Partition, Power Sharing, and the Legitimacy of Post-Conflict Arrangements

Abstract: 

Scholars have produced an impressive compendium of literature pertaining to power sharing and partition as methods of consolidating peace durability. Empirical tests primarily focus on which method is associated with durable peace, however, and stop short of how the method achieves durable peace or why it fails to do so. My research seeks to advance the existing literature by theorizing that any arrangement for the consolidation of durable peace must meet two basic requisites—legitimacy and enforceability—and by exploring the impact partition and power sharing have on the legitimacy of post-conflict arrangements. Using riots and anti-government demonstrations as a proxy for legitimacy, I run a logistic regression analysis to test my hypotheses. I found that, ultimately, while power sharing is significant in increasing the legitimacy of post-conflict arrangements following an ideological conflict, under no circumstances tested is partition significantly associated with legitimacy.

Table of Contents: 

    Introduction

    Although living in the throes of the Cold War may impress one with a sense that the world has become peaceful, this notion could hardly be further from the truth. Instead, there has merely been a transition so that civil conflict has become the dominant form of armed conflict around the globe (Gleditsch et al.., 2002). In light of this information, scholars and decision makers alike have devised numerous methods of peace building, and yet these approaches frequently fail to prevent a relapse into conflict. For example, in arguing that the consolidation of peace is the most pressing issue for peace-builders, Gates & Strom (2007) point out that the initial signing of peace treaties in Rwanda, Angola, and Liberia did nothing to stop the wanton massacre of hundreds of thousands of people in each of those countries. Thus, it is necessary to look beyond stop-gap institutions and the ignition of a transition from conflict to peace. Scholars and decision makers must also dedicate attention to the mechanisms used for the establishment of long-term arrangements conducive to the maintenance of a durable peace.

    This study seeks to advance the existing literature on conflict management by theorizing that any arrangement for the consolidation of durable peace must meet two basic requisites—legitimacy and enforceability—and exploring the impact partition and power sharing have on the legitimacy of post-conflict arrangements. Thus, the chief concern of my research is to ask: what is the impact of the two major methods of conflict management–power sharing and partition–on the legitimacy of post-conflict arrangements in a country following either an ethnic or ideological civil war? Although this leaves room for future examination of enforceability, my research should provide real-world policy implications that cannot be drawn from the existing literature.

    Factors highlighted by previous work focused on identifying the determinants of whether peace proves durable often concentrate on immutable characteristics such as the wealth of the given country as measured by annual gross domestic production per capita (Sambanis 2000; Chapman & Roeder 2007), annual per capita income growth (Sambanis & Schulhofer-Wohl 2009), or overall gross domestic production (Gold 2010). Legitimacy, on the other hand, can be controlled and affected by decision makers if they are informed as to the implications that each method of conflict management has on the provisions of the post-conflict arrangement. In addition to the real world implications, my study also maintains a degree of academic relevance. Although scholars have produced an impressive compendium of literature pertaining to power sharing and partition as methods of consolidating peace durability, empirical tests have largely focused on which of the respective methods is most associated with durable peace and have stopped short of asking how these methods achieve that end or why they fail to do so. Thus, my theory advances the existing literature by examining the process through which power sharing or partition impact the durability of peace in post-conflict societies.

    Using the presence or absence of riots and anti-government demonstrations as a proxy for legitimacy, I run two logistic regression analysis models. Ultimately, the results are more or less consistent between the models but I only find support for the third hypothesis. This suggests that partition is not an effective method of conflict management following ethnic or ideological civil wars. Power sharing, on the other hand, is associated with increased legitimacy in the arrangements of a country following an ideological civil war. It should, then, provide an effective method of consolidating peace durability in such circumstances. On the other hand, neither partition nor power sharing is found to be significantly related to the legitimacy of arrangements following an ethnic civil war. This suggests that decision makers looking to consolidate peace durability in the throes of an ethnic conflict should primarily focus on enforcement mechanisms.

    The first section of this study analyzes the existing literature, identifying the primary methods of conflict management endorsed by scholars in the field of peace studies and civil conflict management. This section also identifies a number of shortcomings in the current compendium of scholarly work pertaining to how these methods relate to the durability of peace in post-conflict societies.  The second section details my theory of how the respective methods of conflict management relate to the legitimacy of post-conflict arrangements following ethnic and ideological civil wars. In this section I also identify my hypotheses. The third section, Research Design, identifies my variables as well as my means of both quantifying and operationalizing these variables. This section also details the methodology used in testing my hypotheses. The Analysis section provides a table and description of my results as well as an interpretation of these results. Finally, I finish up with a discussion of my research and its implications before concluding.

    Literature Review

    Overview

    Historically, literature pertaining to conflict management has been dominated by a debate emanating from competing methods of political power sharing. More recently, however, scholars have also begun to consider military, territorial, and economic power sharing arrangements as well as institutional alternatives to power sharing—power dividing and partition. The former alternative to power sharing was proposed by Roeder & Rothchild (2005) and has received limited scholarly attention. The latter alternative has roused a considerable degree more attention but has not been empirically analyzed with any consistent results. The existing literature has also failed to produce any semblance of consensus among scholars as to the question of which institution (power sharing, power dividing, or partition) provides the best method of initiating a transition from conflict to peace or consolidating peace durability in a post-conflict society. Even among proponents of a particular institution, scholars are divided in a lively debate as to the details of how their arrangements should be implemented and for what purpose.

