoral Manipulation ...16
2.2.4. Electoral Manipulation: Explaining Behavior and Strategies of Incumbents.17 2.2.5. Electoral Manipulation Timing ... 18
2.3 Main Argument of this Study ... 19
3. RESEARCH DESIGN ... 24
3.1 Dependent Variable: Arrests of Former Wartime Leaders ... 24
3.2 Independent Variable: Elections ... 26
3.3 Control Variables ... 27
3.4 Statistical Model……….…..……….. 30
4. ANALYSIS ... 31
4.1 Description of Data... 31
4.2 Findings ... 34
4.3 Discussion... 40
4.3.1 Discussion on Impunity and Arrests in Post-Conflict Nepal ... 41
4.3.2 Discussion on Impunity and Arrests in Post-Conflict Sri Lanka... 44
4.3.3 Discussion on Additional Observations ... 54
4.4. Alternative Explanation ... 57
4.5. Limitations and Potential Bias ... 59
4.5.1 Theoretical Limitation ... 59
4.5.2 Research Design and Statistical Limitation... 60
5. CONCLUSION ... 62
6. ANNEXES ... 65
References ...71
ABBREVIATIONS
CL Civil Liberties
CPA Comprehensive Peace Agreement
CPN-Maoist Communist Party of Nepal-Maoist
DAC Development Assistance Committee
EU European Union
HRW Human Rights Watch
ICC International Criminal Court
ICTR International Criminal Tribunal for Rwanda
ICTY International Criminal Tribunal for Former Yugoslavia
International IDEA International Institute for Democracy and Electoral Assistance
IPU Inter-Parliamentary Union
JVP Janatha Vimukthi Peramuna/People’s Liberation Front
LPM Linear Probability Model
LTTE Liberation Tigers of Tamil Eelam NATO North Atlantic Treaty Organization
NELDA National Elections across Democracy and Autocracy
NPP National Patriotic Party
OECD Organization for Economic Co-operation and Development
PL Political Rights
RO Regional Organization
RRF Rwandan Patriotic Front
SLD Sojusz Lewicy Demokratycznej/Democratic Left Alliance
SLFP Sri Lanka Freedom Party (SLFP)
TRCs Truth and Reconciliation Commissions UCPN-Maoist United Communist Party of Nepal-Maoist
UCDP Uppsala Conflict Data Program
UN United Nations
UNP United National Party
UDMF United Democratic Madhesi Front
YCL Young Communist League
LIST OF FIGURES, TABLES, and ANNEXES
Figure 1: Distribution of Arrests in Post- Conflict Countries...32 Figure 2: Distribution of Elections in Post- Conflict Countries...33 Figure 3: Chi-squared Test for Tabular Association…………...34
Table 1: Logit Estimates on the Likelihood of Arrests in Post-conflict countries, 1989-
2017………..36 Table 2: Logit Estimates on the Likelihood of Arrests in Post-conflict countries after
Omitting the Arrests Carried out after the Immediate Post-election period, 1989-
2017... 39
Annex I: Table of Summary Statistics... 65
Annex II: Logit Estimates of the Likelihood of Arrests in Post-conflict Countries without
Lagged Variables, 1989-2017... ...66
Annex II: An overview of the arrests of former wartime leaders coinciding with elections in
post-conflict countries, 1989-2017...67
Annex IV: Frequency Distribution of Arrests of Former Wartime Leaders in Post-Conflict
Countries, 1989-2017...70
Annex V: Frequency Distribution of Elections of Former Wartime Leaders in Post-Conflict
Countries, 1989-2017...70
1. INTRODUCTION
The transitional justice field has grown exponentially and gained global recognition over the last 20 years (Nesiah 2016: Loyle & Davenport 2016). The growth is observable in its institutionalization within the United Nations (UN), regional organizations, and government agencies, legal and policy developments, and an expanded network of practitioners and scholars working in the field (Nesiah 2016). Such growth has been contributing to establish transitional justice as a customary international norm/law to reckon with the legacy of past atrocities (Nagy 2008; Buckley-Zistel et al. 2014). Based on Kritz (1995), Höglund & Orjuela (2013) define transitional justice as “a concept which has been used to denote mechanisms for addressing past human rights atrocities in countries emerging from violent conflict or authoritarian rule” (302-303). The institutions and mechanisms of transitional justice encompass criminal accountability through International Criminal Court (ICC), hybrid courts, and national court, Truth and Reconciliation Commissions (TRCs), and reparations programs to name a few (Nagy 2008; Loyle & Davenport 2016). Similar to other mechanisms of transitional justice, criminal prosecution is associated with ending the culture of impunity, reinforcing commitment towards the rule of law, and more importantly providing justice to victims, thereby opening the gateway to reconciliation (Oko 2003; Nuzov 2014; Greenstein &
Harvey 2017). However, sometimes the high-level prosecution focused on top leadership and commanding positions are considered sufficient to establish the truth of the past and hear the victim’s voices, especially in response to logistic necessities when a majority of the populations are directly or indirectly linked to past atrocities (Koskenniemi 2002). Individual prosecution also helps to avoid the culture of collective guilt and retribution that can eventually reignite a conflict (Grodsky 2008).
Primarily motivated by the growing field of transitional justice and the focus on the accountability of former wartime leaders/commanders from both sides of a conflict, this study seeks to answer what explains the variation in arrests of former wartime leaders in post- conflict countries
1over time? For example, the conflict/war in Guatemala ended in 1996, but the government carried out the first two high-level arrests for alleged war crimes only in 2006 (Almqvist & Esposito 2011). There was no news of other high-ranking arrests until Hector Bol de la Cruz, the former Police Chief was arrested in 2011 (see Kreutz et al. 2019a Dataset).
1
In this study, post-conflict countries refer to countries that had experienced intrastate armed conflict; therefore,
interchangeable with civil war/war later. Intrastate armed conflict follows the definition of the Uppsala Conflict
Data Program (UCDP) which defines armed conflict as “a contested incompatibility that concerns government
and/or territory where the use of armed force between two parties, of which at least one is the government of a
state, results in at least 25 battle-related deaths in one calendar year” (UCDP (a)).
