Presentation on theme: "Peace through power-sharing? A critical assessment Gerald Schneider University of Konstanz Ukraine, 2007 Afghanistan 2009."— Presentation transcript:
Peace through power-sharing? A critical assessment Gerald Schneider University of Konstanz Ukraine, 2007 Afghanistan 2009
The Dangers of Diversity "The redistribution of national wealth in post- soviet Ukraine has been accompanied by a considerable polarisation of living standards," observes Iryna Prybytkova, a professor at the Ukrainian Academy of Science's Institute of Sociology. Guardian, May 13, 2008 ´…the one-sided nature of the conflict [between Israel and the Palestinians] and the emotions it arouses beyond its boundaries have helped Bin Laden achieve the fourth and most important of his objectives: polarisation” (The Guardian)
But what sort of diversity matters? Polarization vs. fractionalization Source: Montalvo/ Reynal-Querol 2005. Journal of Develop- ment Economics
Ethnic polarization vs. ethnic fractionalization
Power sharing as a solution? “rules that, in addition to defining how decisions will be made by groups within the polity, allocate decision-making rights, including access to state resources, among collectivities competing for power” (Hartzell and Hoddie 2003, 320).
Two papers, two authors: - Power Sharing Provisions and Long-Term Success of Mediation in Internal Conflicts (APSA 2011) - The Illusion of “Peace through Power Sharing”: Constitutional Choice in the Shadow of Civil War (ISA 2012) One is optimistic, the other pessimistic: how can this be ? Paulina Pospieszna Uncle G. University of University of Mannheim Konstanz The Optimist The Pessimist
The foundation of two papers Source: Schneider/ Wiesehomeier JPR 2008 Instituto de Ciências Sociais, Lisbon Nina W.
Power sharing prevents conflicts, depending on the form of diversity Mediated power-sharing might make a positive difference Yet, power sharing is more likely after inconclusive armed conflicts How to reconcile the findings
Can institutions pacify diverse socities? Lipset-Rokkan thesis: Societies with many social cleavages opt for multi-party systems. Power sharing approach (Consociationalism) (A. Lijphart) a) Socially heterogeneous countries should opt for consoctiationalism: 1. Sharing of executive power 2. Federalism 3. Proportional Representation 4. Minority vetoes b) Consociationalist democracies are more peaceful.
Yet, the evidence is inconclusive 1.a. Federalism makes countries more peaceful. (e.g. Cohen 1997. Lake and Rotchild 1999). However, the party system might matter (Brancati 2006) 1.b. Federalism makes countries less peaceful. (e.g. Mansfield and Snyder 2002). 1.c. Selection bias is a problem (Christin and Hug 2011) 2. Systems with proportional representation (Reynal-Querol 2002) or inclusive political systems are more peaceful (Reynal-Querol 2002). See, however, Horowitz vs. Fraenkel/Grofman 3. The number of effective parties exerts a curvilinear influence (Wilkinson 2005) 4. Are presidential systems more peaceful (Fearon and Laitin 2003)?
Our approach Institutional settings lend different opportunities to minorities to participate in power. Hence, the interaction of institutions and diversity is a key factor for understanding civil war. Autocracies: Polarization increases the risk of conflict, dominance decreases it, and fractionalization is not systematically linked to it. Democracies: Polarization decreases the risk of conflict, dominance and fractionalization increase it. Features of democracies: Power-sharing institutions in combination with dominance and fractionalization decrease the risk of conflict, while majoritarian institutions reduce it in combination with polarization
Research Design I: The risk of civil war world wide, 1950-2000 II: The risk of civil war in democracies, 1950- 2000 Method: Conditional risk set model, onsets as ordered multiple events Diversity data: Fearon 2003, Journal of Economic Growth Focus on ethnic divisions, polarization measured as a dummy variable based on Reynal-Querol (2002) Institutional variables based on Golder (2004)
MEASURES OF DIVERSITY: o FRACTIONALIZATION o POLARIZATION --the relative size of the relevant group, be they ethnically religiously or linguistically defined Dominance 1 if the population share of one group is between 55 and 90% Research design (CONT.) Hirschman- Herfindahl (Hirschman 1964) Reynal-Querol (2002)
Polarization increases the risk of conflict in autocracies, but not in democracies In democracies, fractionalization and,to some extent, dominance are associated with internal war.
The marginal effect of fractionalization on the hazard of civil war onset with changes in the squared effective number of parties
The marginal effect of fractionalization and dominance on the hazard of civil war onset with changes in the average magnitude of districts (proportional voting)
How does power-sharing work after an armed conflict? - Why do some successfully mediated peace settlements fail to bring durable peace? - Do power sharing provisions in the peace agreements make a difference?
Power sharing and the duration of peace -political power sharing not necessarily a help in maintaining post-conflict stability or democracy (Adekanye 1998; Binningsbø and Kendra 2009; Norris 2008; Vandeginste 2009) -power sharing concept extension to include four dimensions of power-sharing: political, territorial, military and economic (Hartzell and Hoddie 2003; DeRouen et al. 2009; Jarstad and Nilsson 2008; Mattes and Mukherjee 2006; Savun 2009; Svensson 2009; Walter 2002)
Research Design Data: The UCDP Peace Agreement Dataset (1989 to 2005) (Harbom, Högbladh and Wallensteen 2006). Civil War Mediation (CWM) dataset (1946-2004) (DeRouen, Bercovitch and Pospieszna 2011) 108 mediated peace agreements between 1989 and 2004 Dependent Variable and Estimation Method Mediation Failure: 1 if mediated efforts led to the conclusion of peace agreements but violence with the same parties restarted within 5 years, 0 if mediated efforts that led to the conclusion of peace agreements and violence with the same parties did not restart within 5 years (success of long-term impact of mediation Probit models.