    In addition, while scholars have produced an impressive compendium of literature pertaining to power sharing and partition as methods of consolidating peace durability, empirical tests have largely focused on which of the respective methods is most associated with durable peace and have stopped short of asking how these methods achieve that end or why they fail to do so. Furthermore, factors highlighted by past work interested in identifying the determinants of whether peace proves durable tend to be immutable characteristics such as prewar democracy scores (Hoddie & Hartzel 2001). There is, then, a need to investigate factors contributing to peace durability which can be affected by institutional decision making; moreover, it is neccesary to see how the primary methods of conflict management impact these mutable characteristics.

    Alternative Functions of Power Sharing

    Jarstad (2006) explains that the term power sharing has been employed in two separate strands of literature but that it serves a separate function in each. One literary thread pertains to democratic theory and the other pertains to conflict management. The former generally focuses on political methods of power sharing that offer rival groups a role in the political process and primarily serve to produce functional democracy in divided societies. In the latter, scholars treat power sharing as a method of ending conflict and maintaining peace but focus on territorial and military power as well as political power sharing.

    In the context of democratic theory, the power sharing literature is dominated by a line of argumentation emanating from two alternative approaches to joint governance. The first approach is frequently described as consociational democracy (Lijphart 1969, 1977, 1979, 2004). This method of joint governance is characterized by (A) segmental autonomy and (B) representation within a ‘grand coalition’ for all major factions of a divided society. Delegates within the ‘grand coalition,’ or committee of group-leaders, are selected through a proportional electoral system to represent their respective groups. Delegates in the grand coalition, armed with the power of mutual veto, defend the interests of their groups in a negotiated series of concessions and compromises. According to the theory, this process should produce a stable democracy under terms acceptable to all the major factions in a society. Additionally, consociationalism should have the added benefit of promoting a vertical divide between rival groups that allows each group to compete with all of the others (Lijphart 1969, 1977, 1979, 2004).

    The major alternative to Lijphart’s consociationalism originates with Horowitz (1985) and was labeled centripetalism in Reilly (2001). This approach attempts to eliminate dividing identities within the population of a society by promoting greater integration through a majoritarian electoral system. On the other hand, centripetalism limits the number of political parties and produces a horizontal division such that one group in a society maintains clear dominance over all of the others (Horowitz 1985). Ultimately, however, Lijphart’s theory of consociationalism and Horowitz’s theory labeled centripetalism bothfocus on electoral systems and both seek to produce stable democracy in divided societies through joint governance.

    In the context of conflict management, however, the power sharing literature encompasses institutions that endow former antagonists with political power but also includes institutions pertaining to economic, military, or territorial power. Although much of the literature provides extensive coverage of political power sharing, new debates are arising as to whether this is the most effective method of conflict resolution or peace maintenance. In fact, some advocates of power sharing as a method of maintaining peace durability have argued that political power sharing pacts make no significant difference and only agreements with military and territorial power sharing provisions help to consolidate peace (Jarstad & Nilsson 2008). On the other hand, some scholars have come to the opposite conclusion (Mattes & Savun 2009; Derrouen, Lea, & Wallensteen 2009).

    There is a great deal of disagreement among scholars of conflict management concerning the function of power sharing. For example, Roeder & Rothchild (2005) hold that while power-sharing may be an effective tool for initiating a transition from ongoing conflict to peace, it hampers the process of consolidating a durable peace in post conflict societies. On the other hand, many scholars examining power sharing find it to be a very useful tool for maintaining peace durability and preventing war recurrence in post conflict societies (Hartzell 2009; Hartzell & Hoddie 2003, 2007; Jarstad & Nilsson 2008). Gates & Strom (2007) find that power-sharing may be an effective method of preventing conflict from occurring to start with, and criticizes scholars such as Hartzell & Hoddie (2007) for not having taken this into consideration.

    In summation, scholars of conflict management disagree on whether or not power sharing is effective for (A) preventing conflict, (B) resolving conflict, and/or (C) consolidating peace stability. They generally agree, however, that power sharing institutions encompass not only political power sharing but also military, economic, and territorial power sharing.

    Other Methods of Conflict Management

    In addition to power sharing, two other methods of conflict management have recently warranted an increasing degree of scholarly attention. In the following segment we will discuss the former alternative to power sharing—partition. Following this, we dedicate a section to the latter alternative to power sharing–power dividing.

    Partition and Conflict Management

    Although the attention of the literature on conflict management has generally been monopolized by power sharing, partition has recently become a popular alternative. Scholars have alternatively found partition effective for initiating an end to ongoing conflict (Kaufmann 1996, 1998), consolidating peace durability (Chapman & Roeder 2007), and preventing war recurrence in post conflict societies (Johnson 2010). Literature on partition is also divided on the topic of de facto separation. While Kaufmann (1996, 1998) and Johnson (2010) maintain that de facto separation involving the geographical division of demographic groups into defensible enclaves supersedes sovereignty, Chapman & Roeder (2007) find that de facto separation is of secondary importance as compared to de jure partition dividing a single administrative units into two or more sovereign entities with international recognition. However, while details of pro-partition arguments have varied, all of these scholars agree that partition is a method of conflict management specifically designed for ethnic and nationalist conflicts. For example, Kaufmann (1996, 1998) provides the strongest theoretical argument for partition when he posits the concept of an ethnic security dilemma emerging from adversarial ethnic identities hardened by conflict.

    Some scholars have resisted the notion that partition is an effective method for peace management of any form. Jenne (2009), for example, examined the de facto partition of Bosnia and Kosovo to argue that partition is an ineffective tool during the peace-consolidation stage of conflict management. Sambanis (2000) provided more generalizable results with the first empirical test of the usefulness of partition as a method of conflict management. Although the study found that partition was not useful in either the initiation phase or the consolidation phase of conflict management, it has not proved to be a robust finding. Chapman & Roeder (2007) reanalyzed data from Sambanis (2000) and arrived at opposite conclusions.