Similarly, in Bosnia and Herzegovina, arrests linked to past crimes started shortly after the conflict ended in 1995 but a considerable number of arrests of former high-ranking officials took place only in the latter part of 2004 and early 2005 (Kerr 2005). Adopting from the unpublished codebook of Kreutz et al. 2019b,
2the former wartime leaders/commanders refer to wartime political leaders and senior government officials in government,
3those with commanding positions in military/armed government units,
4leaders/commanders of paramilitary/militia group(s) allied with the government, political representatives/leaders and senior commanders of rebel groups/non-state armed groups (collectively, these high-level actors are hereafter referred to as former wartime leaders).
An examination into past and current literature reveal that the variations in arrests of former wartime leaders have neither been exclusively studied nor been recognized except rarely like in the case of Bosnia and Herzegovina above. Although a research gap exists, past research offers valuable knowledge on what explains the presence (or absence) of transitional justice and its mechanisms in general. Within their various theoretical framework and argument, the past research more or less agrees that transitional justice and its mechanisms at a domestic level is influenced by political power struggles (e.g. Welsh 1995; Szczerbiak 2002; Subotić 2009;
Grodsky 2009; Nuzov 2014).
Considering that political power struggle heightens during elections which makes incumbents resort to different electoral manipulation tools to maintain and consolidate power (further discussed in the theory section), this research presents the arrests of former wartime leaders as a viable electoral manipulation strategy and hypothesize that arrests are likely to be carried out against the former wartime leaders during the election period (i.e. pre-election period, election day and immediate post-election period). The arrests make up a viable electoral manipulation tool because it provides an opportunity to target the top leadership of opposition (e.g. Nouwen & Werner 2010), help downgrade the image of opposition while elevating its image as a moral authority (e.g. Nouwen & Werner 2010), adds credibility to the elimination
2
The codebook is part of an ongoing project currently named, “Arrested Development: The Dangerous Mix of Patronage and Justice in Post-conflict Countries” with the project leaders as Joakim Kreutz (Department of Peace and Conflict Research, Uppsala University and Department of Political Science, Stockholm University) and Anders Themnér (Department of Peace and Conflict Research, Uppsala University).
3
Political leaders and senior government officials concern “heads of state, prime ministers, members of cabinet or legislature, or senior government officials in state agencies/departments” (Kreutz et al. 2019b: 1).
4
Military leaders and those with commanding positions concerns military leaders, generals, chief of staff, senior
commanders, commanders, senior officers while the official armed government units include “armed forces, navy,
air force, police, security agencies” (Kreutz et al. 2019b: 1).
of political opponents (e.g. Kirchheimer 1961), and provides a disguise for other transgressions (e.g. Oko 2003; Loyle & Davenport 2016). Understanding the factors that influence the likelihood of arrests built on the premise of the politicization of justice has twofold benefits.
First, exploring the conditions under which the states employ justice is essential to give clarity to the reasons for the success or failure of transitional justice, its policies, and mechanisms (Grodsky 2008). In other words, understanding the conditions under which justice is employed helps improve the outcome of transitional justice without disregarding the justice mechanism altogether (Ibid). Similarly, Nesiah (2016) points out that when transitional justice stands as a depoliticized technical framework with universal relevance, it gives rise to ill-preparedness in response to political actors whose goals might be different (and even harmful) than serving justice. Therefore, studying the variation in arrests of wartime leaders and exploring its political aspect has both academic and policy implications.
In addition to addressing the theoretical gap by connecting arrests of former wartime leaders with elections and electoral manipulation theory, this study makes an empirical contribution by statistically testing the relationship between arrests of former wartime leaders and election while controlling for the other variables suggested by the past research. The statistical analysis includes 64 post-conflict countries by country-year (N=1412) from 1989 to 2017, predominantly built on unpublished Kreutz et al. (2019a) dataset on arrests of former wartime leaders with some additional coding to complement the dataset.
5The statistical findings are comprehensively discussed in the light of evidence from post-conflict Nepal and Sri Lanka which helps to trace the suggested causal mechanism to some extent.
Initially, a statistically significant positive association between arrests of former wartime leaders and elections is observed, all else equal. However, having a country-year as the unit of analysis invites the problem of some arrests taking place after the hypothesized election period (i.e. pre-election period, election day, and immediate post-election period) in the given election year. Once the arrests that were carried out after the hypothesized election period are omitted, the initially observed statistically significant relationship no longer holds, all else equal. The evidence from Nepal and Sri Lanka suggests that lack of arrests or presence of impunity interplay with domestic politics, alongside providing mild support to the suggested theoretical mechanism. However, few other arrests in Sri Lanka show the possibility of some arrests during the hypothesized election period as coincidental. With the acceptance that this study
5
The additional coding was done based on my internship experience in the project which largely focused on
coding arrests of former wartime leaders. The additional coding was also cross-checked with Anders Themnér
who is one of the project leaders and also my advisor for this paper.
faces a major setback on its research design, the statistically significant relationship that is initially observed between election years and arrests compared to lack of significant association between arrests and other variables based on the past research opens the possibility for further research.
The next chapter on the theoretical framework begins with a brief review of the relevant
literature. After reviewing the past literature, the theory on elections and electoral manipulation
are presented by first defining election, second by establishing the importance of election in
post-conflict countries, third by explaining behavior and strategies of incumbents, fourth by
establishing electoral manipulation timing, and lastly by arguing why arrests of former wartime
leaders serve as a viable electoral manipulation strategy. The third chapter presents the research
design, followed by an analysis in the fourth chapter. The discussion on statistical findings is
complemented with the evidence from post-conflict Sri Lanka and Nepal. The fourth chapter
ends with a discussion on additional statistical observations, an alternative explanation to the
outcome observed, and lastly presents limitations and potential bias of this study. The final
chapter concludes by providing potential directions for further research.
2. THEORETICAL FRAMEWORK
This chapter first presents and discusses existing literature in the field of transitional justice.
It concludes that the existing studies partly agree that the presence (or absence) of transitional justice in the aftermath of a conflict interplays with domestic politics within their main theoretical framework and argument. I connect the premise of politicized justices with elections and electoral manipulation theories to explain variation in the arrests of former wartime leaders in post-conflict countries. Then, the hypothesis to be tested later is presented.
