Democratization and the Mischief of Faction 9781626377363

Citation preview

Democratization and the Mischief of Faction

DEMOCRATIZATION AND THE MISCHIEF OF FACTION

Benjamin R. Cole

b o u l d e r l o n d o n

Published in the United States of America in 2018 by Lynne Rienner Publishers, Inc. 1800 30th Street, Boulder, Colorado 80301 www.rienner.com

and in the United Kingdom by Lynne Rienner Publishers, Inc. Gray’s Inn House, 127 Clerkenwell Road, London EC1 5DB

© 2018 by Lynne Rienner Publishers, Inc. All rights reserved ISBN: 978-1-62637-731-8 (hc : alk. paper)

Library of Congress Cataloging-in-Publication Data A Cataloging-in-Publication record for this book is available from the Library of Congress. British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library.

Printed and bound in the United States of America

The paper used in this publication meets the requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48-1992.

5  4  3  2  1

To my mother Cindy, for everything, to Zac and Tim, as a fraternal gesture of love and respect, and to Monty, for getting me started and keeping me going

Contents

List of Tables and Figures Acknowledgments 1 2 3

4 5 6 7

Democratization and the Mischief of Faction Factionalism in the Post–World War II World Avoiding Factionalism: Senegal, Taiwan, Uruguay Persistent Factionalism: Bangladesh, Bolivia, Zimbabwe Overcoming Factionalism: Chile, Comoros, Estonia, Tunisia Factionalism and Autocratization: Belarus, Central African Republic, Egypt, Thailand Managing the Mischief of Faction

Appendix: The Societal-Systems Process Model References Index About the Book

vii

ix xi

1 31 65 85

113

143 173

191 203 211 215

Tables and Figures

Tables 1.1 2.1 2.2 2.3

2.4 2.5 2.6 2.7 2.8 2.9 7.1

Most-Different Systems Design Matrix Factionalism Onset and Ending Event Code Descriptions Factionalism Cases in Sub-Saharan Africa, 1946−2015 Factionalism Cases in the Middle East and North Africa, 1946−2015 Factionalism Cases in Asia, 1946−2015 Factionalism Cases in Latin America, 1946−2015 Factionalism Cases in Europe, 1946−2015 Frequency Distribution of Primary Faction Types Faction Type Distribution by Region Comparison of Mean Fragility Levels Comparative Analysis Findings

22 34 36

41 42 44 51 56 58 62 175

Figures 1.1 A.1 A.2

The Societal-Systems Process Model Marshall’s Societal-Systems Process Model Marshall’s Societal-Systems Process Model, Modified

ix

16 192 198

Acknowledgments

I am grateful for the enduring support and friendship of Monty G. Marshall, whose Societal Systems Research Inc. funded much of this book’s research and who set me to studying factionalism when I was a first-year doctoral student. I also owe thanks to the hardworking researchers at his Center for Systemic Peace who contributed to the factionalism review project that informed this book, including Min Zaw Oo, Eliot Elzinga, and Gabrielle Elzinga-Marshall. I thank my former Simmons College student Laura Blume for her work assisting me in coding and analyzing factionalism data, and for being a source of inspiration in her own right. I also owe particular thanks to my editor and publisher, Lynne Rienner, for giving this book a chance. Her personal touch and honest feedback have been immensely helpful as the book moved from conference paper to prospectus, to draft, and to the finished manuscript. I also thank her assistant, Nicole Moore; copyeditor Diane Foose; production assistant Allie Schellong; and the entire production team for their work in getting the book to print. I am also grateful for the anonymous reviewers of the manuscript, who provided the most detailed, constructive, and genuinely helpful criticism I have ever received. This research would not have been possible without Simmons College and the tremendous levels of ongoing support, resources, advice, and collegiality I received from Dean Leanne Doherty and Associate Provost Catherine Paden. I also wish to acknowledge my department chair, Denise Horn, administrator Mark Valentine, and colleagues Lena Zuckerwise, Abel Amado, Kirk Beattie, Ambassador Mark Bellamy, and xi

xii

Acknowledgments

Kristina Pechulis for their support and friendship. I owe much inspiration to my current and former students, in particular Sasa Tang, Alex Loughran-Lamothe, Abigail Walters, and Alyssa Conley, and research assistants Lydia Anderson, Kelsi Field, and Shelby Goodwin. While he left this world too soon to appreciate this acknowledgment, I owe a debt of gratitude to my late grandfather and roommate, Harley Cole, for his frequent reminders to work on this book, and for much more frequently distracting me from that task with lessons on tractor repair, among other things. I am also grateful for the support and faith of my grandparents Juergen and Marsha Michel, my parents and brothers, and the rest of my family, immediate, extended, in-law, and honorary. And I thank Andy Clarke, although he really is included under family, for his resolute friendship and for reading and commenting on the book in draft form. Finally, I will always be grateful to my wife, Shannon, for everything she does.