    Although Sambanis and Schulhofer-Wohl (2009) offer some explanation for the contrary conclusions, they do not provide an effective response to the primary pro-partition arguments posited by Kaufmann (1996). While Kaufmann (1996) held sovereignty to be secondary to de facto separation, Sambanis and Schulhofer-Wohl (2009) did not take the degree of population transfer into account with their preferred definition of partition. As theory proposed in Kaufman (1996, 1998) advocates demographic separation, it is neccesary to measure the degree of population transfer in order to test the theory. Perhaps a better test of Kaufmann’s argument is provided by Johnson (2010) as it contains an index for measuring the degree of population transfer. However, Johnson’s result contradicted both the argument of Kaufmann (1996, 1998) and Sambanis (2000) as it held that while partition does help to prevent war recurrence in post conflict societies, it does not help to end an ongoing conflict. Ultimately, however, even within pro-partition circles, scholars remain divided on the role that partition should play as a method of conflict management and more thorough examination is necessary to augment existing literature.

    Power Dividing and Conflict Management

    In contrast to the preceding two methods of conflict management, power dividing is a fairly new idea originating with Roeder & Rothchild (2005) that has received comparatively limited scholarly attention. Power dividing is characterized by checks and balances, multiple majorities, and strong civil liberties. While power sharing focuses on creating organizations to guide a divided society into some semblance of unity, power dividing aims to empower people with universal rights under the assumption that the mutual effort to protect shared liberties will serve as a natural catalyst for unity. Power dividing, as the theory goes, prevents the division of civil society along ethnic lines by empowering the people to establish a means of government protecting mutually held civil rights and liberties.

    However, Roeder & Rothchild (2005) recommend power dividing only for the consolidation of peace durability and the prevention of war recurrence in post-conflict society. They maintain that power sharing is a more pragmatic tool during the initiation phase of the peace process (Roeder & Rothchild 2005). However, Roeder argues that by endowing citizens with universal, individual rights and the freedom of association as opposed to group rights, power dividing ensures that individuals of diverse backgrounds will cooperate in defending constitutional order against challenges that might unravel the system of civil rights from which they all benefit. Additionally, because power dividing swings some control over controversial issues from government jurisdiction to that of civil society by allowing the people to decide how interests are separated, it should produce a resolution more acceptable to the people (Roeder 2010).

    Nevertheless, very little has been done to empirically test the theory. One attempt was made by Gold (2011) but this ultimately concluded on a note of ambivalence with no clear conclusion. This is perhaps to be expected. The cases cited to provide examples of power dividing or to evidence the effectiveness of power dividing as a deterrent for renewed conflict by Roeder & Rothchild (2005) are either instances in which power dividing appears to be present but was not implemented as a method of conflict management (such as the United States) or cases generally described as power-sharing (Switzerland, Belgium, and India). Furthermore, the preponderance of western countries in the pool of examples raises questions as to whether the institution of power dividing is applicable outside of a western context. In any case, the small number of cases in which power dividing was used as a method of conflict management makes it impossible for any reliable, empirical analysis of its effectiveness for either initiating a transition from conflict to peace or consolidating peace durability in a post-conflict society.

    Synopsis of the Status Quo

    The literature pertaining to conflict management has largely been dominated by political power sharing. More recently, however, proponents of power sharing have begun to examine arrangements containing provisions for military, economic, and territorial power sharing in addition to or in lieu of political power sharing. Other scholars have also begun to endorse partition and power dividing as alternative methods of conflict management. Empirical examinations of these approaches, however, have been unsatisfactory. In the case of partition, findings have demonstrated themselves to be consistently contradictory and lacking in robustness. In the case of power dividing, empirical examination of the potential of the institution as a method of conflict management has been made impossible by a general lack of cases available for examination. Ultimately, the literature has failed to produce any definite answers and scholars remain divided in a lively debate. While some agreement can be found by examining factors highlighted in previous literature as determining whether peace proves durable, these factors tend to be immutable characteristics which do not provide real world policy implications. Finally, empirical examinations have dominantly concentrated on which of the respective methods is most associated with durable peace and have stopped short of asking how these methods achieve that end or why they fail to do so.

    Theory

    Theoretically, any arrangement for the consolidation of durable peace must meet two basic requisites regardless of the conflict management institutional decisions. First, the arrangement must be practicably enforceable. Whenever there is no practicable means of enforcing an arrangement, this creates a credible commitment problem similar to that proposed in Fearon (1995) and Walter (1999, 2002) because parties have no assurance that their rivals will uphold their end of the bargain. A security dilemma between former belligerents may arise and, consequently, the parties may violate the provisions of the arrangement to avoid being on the losing side of a zero sum outcome. Second, the arrangement must be perceived as legitimate by the affected parties. To elaborate on the concept of legitimacy, a legitimate arrangement is both just and binding. Whenever the affected parties do not feel bound by the provisions of an arrangement, they cannot be expected to adhere to those provisions. In the face of unjust provisions, the party may feel that it is being oppressed by those measures and therefore feel obliged to resist the arrangement. Thus, my research seeks to advance the existing literature by determining how the primary methods of conflict management (power sharing and partition) impact the legitimacy of post-conflict arrangements. Although this leaves room for future examination of enforceability, my research should also provide useable policy guidance. While the factors highlighted by previous literature as determining whether or not peace proves durable have largely been immutable characteristics such as the prewar democracy score of the country (Hoddie & Hartzel 2001), decision makers can influence the legitimacy of post-conflict arrangements if they are informed as to the implications that each method of conflict management has on the provisions of the post-conflict arrangement.