2.1 Previous Research
The literature on transitional justice has generally been focused on “what is being done and what should be done” to the alleged perpetrators of crimes committed in the past (Loyle &
Davenport 2016). The limited literature explores the presence (or absence) of transitional justice mechanism in general which can be mainly divided into (1) the issues of stability vs.
justice (e.g. Nagy 2008; Grodsky 2008), (2) the relative power of old and new political elites (e.g. de Brito 1997; Grodksy 2008; Nalepa 2010), (3) international pressure and/or support (Subotić 2009), and lastly (4) “politics of the present” (e.g. Welsh 1996; Szczerbiak 2002;
Nuzov 2014).
2.1.1 Issues of Stability vs. Justice
The age-old argument that has been used to justify the absence of transitional justice mechanisms, especially the prosecution of alleged perpetrators for past crimes in a post-conflict period, is the issues of stability vs. justice. It is argued that seeking justice, especially during the fragile transitional period can make conflict parties withdraw their commitment to peace, thus reigniting a conflict (Nagy 2008). Such clemency is argued to harm domestic consolidation as the practice of burying the past can mean burying the ethical values without which safe and peaceful societies are unlikely (Grodsky 2008). Nevertheless, stability vs.
justice remains a contested topic among practitioners and scholars alike (Nagy 2008). Even
political actors in post-conflict countries use the issue of stability when pushed to deal with a
violent past (Ibid). For example, the Kenyan government was sometimes found seeking
deferral of the ICC cases by claiming that the prosecution will hamper the peace and stability
in the country (Hansen 2013). On the other hand, some Kenyan political elite supported
different justice processes which Hansen (2013) claims were based on various private
considerations. One of such considerations was the prospects of targeting political opponents
under the guise of justice (Ibid).
2.1.2 The Relative Power of Old and New Political Elites
The second sphere of argument focuses on the relative power of old and new elites determined by (a) types of political transition and (b) types of state structure following a conflict. Concerning relative power based on political transitions, literature (e.g. de Brito 1997; Peskin & Boduszyn´ski 2003; Pinto 2008; Grodsky 2008) often refer to Huntington (1991) who argues that the types of transition affect political power distribution in a post- conflict period, consequently affecting justice options available to new elites. Huntington (1991) proposes that a harsh or aggressive form of justice is more likely in states where the transition was brought through a revolution as the power base of the former regime that determines their ability to cause problems to new elites is presumably destroyed (cited in Grodsky 2008). On the other hand, the new elites are likely to pursue a softer form of justice if a transition from a conflict to a post-conflict state was negotiated (Pinto 2008). The negotiated transitions are often bound by explicit or implicit amnesties and the share of power offered to the members of old regimes (Grodsky 2008).
Within the relative power theory lies the state structure argument where alleged
perpetrators of past crimes continue to hold powerful military and bureaucratic positions in the
post-conflict period (Grodsky 2008). With a significant influence and resources at their
disposal, pursuing justice against the alleged perpetrators becomes risky and difficult (Smith
2012). Grodsky (2008) adds that even lower-level officials at large numbers with alleged
associations to past crimes can collectively pose an equal threat if they continue to become part
of the state structure. Overall, the broader bureaucracy made of alleged perpetrators in powerful
positions and others with a feeling of complicity in past crimes affects the power and control
of the new elites, thus limiting the pursuit of justice (Grodsky 2008; Subotić 2009; Nalepa
2010). Subotić (2009) points out that types of political transition can have a significant effect
on post-conflict state structures. The transition achieved through stronger compromise or
negotiation is more likely to have loyalists of an old regime embedded in the new state
structures (Ibid). Smith (2012) agrees that the various ways in which conflict ends and various
ways of obtaining and maintaining power shapes the political advantages of those who come
to power which in return determines how much justice is served.
2.1.3 International Support and Pressure
The third sphere of literature focuses on international pressure that partly affects the state’s perusal of justice. Subotić (2009) argues that international actors can induce state compliance by “offering” both carrots and sticks. The carrots are in the form of material rewards such as membership in international organizations and/or foreign aid and investment while sticks are in the form of withholding aid and imposing sanctions (Ibid). However, Subotić (2009) adds that international pressure interplays with domestic politics which affect the degree of compliance with various justice mechanisms. Subotić (2009) elaborates that political elites weigh domestic costs of compliance with justice mechanisms and comply with some international requirements while ignoring others based on its ability to derive domestic benefits. For example, the Sudanese government arrested two senior officials that were indicted by the ICC to defer from arresting then-President Omar Hassan Ahmad al-Bashir (Ibid). The arrests followed with an announcement that they will be holding domestic war-crimes trials for those two who were arrested (Ibid). This move was seen more as a political stunt to improve its international image while controlling the process by conducting trials domestically (Ibid).
Based on the observation of Kenya and its compliance to international pressure, Mueller (2014) finds that state actors and alleged perpetrators decreased their cooperation and even took measures to defy international justice institutions when the political risk of compliance increased. On the other hand, Subotić (2009) finds that the domestic actors are likely to agree to some form of justice if it eliminates political opponents.
2.1.4 “Politics of the Present”
The fourth sphere of literature focuses on the “politics of present” which argues that issues of accountability of the past continue to exist beyond the transition period (Welsh 1996).
Welsh (1996), Szczerbiak (2003) and Kiss (2006) agree that the necessity to deal with past does not just disappear with the time rather it can become even more salient and politicized over time depending on the strategic interests of key political actors (cited in Elbasani &
Lipinski 2011). For example, the Polish politics was polarized along the line of 'communist
forgiving'-'communist purging' axis and aggressive campaigns and sharp anti-communist
rhetoric were often deployed to undermine the opponents (Szczerbiak 2002). Nuzov (2014)
makes similar observations in Russia where the soviet past and alleged crimes committed
during that period was used to discredit the opposition (Nuzov 2014). As a result, the past and
need to address it becomes a recurring issue in the political power struggle where the past
becomes exploited by some to undermine their opponents (Welsh 1996).
The existing literature has its shortcomings. For example, the relative power argument directly/indirectly assumes that power is static after the transition type decides “winners” and
“losers” of conflict, but in reality, political elites continuously face the challenge to maintain their political positions and influence (Grodsky 2008). Besides, there are many instances when authoritarian leaders who were forced out of power were able to maintain varying degrees of influence over the military and political process in a post-conflict period (Peskin &
Boduszyn´ski 2003). Similarly, even in presences of amnesties where criminal prosecution holds little chance, there are examples where new elites have annulled the decisions of predecessors (Grodsky 2008). Moreover, the majority of the literature (example, Welsh 1995;
Kiss 2006; Subotić 2009; Grodsky 2009; and Nuzov 2014) has been guided by the study of Balkan countries and post-communist countries, questioning its ability to explain justice in other types of conflicts. Subotić (2009) likewise points out that transitional justice literature is overwhelmingly focused on a single case study, thus limited in its generalizability.
