Institutional Legacies, Decision Frames and Political Violence in Rwanda and Burundi

Stacey Mitchell

• Publisher: Routledge
• Year: 2018
• Print ISBN: 9780367590048
• eBook ISBN: 9780429019081
• Edition: 1
• Topic: Politics & International Relations
• Subtopic: African Politics
• Index: Politics & International Relations

Rwanda and Burundi are strikingly similar countries that underwent democratization in the early 1990s. In both, resistance to democratic reforms led to coups d'état and presidential assassinations. A conundrum arises in terms of what transpires next. In Rwanda, total genocide was perpetrated by extremist Hutu actors, including government officials, upon the country's Tutsi and politically moderate Hutu populations. In Burundi the coup d'état failed and instead ushered in a lengthy period of civil war. This divergence in outcome is puzzling given the similarity of these two countries, and it is not adequately explained by studies that address collective violence in each.

This book utilizes an integrative approach that facilitates the formation of an explanation that more fully accounts for variation in the type of collective violence that occurred in Rwanda and Burundi. Showing that political actors – during periods of major institutional change – do not all respond to or perceive reform in the exact same manner or in a necessarily rational manner, this book makes an important contribution to the literature on ethnic conflict, collective violence and democratization in Africa.

1 Extant explanations of collective violence in Rwanda and Burundi

This chapter examines three specific groups of theories commonly used to explain the origins of collective violence in Rwanda and Burundi. I categorize each set of explanations by their focus of analysis. Macro-studies explain the violence as a consequence of larger-scale factors, such as economic decline. Meso-level explanations concentrate on the role of institutions as a precipitating force in the outbreak of inter-ethnic bloodshed. Micro-level theories apply a rationalist logic to explain the behavior of individual actors in their decision to commit or take part in genocidal violence.

All of these studies encounter the same two problems: (1) they cannot adequately explain the source of the variation in the responses of political leaders to similar conditions; and (2) they fail to establish a clear connection between context and individual behavior. Highlighting these problems demonstrates the need for an integrative theory that includes a well-defined decision-making model that fully incorporates context and which allows for the possibility that actors use multiple decision rules in their choice-making processes under conditions of risk.

Macro-level explanations of genocide and collective violence

Most of the macro-level studies I consider specifically address the genocide in Rwanda. Others, including those by Krain (1997) and Harff (2003), examine collective violence in Rwanda and Burundi from a broader comparative perspective. All of these studies assume that certain combinations of larger-scale phenomena are responsible for the numerous instances of group violence that have plagued both countries in recent decades. I am keeping this particular analysis relatively brief for the simple reason that I examine many of these same studies in greater detail in Chapter 4 in conjunction with a test of the first alternative hypothesis.

While macro-level explanations of the violence in Rwanda and/or Burundi differ to the extent that they emphasize the causal significance of certain contextual circumstances over others, they all rely on the general assumption that pluralized societies are especially prone to ethnic violence given the right mix of circumstances.

In the genocide literature more broadly, a “pluralized society” is one in which inequalities – political, economic – are superimposed across a society (Kuper 1981, pp. 57–58).¹ The concept of the pluralized society fits well with the concept of “structural violence”; a phenomenon which is manifest in “a deep and widening inequality of life chances; corruption, arbitrariness, and impunity; the permanence of social and economic exclusion; lack of access to information, education, health, and minimal basic needs; and an authoritarian and condescending state and aid system” (Uvin 1998, p. 107). I differ from the macro-level genocide studies in that I do not consider the pluralized society to be a form of structure per se. Instead, I regard it as the institutional foundation for genocide. The pluralized society taps into the impact that informal and formal traditional institutions within a country have on collective violence and group identities.

Another commonality shared by all of the macro-level analyses of the violence in Rwanda and Burundi is the causal significance they attach to a lengthy history of communal violence. In particular, inter-ethnic conflict is assumed to have played a major role in cementing group identities and social cleavages within each country. Concerning the role of group identity specifically, scholars of Rwanda and Burundi strongly contend that the “Hutu” and “Tutsi” identities are by no means static, but have been shaped by historical circumstances and experiences.²

These experiences, moreover, do not just apply to those that occur in one’s home country, but also refer to those taking place in the neighboring country. For example, events such as the 1972 genocide of Hutu counter-elites in Burundi and the October 1993 assassination of the Burundian president, Melchior Ndadaye, of the Front pour la Démocratie au Burundi (“FRODEBU”) party, left an especially negative imprint on Hutu–Tutsi relations in Rwanda – both in a real and abstract sense. The same can be said about the impact that the violence associated with the 1959 Hutu Revolution in Rwanda had on Tutsi elites in Burundi.³

Adding further insult to injury for both countries has been the “culture of impunity” associated with all of these episodes of violence. As this pertains to Rwanda, a number of scholars partially attribute the commission of the 1994 genocide to the lack of accountability the government and its agents faced for the numerous human rights atrocities they perpetrated during the civil war and the democratization process (African Rights 1995; Prunier 1995). These studies suggest that this culture of impunity created an environment that only encouraged further violence and mayhem.⁴

It is important to remember that, in one way or another, these studies assume that violence in Rwanda and/or Burundi was a rational strategy enacted on the part of political elites to retain or acquire more political power. There are other studies which offer the additional assumption that the violence that occurred in both countries during the early 1990s was a rational strategy intended by its perpetrators as a means of population control. The focus of these “resource scarcity” studies is specifically the relationship between population density, available resources and the frequency of collective violence and bloodshed.