    In order to qualify as legitimate, an arrangement must be viewed by the affected parties as both just and binding. A party perceives an arrangement as ‘just’ whenever the provisions of that arrangement are fair or favorable to the interests of the party because such provisions do not place a disproportionate burden or share of the risk on that party. For example, a rebel group would likely view unilateral disarmament as unjust because it requires that the rebel group sacrifice its only form of security and consequently, weaken its position in relation to the government. The provisions of the arrangement must also be conducive to the realization of the primary goals of the party because such provisions allow them to achieve their goals without the violence and resource depletion that accompanies renewed conflict; however, provisions that serve as obstacles hindering the ability of the party to achieve its goals by domestic means require that they either abandon their goals or return to conflict. For example, if the party in question wishes to obtain independent statehood, that party will view a partition providing it with autonomy over a defensible, sovereign enclave outside of the administrative control of the rump state as just because this helps that party to reach its goal of independent nationhood. That same party, however, will likely perceive a power sharing arrangement requiring it to concede the right to self-determination and cooperate in the governance of a shared state with its rivals as unjust because the arrangement serves as an obstacle hindering the ability of the party to achieve independence.

    On the other hand, a party perceives an arrangement as ‘binding,’ whenever that party feels obliged to uphold the provisions of the arrangement either to avoid negative outcomes or promote positive outcomes. If the party from our previous example feels that without conceding to the power sharing arrangement, it will suffer military defeat or that by conceding to the power sharing arrangement, it will eventually move closer to achieving its ultimate goal of partition the party will feel bound by the arrangement. They feel this way because respecting the provisions of the arrangement helps them to achieve a positive outcome (eventual independence, peace, etc) while avoiding a negative outcome (military defeat, resource depleting conflict, etc). As an additional requirement, each party must acknowledge the leader agreeing to the arrangement on their behalf as that representative of the party before the arrangement can be seen as binding. If the party does not acknowledge the leader agreeing to the arrangement as its representative, the party will not feel that it has consented to the arrangement in the first place. As a result, the party may feel oppressed and resort to political violence rather than uphold the arrangement because they feel that the arrangement was imposed upon it.

    In summation, legitimacy is a characteristic held by an arrangement whenever three requisites are satisfied. First, the affected parties must acknowledge the leaders agreeing to the arrangement on their behalf as their representatives. Second, they must feel obliged to uphold the provisions of the arrangement. Finally, they must see the provisions of the arrangement as conducive to the eventual realization of their primary goals. However, as many goals held by parties to a conflict are dependent upon the type of conflict, it should be possible to find general provision types which address common goals held by all parties to other conflicts of that same category. For example, if we are focusing on an ethnic conflict, our method of managing the conflict should contain provisions resolving the ethnic security dilemma. On the other hand, if we are dealing with an ideological conflict, our method of conflict management should contain provisions which help former belligerents to the conflict protect and pursue their political interests. This being said, the nature of both primary approaches to the consolidation of durable peace gives each method a unique implication for legitimacy; therefore, it is necessary to examine the implications of partition and power sharing on legitimacy individually as opposed to merely looking at conflict management as a whole.

    Partition

    I define partition as an institution that seeks to achieve either de jure separation by dividing a single administrative entity into multiple units and/or de facto separation by dividing demographic groups into separate, geographical enclaves. With this definition, it is possible to argue that partition mitigates post-conflict legitimacy in ideological conflicts where ethnic groups see themselves as members of the same community but disagree on how to govern that community. The goals of the parties to such a conflict are to pursue a system for their shared community that protects and pursues their political interests in that community. Dividing these parties into separate enclaves would then hinder their ability to achieve their goal by breaking down the original community and isolating them from the other factions of the original community.  On the other hand, my theory suggests that partition should increase the legitimacy of the post-conflict institutional arrangement in an identity conflict between antagonistic ethnic groups that do not perceive themselves as different factions of the same community because it does not hamper any goal of unified governance and it helps to resolve ethnic security dilemmas (Kaufmann 1996, 1998; Johnson 2010).

    Thus, partition should positively impact the legitimacy of arrangements following a conflict of identity by taking steps towards the resolution of the security dilemma (either by separating rival groups into separate enclaves or by providing them with enclaves in which to separate themselves) without hindering goals of unified governance. On the other hand, partition negatively impacts the legitimacy of arrangements following an ideological conflict by hindering goals of unified governance.

    H1: Partition positively impacts the legitimacy of post-conflict arrangements following conflicts of ethnic identity.
    H2: Partition negatively impacts legitimacy of post-conflict arrangements following a conflict of ideology.

    Power Sharing

    As an alternative to partition, power sharing–defined as an institution of dividing political, economic, military, and/or territorial power among rival groups within a single administrative unit–should increase legitimacy in ideological conflicts. Like partition, power sharing accomplishes this by providing each group with a means of pursuing positive outcomes while avoiding zero-sum and negative outcomes. In the case of power sharing, the positive outcome is the realization of the goals of the party. Power sharing provides each party with a mechanism to pursue its goals by arming them with political and economic power. Former rebel or minority groups will likely see this as a step towards the realization of their goals because such groups are unlikely to have had significant economic or political power. Thus, through power sharing, these groups obtain an influence over government and economic and political power, which equips them with tools that they were formerly lacking. On the other hand, majority and government groups benefit from the arrangement as it prevents the resource depletion that accompanies renewed conflict and allows them to focus on obtaining their other goals. At the same time, power sharing provisions allotting control over military and territorial resources help to resolve security dilemmas and credible commitment problems by giving each group a tangible tool for the protection of its interests should its rivals fail to uphold their end of the bargain (Jarstad & Nilsson 2008). In this way, groups have tangible mechanisms to assure them that they can avoid a negative outcome. Furthermore, the provisions of a power sharing arrangement allow the respective parties to accomplish all of this within a single, unified society. Thus, power sharing has an advantage over partition in cases of ideological conflict because it does not demand that parties isolate themselves from other factions of their community.