Furthermore, the literature has limited leverage in predicting when the politicization of justice is more likely, let alone predicting the likelihood of arrests of former wartime leaders in post- conflict countries.
To fill the gap in our understanding of when arrests of former wartime leaders are more likely, I introduce theories on elections and electoral manipulation in the following section.
The section on electoral manipulation focuses on how elections lead to uncertainty which brings the need to eliminate political opponents while at the same time offers a platform to convey domination of incumbents, thereby consolidating power. Thereafter, I present my theoretical argument on how arrests of former wartime leaders can be considered as one of the viable tools of electoral manipulation, consequently making arrests more likely during the election period.
2.2 Elections and Electoral Manipulation
2.2.1 Defining Elections
Adopting from Hyde & Marinov (2019) codebook, elections in this study refers to general
elections at a national level for “a national executive figure, such as a president, or for a national
legislative body, such as a parliament, legislature, constituent assembly” (2). The person(s)
contesting for the mentioned positions must be directly voted by the people (Ibid). This study
includes national elections for both executive figures and legislative bodies because both have
great costs and benefits associated with it. For example, the election for executive figures by
name highlights the importance of the election, i.e. electing the highest position of a country.
Similarly, elections for legislative bodies encompass different stakes associated with different governmental bodies and institutions (Simpser 2013). For example, the plurality of seats in legislative elections ensures the privilege of forming the government (Ibid). Sometimes parties can unilaterally amend the constitution in case of an absolute majority or two-thirds majority of the seats in a legislative body (Ibid). Overall, elections in this study constitute national elections for national executive figures, or national legislative bodies voted directly by the people. These elections do not have to be free and fair, rather the existence of electoral politics is considered sufficient for this study which is further addressed in the research design.
2.2.2 Election in Post-Conflict Countries
In the last two-decade, democratization has become a prescribed post-conflict rebuilding mechanism (Themnér 2017). While ample argument can be made about the quality of democracy that follows as a part of a rebuilding mechanism in post-conflict countries, elections have become a common phenomenon even in “pseudodemocracies,” “virtual democracies,” and “competitive authoritarian” regimes (Levitsky & Way 2002; Magaloni 2010; Hyde & Marinov 2012; and van Ham & Lindberg 2015). In other words, elections have become a common feature in the majority of regime types that act as a formal requirement to gain and/or retain power (Birch 2011). This establishes that we are likely to see elections in post-conflict countries although it might be less frequent and less free and fair in some countries than others.
2.2.3 Defining Electoral Manipulation
The process of influencing the electoral process is known as electoral manipulation which is prevalent in many political systems and can take many forms (Lehoucq 2003; van Ham &
Lindberg 2015). The manipulation of electoral institutions relies on legal tactics where
incumbent governments alter electoral rules in their favor; the renowned example is
gerrymandering where electoral boundaries are delineated to increase the likelihood of a
favorable outcome (Birch 2011). Other common manipulation strategies involve an arbitrary
denial of candidates or party registration/nomination, removal from contestation on arbitrary
reasons, and vote-buying (Ibid). Sometimes incumbents can go to the extent of spying,
threatening, harassing, and arresting opposition politicians as well as other government critics
(Levitsky &Way 2002). In the past, opposition members have also been jailed, exiled, and in
some cases assaulted or murdered (Ibid). Birch (2011) points out that there are few forms of
electoral manipulation that non-incumbent political candidates or parties undertake though most evidence shows the majority of electoral manipulation or violation is committed by an incumbent government. Following that majority of electoral manipulation is carried out by an incumbent government, this paper is limited to exploring theories on electoral manipulation concerning incumbents.
2.2.4 Electoral Manipulation: Explaining behavior and Strategies of Incumbents Elections bring a risk of losing the power which leads to the necessity of eliminating political opponents while at the same time elections offer the opportunity to convey the dominance for power consolidation. First, the necessity of holding elections comes with the uncertainty of who will become “losers” and who will become “winners” (Birch 2011). The uncertainty translates into a threat of losing the dominant position for incumbents (Hafner- Burton et al. 2014). The possibility of losing power is riskier in post-conflict countries because political control often becomes only means to access and ensure political, economic, and social benefits and in some instances, only means of physical security as losing elections can invite dire consequences, such as harassment, exile and even execution (Höglund 2008 cited in Themnér 2017). The high stakes associated with election prompts incumbents to influence the electoral process (Davenport 1997; Höglund 2009; Hafner-Burton et al. 2014). There is even a common conviction that political players are always ready to alter the rules to secure its benefits (Kaminski 2002). van Ham & Lindberg (2015) argue that incumbents seek to influence election outcomes by intimidating opposition candidates/parties and voters to an extent that it sufficiently reduces competition in favor of incumbents. The political opposition is likely to boycott the election when faced with harassment, such as imprisonment and torture of opposition candidates (Hafner-Burton et al. 2018). The withdrawal of opposition parties before the election significantly increases the probability of the incumbent government winning the election (Hafner-Burton et al. 2014). It also creates a bias in the voter turnout in favor of incumbents as it coerces opposition supporters either to boycott voting altogether or intimidates them into voting for the incumbent government (Hafner-Burton et al. 2018).
In addition to causing opposition boycott and biasing voters, electoral manipulation is
deployed to convey the domination of incumbents, thereby influencing the decision and
behavior of social and political actors. Simpser (2013) highlights that elections, alongside
providing the platform for power contests, offer an opportunity to transmit or distort
information about the incumbent’s resources, institutional dominance, and organizational
capacity. By signaling their power through electoral manipulation, incumbents can influence
the behaviors and decisions of key political and social actors in favor of incumbents and against oppositions (Greenstein & Harvey 2017). For example, when incumbents show the capability to maintain their power through different means, donors who might have originally planned on financing opposition party/candidates or were indecisive on whom to support are likely to choose incumbents over oppositions (Simpser 2013). The showcase of strength and dominance of incumbents is also likely to prevent possible elite splits, attract ambitious politicians, and ensure compliant behavior from bureaucrats, voters, and other socio-political actors (Greenstein &Harvey 2017).