1 Democratization and the Mischief of Faction

In 1787 James Madison decried the “mischiefs of faction” in “Federalist 10,” saying that factions, which he defined as parochial conglomerations of interests averse to the rights of others or the interest of the community, produce the “mortal diseases under which popular governments have everywhere perished” (1787: 10). Madison’s concern— derived from his analysis of the Athenian and Roman experiments with democracy—was prescient. Some 220 years later, the US Political Instability Task Force (PITF), combining senior scholars and policymakers, published its Phase IV findings on the causes of political instability (i.e., civil war, revolution, genocide) in the 2010 volume of the influential American Journal of Political Science (Goldstone et al. 2010). In a distinct echo of Madison’s warning, the PITF found that a single condition, factionalism, derived from Madison’s definition and coded in the Polity IV data series, was by far the most important predictor of political instability events around the world since the 1940s. The PITF’s Phase IV findings, which confirmed the results of its previous analyses, spurred significant research into the problem of factionalism, including a 6-year study by me and a research team at the Center for Systemic Peace led by Monty G. Marshall, reviewing every change in governance for 167 states (all countries with populations greater than 500,000), back to 1955 and, in some cases, independence.1 In addition to more than 100 country reports, this study also produced a body of research, conference papers, and publications by me, alone and in collaboration with Marshall and others, situating factionalism in the context of governance more broadly and analyzing the relationships between factionalism and related phenomena.2 The results of these 1

2

Democratization and the Mischief of Faction

analyses confirm both Madison’s warnings and the PITF’s findings: factionalism is a distinct condition of state-society interaction that is endemic to the democratization experience, and it is one of the most mortal diseases faced by government. Madison called for any “friend of popular government” to “provide a proper cure” for the factional condition, preferably without violating liberty and diversity (1787: 10). I have written this book to that end. In these pages I seek to show within the myriad experiences of democratizing states, common factors, policies, and strategies that may point toward smart practices for managing (or even avoiding) overt factionalism in the democratic transition process as well as those policies and strategies that states should avoid. Before that, however, I need to define our terms and place factionalism into a theoretical framework, mindful of its position vis-à-vis other, more established bodies of scholarship. Thus, I organized this book into three key parts: (1) defining factionalism and situating the condition within a broader theoretical framework of state-society relations (this chapter and the Appendix); (2) investigating the nature of factionalism from a global (macrocomparative) perspective, describing all cases of factionalism from 1946 to 2015 (Chapter 2); and (3) analyzing the experiences of 14 countries in a panel of most-different systems (MDS) analyses to distill smart practices for avoiding or managing factionalism in the democratization process (Chapters 3 through 6). In the conclusion (Chapter 7), I synthesize the book’s findings and briefly consider ramifications for mature democracies, including the United States. I begin this introductory chapter by defining factionalism, then briefly situate the concept of factionalism in the context of recent research into polarization and democratization across the social sciences. After that, I concisely introduce a theoretical framework for conceptualizing factionalism (a full theoretical treatment can be found in the Appendix) before concluding with an outline of the book. Defining Factionalism Factionalism is a naturally occurring condition of political participation, common to but latent in autocratic regimes, characterized by persistent systemic polarization of political society, which frequently emerges during the political liberalization process associated with the initial stages of democratization. While the definition of factionalism used here is based on Madison’s concept of faction from “Federalist 10,” Madison’s defini-