    Therefore, my theory suggests that power sharing increases legitimacy in ideological conflicts by arming rival groups with a mechanism (political/economic power) with which to diplomatically pursue their political interests while simultaneously arming them with security producing tools (military/territorial control) to reassure them against the occurrence of negative or zero-sum outcomes. However, because it accomplishes both of these goals within a single unified society, it decreases legitimacy in identity conflicts where groups are not dedicated to existing as a single community.

    H3: Power sharing positively impacts the legitimacy of post-conflict arrangements following conflicts of ideology.
    H4: Power sharing negatively impacts the legitimacy of post-conflict arrangements following conflicts of ethnic identity.

    Research Design

    This study explores the impact of power sharing and partition on the legitimacy of post-conflict arrangements following ethnic and ideological civil wars. Instances of civil war were identified using the peace duration dataset provided by Mason et al. (2011). However, as this examination focuses on post-conflict arrangements, some period of peace duration following the civil war exceeding at least one year is required before a post-conflict period constitutes an observation in the analysis. For example, although Afghanistan experienced numerous civil wars within the temporal parameters of the study, it does not constitute an observation in the analysis because there has been no significant ‘post-conflict’ period following any of these civil wars. Thus, the list of observed cases in this study encompasses all countries to have experienced civil war followed by at least one year of peace between 1946 and 2002.

    Dependent Variable

    The dependent variable for this study is the legitimacy of institutional arrangements in post-conflict countries. It had been hoped that the information gathered by various barometers and the World Values Survey could be used as an indicator of legitimacy, particularly in their questions regarding satisfaction with the current government. However, as surveyors are unwilling to enter into post-conflict countries where violence is likely to occur, these sources do not provide consistent data on many of the observations in this study. As a result, it is neccesary for any systematic analysis of legitimacy in post-conflict countries to find an alternative to the information derived from these surveys. After an extensive examination of all the available data, it was determined that the best method of resolving this issue would be to create a proxy using indicators of the presence or absence of legitimacy. As riots and anti-government demonstrations suggest popular dissatisfaction with the current institutional arrangements as well as an inability of the administration to maintain order, both events provide a reasonable proxy for the presence or absence of legitimacy in a given country during a given time period. In this case specifically, instances of anti-government demonstrations and riots in post-conflict societies during a period of peace exceeding at least one year were identified using data derived from the Cross National Time series data provided by Banks (2007). Thus, in order to determine whether legitimacy is present or absent in a given case, the event-count data of Banks (2007) was used to construct a dichotomous variable such that legitimacy is equal to one if one or more riots occurred within the first five years following a civil conflict. If no riots occurred, legitimacy is equal to zero. A similar variable was created using the occurrence of anti-government demonstrations as derived from the same dataset. The threshold was set at one because well over half of all observations in the Banks dataset show that no riots (or antigovernment demonstrations) occurred in that country-year. In either case, one represents the absence of legitimacy in post-conflict arrangements and zero represents the presence of legitimacy in post-conflict arrangements.

    Independent Variables

    In this study, partition is a dichotomous variable equal to one if partition did occur and equal to zero if partition did not occur. The variable was constructed using the dataset of Tir (2005). The variable for power sharing is also dichotomous. Zero signifies that power sharing was not used as a method of conflict management in the given country during a given year. The variable for power sharing was constructed using the data collected by Hartzell & Hoddie (2007) (pp. 47-49). Thus, like the dependent variables, the independent variables in this study are dichotomous. In this case, a zero signifies that the method of conflict management (partition or power sharing) did not occur in the respective country during that year. A one signifies that the method of conflict management did occur in the given case. For example, following the 1968 civil war in the Dominican Republic, power sharing provisions were used as post-conflict institutional arrangements. Thus, for the Dominican Republic during 1996-1970, power sharing is categorized as a one and partition is categorized as a zero.

    Furthermore, using the dataset of Mason et al. (2011), civil wars are categorized as either ethnic or ideological using another dichotomous variable. This allows for the creation of two interaction variables. The former examines the interaction between power sharing and whether the civil conflict was ethnic. The latter variable examines the interaction between partition and whether the civil conflict was ethnic. This allows for the analysis to observe the impact of partition and power sharing on the legitimacy of post-conflict arrangements following ethnic and ideological conflicts separately.

    Finally, the data provided by Hartzel & Hoddie (2007) was also used to produce a third category. The third category, ‘other,’ was created to include military victories and negotiated truces without power sharing provisions. However, this variable was only generated to provide a baseline category and was not tested in the analysis. Power dividing was also excluded from the analysis because of the limited number of cases in which it appeared as a method of conflict management.

    Control Variables

    I use several control variables to include gross domestic production, conflict duration, conflict intensity, the presence or absence of a peacekeepers mission within the country, and the degree of democracy present in a country prior to its civil war.  In the prior literature, there has been some inconsistency in the method of measuring these variables. For example, Sambanis (2000) measured conflict intensity as a continuous variable based on conflict deaths. Jarstad and Nilsson (2008), on the other hand, used a dichotomous variable based on whether or not the conflict reached the level of civil war (1,000 conflict deaths). However, despite definitional disagreements, these factors have maintained a consistent presence in conflict management literature and were identified by Hoddie and Hartzell (2001) as significant to the durability of peace settlements. The inclusion of these variables as controls in my study, then, seems consistent with the prior literature of the field. As an additional control, I include a variable for ethnic-linguistic fractionalization derived from Alesina (2003).