Research Design Any Power sharing if the agreement specifies at least one type of power-sharing dimension (political or military or territorial). Power Sharing Comprehensiveif at least two power-sharing dimensions are specified in the agreement Political Power Sharingif the agreement included at least one of the following: - integration of rebels into government or civil service; - rebel integration into the interim government; - power-sharing in new government. Territorial Power Sharing if a peace agreement mentioned at least one of the following: Autonomy (granted the disputed region autonomy), Federation (provided for a federal state solution); Independence; Referendum (the agreement provided for the holding of a referendum on the future status of the disputed region; Local Power Sharing (granted the disputed region power-sharing in the local government; Local self-government (granted the disputed region local governance); or Demarcation of the border. Military Power Sharingif the agreement provided for the creation of a new national army or the integration of rebels into the army Independent Variables
Research Design Other Factors: -Mediator’s Profile: Powerful Mediator, IGOs -Directive Mediation Strategy -War Duration -Peacekeeping Operations (PKO) -Commission or committee to oversee implementation -Peace Spoilers -Territorial Wars -Peace Agreements Ended -Peace Agreement End*Territorial Wars
Empirical Analysis Probit Analysis (Dependent Variable: Mediation Failure) Cell entries report coefficients. For the interaction-terms, cell entries report marginal effects. Robust standard errors in parentheses; ∗∗∗ p <0.01, ∗∗ p < 0.05, ∗ p < 0.10
Yet, are power-sharing societies a random sample of all post-conflict societies? - Which post-conflict countries opt for power sharing institutions? - Do these institutions affect the chance of medium-term peace? War outcome Democracy? If yes: Power sharing? Effectiveness of power sharing?
H1: (Adoption of power sharing) A: If a civil war had ended with a victory of one side or inconclusively, a country is less likely to experience free and fair elections and to maintain or introduce power-sharing institutions in the first period of post-conflict reconstruction. B: Ethnic polarization increases the chance that a country relies on power sharing institutions, while ethnic fractionalization decreases this chance. H2: (Effectiveness of power sharing) Power sharing in post-conflict societies increases the chance that large groups are adopted into the political system and reduces the risk that a civil war recurs. Hypotheses
551 conflict terminations, 1946-2009 CONFLICT TERMINATION TYPE o VICTORY o INCONCLUSIVE (ceasefire agreement with conflict regulation; ceasefire agreement; low activity; joining alliance; other) CIVIL WAR RECURRENCE Recur5_10: violence recurrence between 5 and 10 years from the end of the war. Dataset: UCDP Conflict Termination dataset v. 2010-1, 1946 - 2009 (Kreutz 2010). Research design
FREE AND FAIR ELECTIONS = 1 if a country i had between t-5 and t legislative elections that were fair and free Frequency of Elections per Post-Conflict Country in the Five Years After the End of a Conflict, 1946-2009 Datasets/Sources: Quality of Election Data (QED) by Judith Kelley, OSCE Election reports, Electoral Observation Mission Reports by the Organization of American States; Country Reports on Human Rights Practices submitted to US Congress. Research design (CONT.)
POWER SHARING INSTITUTIONS o PROPORTIONAL REPRESENTATION Datasets/Sources: the Democratic Electoral Systems Around the World dataset assembled by Matt Golder (2005) for the period until 2000 and the Database of Political Institutions (DPI) provided by the Worldbank (Beck 2001) for data after 2000 o FEDERALISM Datasets/Sources: AUTON variable from Database of Political Institutions (DPI). In case of missing data we consulted GOVSTRUCT and REGSTRUCT variables from dataset by Reagan et al. (2009) : The Institutions and Election Project (IAEP) Conflict Termination (t-5) Research design (CONT.) 12345 Period t
Categories of Dependent variable and their frequencies, per Post-conflict termination, 1946-2009 Source: UCDP Conflict Termination dataset v.2010-1, 1946 – 2009; Quality of Election Data (QED) by Judith Kelley, OSCE Election reports, Electoral Observation Mission Reports by the Organization of American States; Country Reports on Human Rights Practices submitted to US Congress; Democratic Electoral Systems Around the World by Matt Golder, Database of Political Institutions (DPI) by the Worldbank, The Institutions and Election Project (IAEP).
Results: Adoption of power sharing Results from Multinomial Probit Regression Models to Predict Adoption of Power-sharing Institutions in Post-conflict Countries, 1946-2009 Cell entries report coefficients. Robust standard errors in parentheses. Std. Err. adjusted for 345 clusters in dyadic group identity; ∗∗∗ p <0.01, ∗∗ p < 0.05, ∗ p < 0.10
Effectiveness of power sharing Dependent Variable: Conflict Recurrence 5-10 years After Cell entries report coefficients. Robust standard errors in parentheses. Std. Err. adjusted for 288 clusters in dyadid; ∗∗∗ p <0.01, ∗∗ p < 0.05, ∗ p < 0.1
A closer look: Majoritarian democracies without conflict recurrence
Conclusion Power-sharing helps maintain peace, but its effectiveness depends on the predominant form of diversity Mediated peace agreements that stipulate the sharing of power between the former adversaries pacify states for at least five years after the first end of conflict However, two institutions closely associated with power sharing have no effect on the risk of a new civil war. The chance of power sharing depends on the outcome of a conflict.
Thank you! How to get in touch with uncle G.: Gerald.Schneider@uni-konstanz.de