Although incumbents would prefer to maintain and consolidate power, they face a trade- off between electoral success and domestic and international legitimacy when deploying electoral manipulation tools (Birch 2011). The loss of domestic legitimacy affects the ability of leaders to rule without coercive means (Ibid). There are instances where the loss of domestic legitimacy due to alleged electoral manipulations have led to large-scale protests and attempted coups d’état or violent rebellions (van Ham & Lindberg 2015). It has also become harder to gain international legitimacy through a poor-quality electoral process as, in the last two-decade, electoral integrity has gotten momentum (Birch 2011). Hafner-Burton et al. (2014) add that the presence of institutional constraints that seek accountability of the government decisions and actions affects the incumbent’s choice of deploying election violence. Therefore, leaders seek to adopt electoral strategies that would both retain and consolidate power by maintaining an acceptable level of legitimacy (Birch 2011).
2.2.5 Electoral Manipulation Timing
When studying election violence, a type of electoral manipulation tool, Höglund (2009) points out that election violence is often clustered around election times: (1) the pre-election period, (2) election day/s, and (3) the immediate post-election period. According to Höglund (2009), “a number of events and shifts focus from the day-to-day politics to elections” can mark the initiation of a pre-election phase (416). Such events are usually the start of voters or party registration or the start of a campaign period (Ibid). However, different countries vary when it comes to the span of pre-election as some countries might take a longer period even for voter registration due to lack of basic documentation and popular census (Ibid).
Regarding the post-election period, Höglund (2009) defines it as “the period leading up
to the inauguration of the newly elected body,” and during this period violence can take place
during vote counting, analyzing results and even continue after result announcement when
losing parties are reluctant to accept the result (416). She mentions that “The incumbent
government that does not want to leave its position of power may clamp down hard on the opposition” (419). Elklit & Reynolds (2005) and Norris (2014) also agree that electoral manipulation starts as early as from the pre-election legal framework to party/candidate registration and campaigning, to the voting day, and continues to vote counting and the formal announcement of results in the post-election period (cited in van Ham & Lindberg 2015).
However, regarding election violence, Höglund (2009) observes that it varies in its form and intensity during different periods clustered around the election. Other electoral manipulation tools could suffer from similar faith depending on situations, like the presence of international election observers, which is beyond the scope of this study.
2.3 Main Argument of this Study
To recapitulate, incumbents generally manipulate elections as a strategy to maintain and consolidate power. Since previous research more or less agrees that domestic actors can politicize transitional justice mechanisms to eliminate political opponents among pursuing other private agendas, I anticipate that arrests of former wartime leaders follow a similar pattern as other tools of electoral manipulation in this regard. As other tools of manipulation suffer from legitimacy issues and other costs, my assumption is that incumbents would still like to continue manipulating elections to retain and consolidate power, but they will also opt for arrests in the guise of transitional justice to minimize costs associated with electoral manipulation and, at the same time, fulfill the goal of eliminating oppositions and consolidating power. The arrests of former wartime leaders in post-conflict countries is a viable electoral manipulation tool for incumbents because it provides an opportunity to target the top leadership of opposition (e.g. Nouwen & Werner 2010), help downgrade the image of opposition while elevating its image as a moral authority (e.g. Nouwen & Werner 2010), adds credibility to the elimination of political opponents (e.g. Kirchheimer (1961), and provides a disguise for other transgressions (e.g. Oko 2003; Loyle & Davenport 2016). While all these “qualities” make arrests of former wartime leaders a viable electoral manipulation tool, in reality, one or combination of these “qualities” could be sufficient to make an arrest.
First, in post-conflict countries, an increasing number of past leaders enter the electoral
arena rather than altogether quitting from the present political scene, especially with the
majority of conflict ending with peace agreements (Schedler 2002). The former wartime
leaders transition into a post-conflict period in different ways; some transform their armed
groups into political parties, some join established political parties, and some opt for forming
a completely new political organization with no ties to the formal armed group (Themnér
2017). They also differ in political positions they aim for as some run for the presidency and some for senator position or house of representatives, regardless of positions, they continue to function as key players in forming political alliances during national elections (Themnér 2017).
But the political and influential position does not exempt them from their past (Ibid). Also, as established before, there is an expectation that at least senior officials responsible for worst atrocities should be prosecuted (Kauffman 2005). Hence, transitional justice provides a reason to target these top leadership, and as per Nouwen & Werner (2010), removal of top leadership is expected to collapse the whole group. In Liberia, for example, Charles Taylor (former president of Liberia) was regarded as the “godfather” of the National Patriotic Party (NPP) (Themnér 2017). When Taylor had to resort to exile due to alleged connection to past atrocities, NPP started weakening with its remaining key leaders striking deals with political opponents (Ibid). Even if the whole group does not collapse, Silitski (2005) argues that when the stronger opposition candidates are denied entry or disqualified from an electoral race, the victory of incumbents is almost undeniable.
Second, transitional justice is becoming a fundamental universal norm to address atrocities committed during the conflict period (Nouwen &Werner 2010). Such universalism provides an opportunity to label opponents as the “enemies of mankind” (Schmitt 1996 cited in Nouwen & Werner 2010). Mamdani (2007) even adds that “war criminals” “genocidaire”
have become labels “to be stuck on your worst enemy, a perverse version of the Nobel Prize, part of a rhetorical arsenal that helps you vilify your adversaries while ensuring impunity for your allies” (Ibid: 957). This logic could be followed not only to those who commit crimes but also to those who do not support the accountability of such atrocities. Nouwen & Werner (2010) claim that vilifying political opponents and its supporters emphasize actions of political elites who unlike others are willing to respect the universal norm. They even predict that the incumbents are more likely to execute ICC-issued arrest warrants when there is a greater likelihood of its suspect being recognized as the “enemies of mankind.”