    I control for the Gross Domestic Product (GDP) of the country by using a measure of the GDP of the country in constant US dollars. The data for gross domestic production was obtained from the World Bank Development Indicators (www.data.worldbank.org). Finally, I logged the variable to control for possible outliers.

    Conflict duration and conflict intensity are both derived from the dataset used in Mason et al.. (2011). Both of these variables are continuous measures of the respective variable. Conflict duration is measured by the number of years the previous conflict spanned. Conflict intensity, on the other hand, is measured in battle-related deaths. In addition, the variable is logged to account for possible outliers.

    I also control for the presence of a peacekeeping operation. The dichotomous variable for the presence of peacekeepers was identified using the Third Party Peacekeeping Mission 1946-2006 dataset by Mullenbach & Dixon (2006). A one for the peacekeeping variable represents the presence of a peacekeeping mission in the given country during a given time period. A zero represents the absence of a peacekeeping mission in a given country during a given time period.

    Consistent with previous work (Chapman & Roeder 2007; Jarstad & Nilsson 2008; Sambanis 2000), the degree of prewar democracy present in a country was derived from the latest polity data series. In this case specifically, the prewar democracy of an observed case is measured by its Polity2 variable in the Polity IV Political Regime Characteristics and Transitions Dataset (Marshall and Jaggers 2000). The polity score of a country is measured on an ordinal scale ranging from -10 for the most autocratic to +10 for the most democratic.

    Methodology

    Given that the dependent variable for this study, legitimacy, was measured in the form of two variables–one based on the presence or absence of anti-government demonstrations and the other based on the presence or absence of riots–it is neccesary to analyze the data in two separate models. However, as the dependent variable is also dichotomous, I was able to run a simple logistic regression analysis in both models. The results using a binomial logistic model are provided in Table 1 on page 28. In binomial logit models, we can observe whether the coefficients are statistically significant or not. However, the actual coefficients produced by the logit regression are limited in their interpretation. When using a binomial logit model, one cannot interpret the coefficients directly in terms of a change in the dependent variable, y, for a unit change in the independent variable, x (Long 1997). In Table 2 on page 29, I present the odds ratio of the model, which allows for easier interpretation of the model as well as the proposed relationships in the hypotheses. In this manner, the coefficients are easier to interpret but the statistical significance remains the same from the previous model.

    Analysis

    In order to test my respective hypotheses, I ran two logistical regression analysis models. Each tested the impact of both power sharing and partition on the legitimacy of institutional arrangements following ideological civil wars; furthermore, both contained interaction variables to test the impact of power sharing and partition on the legitimacy of institutional arrangements following ethnic civil wars. The results of each model can be found in both Table 1 and Table 2Table 1 contains the standard error and coefficient of each respective variable. Table 2 contains the odds ratio of each respective variable.

    Model One

    In model one, I ran a logistic regression analysis using the presence or absence of riots as a proxy for the legitimacy of institutional arrangements during a post-conflict period. Thus, the dependent variable in model one is a dichotomous variable equal to one if riots occurred and equal to zero if no riots occurred. As demonstrated in both Table 1 and Table 2, model one finds that the use of power sharing as a method of conflict management does not significantly impact the legitimacy of institutional arrangements following ethnic civil wars but that it does significantly increase the legitimacy of institutional-arrangements following ideological civil wars.  Specifically, the model suggests that the use of a power sharing institutional arrangement after an ideological civil war will decrease the probability that riots will occur during the post-civil conflict period by a factor of .53. This finding implies that the establishment of a power sharing institutional arrangement after an ideological civil war will be viewed by the population as legitimate and consequently, they will be less likely to riot against the government. Although these results are consistent with my third hypothesis, the results are actually contrary to my fourth hypothesis because the use of power sharing institutional arrangements after an ethnic civil war was not statistically significant. Similarly, partition is not significantly related to the legitimacy of institutional arrangements following either an ethnic or ideological conflict; therefore, neither my first nor my second hypothesis is supported by model one. Of the control variables, only a country’s pre-war democracy score significantly impacts the legitimacy of institutional arrangements following a civil conflict and actually operates in an unexpected manner by increasing the probability of riots.

    The findings of model one suggest that while power sharing may increase the legitimacy of institutional arrangements following an ideological conflict, decision makers may want to consider alternative methods of conflict management for countries in the throes of an ethnic conflict. Furthermore, the results of model one suggest that partition may not be an effective method for increasing the legitimacy of institutional arrangements following an ethnic or ideological civil conflict; therefore, decision makers seeking to consolidate peace in a post-conflict country using partition–or power sharing, if the conflict was ethnic in nature–should primarily focus their efforts on establishing effective enforcement mechanisms.

    Model Two

    In model two, I ran a logistic regression analysis using the presence or absence of anti-government demonstrations as a proxy for the legitimacy of institutional arrangements in a post-conflict period. Once again, this means that the dependent variable is dichotomous. One represents the presence of demonstrations and zero represents the absence of demonstrations. Similar to model one, the results for model two are present in both Table 1 and Table 2. In this case, however, neither of the main independent variables is found to significantly impact the legitimacy of institutional arrangements following either ethnic or ideological conflicts. Of the control variables, only gross domestic production, ethno-linguistic fractionalization, and the presence of a peacekeeping mission have a statistical significant effect on the legitimacy of institutional arrangements following a civil war.