Uvin (1998) makes a useful distinction between arguments that take a “hard” and “soft” Malthusian approach to addressing the causal role of resource scarcity in Rwanda. Hard Malthusian arguments suggest that collective violence is the direct result of the disproportion between a country’s population growth and its natural resources (pp. 180–181). The theoretical perspective most commonly applied to the case of Rwanda is a “soft” Malthusian approach, which suggests that certain elements in society mediated the impact that environmental factors had on group relations (pp. 182–183).⁵ In other words, how much of a threat resource scarcity posed to Rwanda was a function of the particular political and social institutions in place within Rwanda. The less symmetrical the distribution of power, the worse the outcome.

Closely related to resource scarcity arguments are those which consider economic crises to have been a variable of particular importance in the violence that erupted in Rwanda. All of the qualitative studies written about the genocide, including those by Des Forges (1999), Prunier (1995) and African Rights (1995), agree that a severe bout of economic distress – brought on by the collapse of the international coffee market, the imposition of a structural adjustment program by the International Monetary Fund (“IMF”), and civil war – increased the likelihood that genocide would occur by exacerbating inter-group hostilities and by decreasing the security of ruling elites.⁶ Under these conditions, economic tensions and frustrations were manipulated by incumbent elites to gain political advantage.

The most obvious problem that all of the macro-level genocide studies encounter is that they cannot explain why similar conditions in Burundi did not lead to a similar outcome. The data presented in Chapter 4 will demonstrate that circumstances in Burundi in the late 1980s/early 1990s were similar – if not worse in many ways – to those in Rwanda. Moreover, as a “repeat offender” of mass violence, the likelihood that a second genocide would occur in Burundi was quite significant.⁷ Unfortunately, few of the above-mentioned studies offer any sort of systematic comparison between the two countries. Consequently, the conclusions they reach concerning the relationship between macro-level factors and the genocide in Rwanda are tenuous at best.

The few scholars who do address this issue of variation attribute the difference in the scale of violence that occurred in both countries to the number of “accelerating events” which immediately preceded these conflicts. The works by Harff (1998, pp. 70–78) and Davies et al. (1998) suggest that the occurrence of a greater number of accelerating events in Rwanda, in conjunction with various other pre-conditional circumstances – both background and intervening – resulted in total genocide. In Burundi, a lesser number of very similar events led to a completely different outcome: civil war. Yet, neither study can fully explain why a certain frequency of events prompted decision-makers to select a policy of genocide in one case but not the other.

The source of this explanatory deficiency is an attribute that is common to all of the macro-level studies discussed above, as well as to macro-level explanations of genocide more generally. It is the problem of structure and agency, in which human behavior, preferences and interests are largely pre-determined by macro-level phenomena.

In his analysis of macro- and micro-level explanations of collective revolutionary behavior, Berejikian (1992) suggests that, in order to overcome a structure–agency dilemma, the former explanations indirectly rely on an assumption that actors “employ a decision rule akin to a microeconomic rational maximizer and are thus driven by the desire to secure maximum benefit” (p. 649). I suggest that macro-level studies written about the occurrence of collective violence in Rwanda and/or Burundi depend on a similar assumption.

As this applies to Rwanda more specifically, the studies by Prunier (1995), Des Forges (1999) and others assume that extremist Hutus rationally opted to perpetrate genocide because they considered multiparty rule with the Tutsi to be the costlier alternative. Unfortunately, none of these studies provides any evidence that demonstrates political elites in Rwanda conducted any type of cost-benefit analysis in their decision-making processes. Because the focus of these studies remains at the systems-level, the preferences of actors are instead derived from their policy choices. Consequently, it is difficult to fully discern how and why factors such as resource scarcity and/or economic crises affected political elites in Rwanda and Burundi in the particular ways that they did.

Moreover, the notion that political elites employed a “single and mutual decision rule” suggests that political elites in Rwanda and Burundi should have responded to similar crises in a very similar way. That this was not the case raises some doubts as to the applicability of rational choice explanations to both examples. Additionally, macro-level studies that rely on the auxiliary assumption that the Hutu extremists in Rwanda were behaving as rational maximizers cannot explain why other more cost-effective policy options were completely discarded. Put another way, these studies cannot adequately explain why or how Hutu politicians and party members came to perceive genocide as the optimal strategy relative to all others.

Meso-level explanations: democratization and collective violence

There is some degree of overlap between studies that attempt to explain the occurrence of genocide in Rwanda from a macro perspective and those which take into account the role of democratization as a factor in explaining inter-group violence. Much of this overlap is due to shared assumptions regarding the causal influence of contextual elements on the occurrence of group violence as well as the existence of rational political actors.

Where the two groups diverge is with respect to the role of institutions in communal violence. Macro-level analyses about the violence in Rwanda and/or Burundi depict the process of democratization as a trigger event that unleashed group violence in countries already heavily burdened by a number of other serious problems. Democratization studies consider institutions and the process of institutional change to be primary causal factors in the occurrence of ethnic violence in Rwanda, Burundi, and elsewhere in the developing world.

In this study I focus specifically on those democratization studies that address the prevalence of what are termed “semi-democracies.” Also referred to as “virtual democracies,” they are political systems which are neither fully democratic nor fully authoritarian and which possess several distinct features, including:

• a formal basis in citizen rule, but with key [decision-making] (especially economic) insulated from popular involvement;

• manipulation of democratic transitions by political incumbents, including the use of violence and electoral fraud, to [re-legiti...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright page
• Table of Contents
• List of figures
• List of tables
• List of abbreviations
• Introduction: the puzzle
• 1 Extant explanations of collective violence in Rwanda and Burundi
• 2 An alternative explanation of collective violence
• 3 Methodology and research design
• 4 A story of twins
• 5 Social structures and collective violence in Rwanda and Burundi
• 6 Traditional political institutions and collective violence in Rwanda and Burundi
• 7 Reference points, decision frames and the choice to perpetrate genocide
• 8 Conclusions
• Afterword
• Index