Third, Kirchheimer (1961) points out that the political usage of criminal prosecution
does not necessarily preclude its legal character. Rather, the legal means allows using legal
fiction to conceal any political motivation of prosecuting the opponent (Ibid). In other words,
unlike other means to eliminate a political competition, legal means offer legitimation to
eliminating opponents (Ibid); however, the legitimacy gained through this process is likely to
vary based on how much the judicial system is considered independent and impartial. For
example, the ICC itself has been perceived as biased towards the “West” as the crimes
committed by Western countries have been considered ignored while African leaders share the
feeling of being scrutinized for their actions (Mills & Bloomfield 2018). Similarly, domestic judicial systems, especially in post-conflict countries, might suffer from corruption and other shortcomings. Therefore, how much justification transitional justice provides to eliminate political opponents may depend on perceived independence of domestic and international judicial institutions from national and international political struggle. Nevertheless, Posner (2005) when discussing the usage of legal means to eliminate opponents argues that when incumbents believe that political opponents have more chances to win in the election, prosecuting them remains an attractive choice regardless of questionable judicial institutions.
Fourth, the issue of accountability for past crimes, which in this case is brought by the arrest of former wartime leaders, provides cover for other transgressions of the incumbent (Oko 2003). Loyle & Davenport (2016) agree that the justice mechanism can be used as a disguise to curtail civil liberties and legitimize state repression. Repression and similar activities have often been established as a means for political consolidation (Ibid). Thus, the implementation of the justice process holds the possibility to divert domestic and international attention from repressive activities which contribute to the consolidation of power (Ibid).
Lastly, one can argue that political opponents can use the same rhetoric if the
incumbents have ties to past atrocities. Meaning, if incumbents themselves were part of past
atrocities or have alliances with former wartime leaders, there are chances that oppositions use
alleged links to past crimes and lack of accountability to vilify incumbents. In such cases, I
maintain that arrests will still take place so that incumbents can cut ties to such a past, especially
considering that political actors turning against their allies is neither a new concept nor a new
practice. For example, Mainwaring & Zoco (2007) observe that political parties seek to
distance themselves from their party labels in the presence of widespread dissatisfaction with
what they represent. Such activities are common during elections as those party labels become
hindrances to gaining positive popularity, giving little reason to remain faithful to party labels
and sometimes even the party altogether (Ibid). Even earlier in the case of Liberia, we observed
that key leaders of NPP struk deals with political opponents when former President Taylor had
to resort to exile due to his alleged connection to past atrocities (Themnér 2017). In Sri Lanka,
it is even legal for elected parliamentarians to switch from the party that they got elected from
to another party and still keep their seat (DeVotta 2014). Following such practices in the game
of political struggle, we can expect that incumbents are likely to carry out arrests against the
former wartime leaders even if they are allies to distance themselves from its consequences
during the election period.
While I have established that arrests of former wartime leaders can act as an attractive electoral manipulation tool for the incumbents, there is an essential consideration to be made before hypothesizing the relationship between arrests and elections. In addition to a government/incumbent making the arrests and being the indicting actor, there is a possibility that in some arrests the indicting actors are ICC or other states. Since both ICC/other states interact with domestic politics, the inclusion of arrests made within the relevant post-conflict countries regardless of such possibility is reasonable. There is a common argument in the transitional justice literature that resonates with the ICC being only as strong as its enforcement capacity (e.g. Penrose; Zhou 2006; Neumayer 2009). Meaning, the ICC does not have a police force of its own and is tied by the issue of state sovereignty; therefore, it requires state cooperation for the arrest of accused persons, their extradition, and other support throughout prosecutions (Penrose 2000; Zhou 2006). This holds for other international indicting actors who are also constrained by issues of sovereignty and voluntary compliance of the relevant country (Penrose 2000). For example, the Rwandan government sometimes strategically constrained the functioning of the International Criminal Tribunal for Rwanda (ICTR) by interfering with investigations, withholding evidence, denying passports and visas to witnesses (Simmons & Danner 2010). This resulted in the ICTR mandate being restricted in its prosecution of the Rwandan Patriotic Front (RPF), the ruling political party, atrocities (Peskin 2005). Loyle & Davenport (2016) point out that such cases serve as the reminder that international prosecution will cease to function without government cooperation. In addition, the success of ICC is largely measured by its completed cases which makes it essential to maintain a cordial relationship with the state actors (Nouwen & Werner 2010). Therefore, the issue of sovereignty and necessity to maintain relationships with the host country government is likely to align arrests with incumbents’ private gains though it should be accepted that international indictments can bring with itself a unique power compared to domestic indictments.
In sum, past literature interlinks transitional justice with domestic politics which in some cases might be used to eliminate political opponents and pursue other private agendas.
Considering that incumbents also seek to manipulate elections to maintain and consolidate
power, I anticipate that arrests of former wartime leaders follow a similar pattern as other tools
of electoral manipulation. Since other electoral manipulation tools suffer from legitimacy
issues and other costs, my assumption is incumbents would like to continue manipulating
elections. Hence, they also opt for arrests in the guise of transitional justice to minimize the
costs associated with electoral manipulation while securing the elimination of political
opponents and consolidating power. Establishing that the arrests are an attractive tool of electoral manipulation for the incumbents and considering that electoral manipulation is expected to cluster around (1) the pre-election period, (2) election day/s, and (3) the immediate post-election period, I hypothesize that:
H1: Arrests are more likely to be carried out against the former wartime leaders during the
election period (pre-election period, election day, and immediate post-election period).
3. RESEARCH DESIGN
To statistical test the hypothesis suggested in the previous chapter, this study compiles a dataset predominantly deriving from the unpublished dataset by Kreutz et al. (2019a) on arrests of former wartime leaders with some additional coding to complement the dataset and updated NELDA dataset by Hyde & Marinov (2012) on national elections. The unit of analysis is country-year, covering the period between 1989 to 2017. The chapter ends with discussions on control variables and the choice of a statistical test.
3.1 Dependent Variable: Arrests of Former Wartime Leaders
The unpublished dataset by Kreutz et al. (2019a) for the project currently named “Arrested Development: The Dangerous Mix of Patronage and Justice in Post-conflict Countries”
includes arrests of former wartime leaders made between the period of 1989 to 2017 in post- conflict countries. The dataset is built on conflict termination dataset by Kreutz (2010) on state- based conflicts between 1946 and 2015 to identify post-conflict cases (Kreutz et al. 2019b).