    The results of model two suggest that whenever confronting a country in the throes of either ethnic or ideological conflict, decision makers should primarily focus their efforts on the establishment of enforcement mechanisms. Furthermore, as neither partition nor power sharing is significantly related to the legitimacy of institutional arrangements during the consolidation phase of the peace process, it may be neccesary for scholars to produce a new method of conflict management in order to maintain post-conflict institutional arrangements with legitimacy. In this case, the power dividing approach of Roeder and Rothchild (2005) may provide a promising alternative. The reliability of the implications of the model is, however, limited by the lack of  reliable data for the quantification of legitimacy.

    Synopsis of Findings

    Although there was some variance between the two models, neither found partition to be an effective method of increasing the legitimacy of institutional arrangements following ethnic or ideological conflicts. Similarly, neither model found a significant relationship between power sharing and the legitimacy of institutional arrangements following an ethnic conflict and only the former found a significant relationship between legitimacy and power sharing following an ideological conflict. Thus, while the analysis contained within this study provides some evidence for my third hypothesis, further research is neccesary to confirm the relationship and no support was found for any of the other hypotheses.

    Limitations

    Although no support was found for three of my hypotheses, this study is severely limited by the existence of data used to quantify legitimacy. Using riots and anti-government demonstrations as a proxy was a compromise due to incomplete datasets among barometers and the World Values Survey. Thus, future research using a more direct measure of legitimacy may find a more significant relationship between the partition and power sharing and the legitimacy of post conflict arrangements following both ethnic and ideological conflicts.

    Conclusion

    My research set out to determine what impact power sharing and partition have on the legitimacy of post-conflict arrangements in a country following either an ideological or ethnic civil war. Although more research is necessary to test my theory in full, this study helps to advance the existing literature by matriculating from past examinations which have merely looked at which of the respective methods of conflict management are most associated with durable peace to studying how these methods achieve that end or why they fail to do so. Furthermore, by examining legitimacy as a factor in determining whether conflict will recur, this project provides some policy implications that cannot be found in past literature focusing on less mutable characteristics. The lack of findings to evidence a significant relationship between partition and the legitimacy of post conflict arrangements in cases of either ethnic or ideological civil war suggest that it may not be an effective tool for the consolidation of peace durability. When partition is used as a method of conflict management, decision makers should primarily focus their efforts on improving enforcement mechanisms. Similarly, as power sharing shows no significant implications for the legitimacy of arrangements following ethnic conflict, decision makers attempting to use power sharing as a method of conflict management following an ethnic conflict should invest more in the establishment of effective enforcement mechanisms. However, since power sharing demonstrates some promise for promoting legitimacy following ideological conflicts, further research should be conducted to explore in greater detail the relationship between power sharing of the institutional arrangements during a post-conflict period.

    However, this study leaves room for future research. Alternative measures of legitimacy should be tested and research should be conducted examining the implications that partition and power sharing institutions have on the enforceability of post-conflict arrangements. Future studies should also expand upon the theory by analyzing specific provision types most conducive to improved measures of legitimacy and enforceability. For example, one might ask whether power sharing institutions should include provisions primarily focused with military as opposed to political power sharing. To make this determination, a study must look at not only the presence or absence of power sharing measures or partition but the nature of those institutions. Thus, although this project provides a theoretical stepping stone for the advancement of existing literature, further work is neccesary to further explore this avenue.