The point of departure from the UCDP’s data is Kreutz et al. (2019 a) focus on conflict dyads between a government and a non-state armed group (Ibid). Hence, the Kreutz et al. (2019b) dataset includes arrests of former wartime leaders that were carried out after the conflict dyad was terminated by a victory (for either side), or a peace agreement, or a ceasefire agreement, or low activity, or actor ceasing to exist (Ibid). In cases where conflict-dyads re-erupts, arrests are not coded until the conflict terminates again (Ibid).
The former wartime leaders, as mentioned in the introduction section, includes (a) political
leaders of rebel groups or militias even without any military background, (b) “political leaders
of governments, i.e. heads of state, prime ministers, members of cabinet or legislature, or senior
government officials in state agencies during the conflict,” and (c) “military leaders including
senior commanders, commanders, senior officers, generals, chiefs of staff of security forces or
non-state armed actors (rebel groups, militias, paramilitaries)” (Ibid: 2). It is important to note
that some sources fail to distinguish between senior/low-ranking commanders, in which case
the “commander” term was considered enough to be coded as the arrests (Ibid). Similarly, the
dataset includes arrests for any alleged crimes related to the armed conflict rather than “war
crimes” defined in international law because what encompasses war crimes on itself is
politicized and debatable (Ibid). Hence, the alleged crimes that the former wartime leaders are
arrested for can range from war crimes, crimes against humanity, serious violations of
international humanitarian law, unlawful killings, physical violence to murder, looting, use of
child soldiers, corruption and so on during the conflict period (see Kreutz et al. 2019a Dataset).
The following shortcomings of the data on arrests of former wartime leaders may be noted:
1. The data on arrests of former wartime leaders are collected mainly from news media, NGOs, and other official documents. Some conflicts like the Balkan conflict have gotten more international media attention than other countries. Therefore, an unequal amount of information is available for different post-conflict countries; consequently, some countries have more arrests and some arrests had to be excluded due to lack of complete information.
2. In Bosnia and Herzegovina, arrests are present almost every year in the post-conflict period which might create bias in the statistical findings. Statistical models by controlling for Bosnia and Herzegovina is also presented to ensure the robustness of the findings.
Overall, in the compiled dataset used for this study, the dependent variable (arrests of former wartime leaders) is coded as a binary variable where 1 is coded if the arrests take place and 0 if no arrests. The arrests that were made outside the country were omitted because the theory is limited to explaining the arrests over which incumbents of the relevant post-conflict country have authority. Only exceptions are an arrest in 1993 coded for Israel that took place in Palestinian area and four arrests coded for Indonesia between 1992 and 1997 that took place in East Timor. The arrest in the Palestinian area is included because it was made by Israeli authority (Middle Eastern Journal 1993), and also the indicting actor is Israel itself ( see Kreutz et al. 2019a). Similarly, arrests in East Timor are included because the Indonesian soldiers left the territory only in 1999 (UCDP (c). Moreover, Indonesia was the indicting actor for all those arrests (see Kreutz et al. 2019a). Thus, it can be argued that Indonesia had a strong influence and authority over those arrests compared to other excluded arrests made outside of the country. Considering that Indonesian authority in East Timor could have declined after 1999, another arrest made in 2000 in East Timor was consciously excluded. Similarly, an arrest made in Georgia was excluded because it was made in Abkhazia. Abkhazia is a self-proclaimed independent state, and UCDP (b) maintains the Georgia-Abkhazia conflict as a “frozen”
conflict since the tension between them remains intact despite low to no battle death over the
years. In addition, the arrest was made by Abkhazia security forces and even the indicting actor
is Abkhazia itself (see Kreutz et al. 2019b). This study also excludes arrests made by
peacekeeping forces even if it was carried out within the country; hence, 27 arrests (including
surrender to North Atlantic Treaty Organization (NATO) and another to International Criminal
Tribunal for the former Yugoslavia (ICTY) office in Bosnia and Herzegovina) made by
peacekeeping forces in Bosnia and Herzegovina, 1 in Liberia, and 2 in the Central African
Republic) are excluded. Likewise, the surrender cases where those who were indicted by ICTY
voluntarily flew themselves to the Hague were excluded from the dataset. Regarding arrests in
post-conflict countries in general, possible arrests that are unclear due to key missing information; for example, lack clarity in whether they were arrested for crimes committed during the conflict or post-conflict period were omitted.
3.2 Independent Variable: Elections
The data for elections (independent variable) is drawn from the National Elections across Democracy and Autocracy (NELDA) Version 5 dataset by Hyde & Marinov (2012). The Version 5 dataset includes elections held for executive positions, and/or legislative positions and/or for Constituent Assembly at a national level from 1945 to 2015 (Hyde & Mirnov 2019).
Voters must directly be electing the person/s for the respective national level position to be coded as an election (Ibid). While the original dataset treats presidential, legislative, and constituent assembly elections as separate observations even if it occurs on the same day (Ibid), this study uses a collapsed version that treats them as combined observations based on the year they take place. Similarly, the dataset provides information only until 2015; therefore, missing values of elections are added for the years 2016 and 2017 from the International Institute for Democracy and Electoral Assistance (International IDEA), an intergovernmental organization.
The shortcoming of using election data, in general, is that it does not capture the competitiveness and other ostensible flaws of those elections (Hyde and Marinov 2019). The shortcoming is managed to some extent by controlling for the level of democracy which is discussed later.
It is also important to recall the hypothesis which states that arrests are more likely to be carried out against the former wartime leaders during the election period (pre-election period, election day, and immediate post-election period). As mentioned previously, various events, such as voters and/or political parties’ registration can mark the initiation of a pre-election period and different countries vary in its longevity (Höglund 2009). Similarly, the immediate post-election period can be marked as a period after an election and before the inauguration of a newly/reelected body (Ibid). Since different countries have varying time spam concerning pre and immediate post-election periods, the election year is code as 1 and other non-election years as 0.
By coding election years as the binary variable and having a unit of analysis as country-
year, it invites the problem of some arrests taking place after the immediate post-election period
in the given election year. To ensure the robustness of the study and avoid falsely accepting
the hypothesis, an analysis by omitting the arrests that took place after the immediate post-
election period is also included. Such arrests are omitted by tallying arrest dates with election
dates and inauguration dates of a newly/reelected executive or legislative body for all the relevant observations where arrests and elections are in the same year, available in Annex III.
The inauguration dates are compiled mainly from Inter-Parliamentary Union (IPU) election archives and others from different books.