    Bibliography

    • Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain Wacziarg. 2003. “Fractionalization.” Journal of Economic Growth 8:155-194.
    • Banks, Arthur S. 2007. “CNTSDATA” (Data file) Cross National Time Series Data Archive.
    • Chapman, Thomas & Philip Roeder. 2007. “Partition as a Solution to Wars of Nationalism: The Importance of Institutions.” The American Political Science Review 101 (Nov) pp. 677-691.
    • Derrouen, Karl, Jenna Lea, & Peter Wallensteen. 2009. “The Duration of Civil War Peace Agreements.” Conflict Management and Peace Science 26 (Sep) 367-387.
    • Gates, Scott and Kaare Strøm. 2007. “Power-sharing, Agency and Civil Conflict.” Draft.
    • Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg & Håvard Strand. 2002. “Armed Conflict 1946–2001: A New Dataset.” Journal of Peace Research 39 (Sep) 615–637.
    • Gold, Valentin. 2011. “Power-Sharing or Power-Dividing? Walking out of the Maze.” Paper prepared for presentation at the ECPR Joint Sessions of Workshops. April 12-17, 2011.
    • Hartzell, Caroline & Matthew Hoddie. 2003. “Institutionalizing Peace: Power Sharing and Post-Civil War Conflict Management.” American Journal of Political Science 47 (April) pp. 318-332.
    • __________ . 2007. Crafting Peace: Power-Sharing Institutions and the Negotiated Settlement of Civil War. University of Pennsylvania Press.
    • Hartzell, Caroline. 2009. “Settling Civil Wars: Armed Opponents’ Fates and the Duration of the Peace.” Conflict Management and Peace Science 26 (September) pp. 347-366.
    • Horowitz, Donald. 1985. Ethnic Groups in Conflict. University of California Press.
    • Jarstad, Anna & Desiree Nilsson. 2008. “From Words to Deeds: The Implementation of Power-Sharing Pacts in Peace Accords.” Conflict Management and Peace Science 25 (Aug) pp. 206-223.
    • Jarstad, Anna. 2006. “The Logic of Power Sharing after Civil War.” Ph.D. diss. Uppsala University.
    • Jenne, Erin. 2010. “The Paradox of Ethnic Partition: Lessons from de facto Partition in Bosnia and Kosovo.” Regional and Federal Studies 19 (May) pp. 273-289.
    • Johnson, C. R. 2010. “Partition as a Solution to Ethnic Civil War: Statehood, Demography, and the Role of Post-War Balance of Power for Peace.” (Doctoral dissertation). Retrieved from http://drum.lib.umd.edu/handle/1903/10074.
    • Kaufmann, Chaim. 1996. “Possible and Impossible Solutions to Ethnic Civil Wars.” International Security 20 (Spring) pp. 136-175.
    • __________. 1998. “When All Else Fails: Ethnic Population Transfers and Partitions in the Twentieth Century” International Security 23 (Autumn) pp. 120-156.
    • Lijphart, Arend. 1969. “Consociational Democracy.” World Politics 21 (Jan) 207-225.
    • __________.  1977. Democracy in Plural Societies: A Comparative Exploration. Binghamton: Yale University Press.
    • __________. 1979. “Consociation and Federation: Conceptual and Empirical Links.” Canadian Journal of Political Science 12 (Sep) pp. 499-515.
    • __________. 2004. “Constitutional Design for Divided Societies.” Journal of Democracy 15 (April) pp. 96-109.
    • Long, Scott J. 1997. Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks, CA: Sage.
    • Marshall, Monty G. and Keith Jaggers. 2000. “Polity IV Project: Political Regime Characteristics and Transitions, 1800-2000”. Retrieved at http://www.cidcm.umd.edu/inscr/polity/.
    • Mason, T. David, Mehmet Gurses, Patrick T. Brandt, and Jason Michael Quinn. 2011. “When Civil Wars Recur: Conditions for Durable Peace after Civil Wars.” International Studies Perspectives 12: 171-189.
    • Mattes, Michaela & Burcu Savun. 2009. “Fostering Peace after Civil War: Commitment Problems and Agreement Design.” International Studies Quarterly 53 (Sep) 737-759.
    • Reilly, Ben. 2001. Democracy in Divided Societies: Electoral Engineering for Conflict Management. Cambridge: Cambridge University Press.
    • Roeder, Philip G. 2005. “Power Dividing as an Alternative to Ethnic Power Sharing.” In Sustainable Peace: Power and Democracy After Civil Wars, eds. Roeder & Rothchild. New York: Cornell University Press, pp. 29-50.
    • __________. 2011. “Power Dividing.” In S. Wolff and C. Yakinthou (Eds) Conflict Management in Divided Societies: Theories and Practice. London: Routledge.
    • Roeder, Philip & Donald Rothchild. 2005. “Power Sharing as an Impediment to Peace and Democracy.” In Sustainable Peace: Power and Democracy After Civil Wars, eds. Roeder & Rothchild. New York: Cornell University Press, pp. 29-50.
    • Sambanis, Nicholas & Jonah Schulhofer-Wohl. 2009. “What’s in a Line? Is Partition a Solution to Civil War?” International Security 35 (Fall) pp. 82-118.
    • Sambanis, Nicholas. 2000. “Partition as a Solution to Ethnic War: An Empirical Critique of the Theoretical Literature.” World Politics 52 (Jul) pp. 437-483.
    • Tir, Jaroslav. 2005. “Dividing Countries to Promote Peace: Prospects for Long-Term Success of Partitions.” Journal of Peace Research 42(5): 545-562.
    • World Development Indicators, The World Bank. www.data.worldbank.org.

    Table 1: Logit Models

    Variables

    Model 1
    (Riots)

    Model 2
    (Demonstrations)

    Partition

    -.2458433
    (.3285129)

    -.279376
    (.3322582)

    Power Sharing

    -.6413976*
    (. 3601905)

    -.2227964
    (.3362909)

    Conflict Type
    (Ethnic or Ideological)

    -.3524169
    (.4366638)

    -.3440288
    (.4518329)

    Conflict Intensity

    .0745491
    (.0622269)

    .0120025
    (.064535)

    Conflict Duration

    -.0255395
    (.0209414)

    -.0230821
    (.0207174)

    Interaction Variable, Power Sharing and Ethnic Conflict

    -.1469779
    (1.199397)

    -.1733959
    (1.27456)

    Interaction Variable, Partition and Ethnic Conflict

    .4772114
    (.802171)

    -1.67633
    (1.167532)

    Prewar Democracy

    .0520139**
    (.0228616)

    .03259
    (.0230382)

    GDP

    (.112348)
    .0876588

    .2909668***
    (.0899066)

    Presence of peacekeeping
    mission

    .1364663
    (.3768723)

    1.008484**
    (.3524285)

    Ethno-Linguistic
    Fractionalization

    -.6374887
    (. 4952636)

    -1.183319
    (.5067892)**

    *p<.10 
    **P<.05
    ***p<.001

    N =  379
    Pseudo R2 = 0.0341

    N =  379
    Pseudo R2 = 0.0685

     

    Table 2: Logit Models with Odds Ratios

    Variables

    Model 1
    (Riots)

    Model 2
    (Demonstrations)

    Partition

    Odds Ratio
    .7820448

    Odds Ratio
    .768107

    Power Sharing

    .526556*

    .8019339

    Conflict Type
    (Ethnic or Ideological)

    .702987

    .7983122

    Conflict Intensity

    1.077398

    .004966

    Conflict Duration

    .9747838

    .9793025

    Interaction Variable: Power Sharing and Ethnic Conflict

    .863313

    1.005458

    Interaction Variable: Partition and Ethnic Conflict

    1.611574

    .1426139

    Prewar Democracy

    1.05339**

    1.001247

    GDP

    1.118902

    1.332172***

    Presence of peacekeeping
    mission

    1.146216

    2.389042**

    Ethno-Linguistic
    Fractionalization

    .5286183

    .2911483**

    *p<.10
    **P<.05
    ***p<.001

    N =  379
    Pseudo R2 = 0.0341

    N =  379
    Pseudo R2 = 0.0621