Rather than only providing findings by omitting the arrests that took place after the inauguration period in the given election year, I have included both findings because the election year might be a unique event that makes arrests of former wartime leaders more likely, hence opening a path for future research (explored in next chapter). In other words, there has not been any quantitative study to explain the systematic variation in the arrests of former wartime leaders, so election years might offer unique insight and open a gateway for future exploration, alongside testing relevant past theories. On the other hand, there is a probability that some elections take place at the beginning of the year due to which the effect of a pre- election period on the dependent variable may not be captured. Therefore, I also include a model after re-coding the election variable by assigning 1 value to all the election year and a year before the election, and 0 otherwise.
Lastly, in this compiled dataset, the post-conflict period starts from the conflict termination year. The Kreutz et al. (2019a) dataset does not code for the arrests that happened before the exact conflict termination date (year/month/day), so arrests that took place before that date are automatically excluded. There is also a likelihood that some elections take place in the same year as conflict termination year, so elections that concluded month/s before the exact conflict termination date as provided by Kreutz et al. (2019a) are excluded from the study. For example, if the conflict has terminated in 1991/2/15, but the election took place in 1991/1/23 then those elections are excluded. Whereas, elections that take place after the exact conflict termination dates are included.
3.3 Control Variables: Stability, Relative Power, International Support and Pressure, and Democracy
Other variables likely affect the possible relationship between arrests and elections. Based on the past literature, the control variables identified are stability, the relative power of new and old elites, international support, and pressure with the addition of a democracy variable.
First, as discussed in the earlier chapter, there exists the issue of stability vs. justice as seeking
justice, especially during the fragile transitional period can make conflict parties withdraw their
commitment to peace which raises the possibility of reigniting a conflict (Nagy 2008). This
could mean that the violent past would be addressed only when the transitional period passes
or the “stability” is achieved in the post-conflict period. To control this effect, I code stability as a count variable where the conflict termination year is counted as 0 and years after that as 1, 2, 3, and so on. If another conflict-dyad is active during that period, it is coded as 0.
Second, the previous literature suggests the relative power of old and new elites determined by (a) types of political transition and (b) types of state structure following a conflict affects how much justice takes place in the post-conflict period. The political transition theory expects justice to be more likely when a past government is completely overthrown and least likely when a transition is brought through agreements (Huntington 1991 cited in Grodsky 2008).
Similarly, the state structure theory expects minimum or no justice when old elites remain influential in the post-conflict period (Grodsky 2008; Smith 2012). Since the type of political transition can also have a significant effect on state structure in the post-conflict period (Subotić 2009; Smith 2012), the possible effect of both political transition and state structure is controlled through the type of conflict termination. For this, I rely on Kreutz (2010) conflict termination dataset. When the conflict ends with a victory (either for rebels or government) it is coded as 1 and if the conflict ends with Peace agreement/Ceasefire /Low activity/Actor ceases to exist it is coded as 0. Even government victory is coded as 1 because there are fewer chances of having a bidding agreement to defer from arrests, unlike those conflicts that end with peace agreements. In cases where conflict ended in diverse ways due to the presence of different conflict dyads, it is coded as 1 when at least one of the dyads ended in either government victory or rebel victory. It is necessary to code it as 1 because based on theory, the arrest would still take place from the dyad which ended with a victory.
Third, Subotić (2009) argues that international actors can induce state compliance to justice
by “offering” both carrots and sticks in some cases. To capture the effect of international
pressure or support, I account for the amount of foreign aid and the presence of peacekeeping
in post-conflict countries. For foreign aid, I rely on “Net bilateral aid flows from DAC donors
(current US$)” derived from the World Bank data from 1960 to 2018. The Development
Assistance Committee (DAC) was established in 1960 and consists of 30 member countries,
like the United States (U.S.), United Kingdom, France among others (The Organisation for
Economic Co-operation and Development (OECD). When examining the impact of aid on
democratic outcomes, Zürcher et al. (2013) argue that aid donors that influence the timing and
terms of aids are more important than the aid itself. In a similar manner, DAC aid captures
certain aid donors that can arguably be considered powerful and influential countries. Since
almost all countries throughout different years have gotten DAC aid (see the Word Bank
dataset), so the aid is taken as a net amount (current US$).
Regarding peacekeeping, Berdal & Ucko (2015) point out that peacekeeping heavily relies on funding and resources from the powerful member states. Such dependence, in many instances, translates into powerful countries determining the size and cost of a mission (Fortna 2008). If this is true then inferences can be drawn that international attention or pressure can be observed through peacekeeping to some extent, especially through the cost and size of peacekeeping forces. To control for this effect, I rely on Bara & Hultman's (2020) global peacekeeping dataset from 1993 and 2016. I include the total number of uniformed peacekeeping personnel (police, observers, and troops) on a yearly basis which Bara &
Hultman's (2020) dataset primarily derives from Stockholm International Peace Research Institute (2017). The total number of uniformed peacekeeping personnel for the United Nations (UN) and Regional Organization (RO) are included separately to capture their individual effect on the dependent variable. Since the peacekeeping dataset starts from 1993 compared to this study which starts from 1989, the missing values are likely to affect the results.
Lastly, I control for the level of democracy in the country for the given year to address the shortcoming of using elections in general as it does not account for the competitiveness of the elections and its other shortcomings (Hyde & Mirnov 2019). In addition, transitional justice is often interlinked to democratic values (Oko 2003; Nuzov 2014), so the arrest of former wartime leaders would become more likely as a post-conflict country achieves higher democracy scores.
I control for this effect by using the Freedom House (2020) dataset from 1973 to 2020. The Freedom House data divides countries as Free, Partially Free, and Not Free based on civil liberties (CL) and political rights (PL), measured on the scale of 1 to 10.
6The political right aspect of the dataset captures the ability of people to freely take part in the political process, such as being able to freely vote for distinct alternatives in legitimate elections (Freedom House 2018 cited in Coppedge et al. 2020). Similarly, civil Liberties capture the freedom of expression and belief, rule of law, organizational and associational rights, and personal autonomy (Freedom House 2018, and Teorell et al. 2018 cited in Coppedge et al. 2020). The ability of the Freedom House dataset to account for whether the regular elections are relatively free among other aspects of democracy justifies the use of this dataset. The democracy status of the country is coded as a binary variable where Free is coded as 1 and Partially Free and Not Free as 0.
6