Democratization and the Mischief of Faction

3

tion is problematic in that it could also apply to interest groups or, indeed, political parties, in a conventional mode of political participation. Who decides when a party’s, interest group’s, or labor union’s activities have become averse to the rights of the community? The definition of factionalism used here (and operationalized in Polity IV) avoids these problems by describing a distinct systemic condition of political participation and state-society interaction, which reflects patterns of political behavior that are both systemic and sustained, characterized by extreme polarization (beginning with elites but including, typically, mass mobilization), antisystem rhetoric, disruptive political behavior, coercive behavior by the state, and parallel and reinforcing identity-based, economic, and political cleavages. A faction, in turn, is a conglomeration of individuals, groups, and often political parties, organized around a common goal of controlling the polity and removing the opposition faction from the political landscape, without regard for the effects of their efforts on the political system or the welfare of the society as a whole. While an interest group, political party, or identity group may form part of a faction, and may often serve as the most visible and identifying leadership within a faction, such a group cannot, itself, be a faction in this sense. Indeed, it is typical for factions to fracture if they are successful in eliminating the opposition faction, as they often include groups that otherwise possess distinct, competing, and incompatible interests. While recognition of the importance of this kind of factionalism has only recently reemerged, the concept itself is old. Among political philosophers, Madison’s definition of faction was quite common in the late eighteenth century, and can be seen in George Washington’s Farewell Address (drafted by Alexander Hamilton) and in the work of Thomas Jefferson (Sartori 1976). Henry Bolingbroke, David Hume, Edmund Burke, and Voltaire engaged in a vigorous debate over the meaning and import of the terms party and faction, decades before the protagonists of the American Revolution adopted the vernacular of the argument (Bolingbroke 1734; Burke 1770; Hume 1742). Epitomized by Voltaire, “the term party is not, in itself, loathsome; the term faction always is” (1764: 765). Intriguingly, as the term party came into common use and was developed theoretically by party theorists such as Moisey Ostrogorski, Robert Michels, and Harvey Mansfield, the term faction, as used by Madison and refined here, gradually left the vocabulary of most political scientists. Giovanni Sartori, in Parties and Party Systems, traces the evolution of the term party and its meaning as distinct from this older concept of faction:

4

Democratization and the Mischief of Faction

Parties are not factions; that is, unless a party is different from a faction, it is not a party (but a faction). . . . Parties are often criticized, but they are not an evil by definition . . . factions are an evil. . . . To be sure, party members are not altruists, and the existence of parties by no means eliminates selfish and unscrupulous motivations. . . . The difference is, then, that parties are instrumental to collective benefits, to an end that is not merely the private benefit of the contestants. Parties link people to a government, while factions do not. Parties enhance a set of system capabilities, while factions do not. . . . Parties are functional agencies . . . while factions are not. . . . [F]actionalism is the ever-present temptation of a party arrangement and its ever-possible degeneration. (1976: 22)

Unfortunately, Sartori’s plea that the distinction between parties and factions “should be kept conceptually firm” went unheeded, and the term factionalism evolved to describe a distinct subgroup or caucus within a political party. In the US case, for example, political scientists might refer to the libertarian faction of the Republican Party. However, the mischief of Madisonian faction—systemic sustained polarization of political society, organized around parallel and reinforcing identity, economic, and social cleavages, employing antisystem rhetoric and disruptive or coercive political tactics, with factions pursuing each other’s destruction without regard for the cost to the welfare of the community—certainly did not fade away, and neither have its deleterious effects on polities. Indeed, the recent “waves” of democratization have made factionalism all the more common in the past century, affecting nearly every case of democratization. Empirical work on defining and understanding these different experiences associated with factionalism has continued across multiple academic fields, in many ways as the central problem of the post−World War II era, although this scholarship has not typically used the terms faction or factionalism. In political science and sociology, studies of ethnic violence, political polarization, and sectarian violence have emerged to analyze and explain parts of this broader concept, but in isolated and disaggregated fashion. In social psychology, studies of in-group and out-group dynamics have focused on the same phenomenon from a different lens, offering explanations for how factions form, again without using the term. Substantial portions of the fields of peace studies and conflict analysis and resolution are dedicated to the problem of factionalism, without using the term. These studies have all approached a single phenomenon, factionalism, but in searching for the cause have identified a particular aspect (e.g., ethnicity or religious identification) of the factions involved—the symbolic issue

Democratization and the Mischief of Faction

5

that appears to divide the polarized groups, the effects of polarization in political behavior, or the process by which groups form—without recognizing that all may in fact have been researching a common phenomenon: factionalism. Factionalism, as discussed here, is thus not a new term or concept, but reflects the original meaning of the term faction. It is, I hypothesize, a common phenomenon behind the apparent social cleavages, polarization, and sectarian violence, among other outcomes, that have been well researched (separately) by generations of scholars from different subfields using different terms. Of these, the most important for this analysis of factionalism is the study of sociopolitical polarization—the study of how and why societies separate into two groups with distinct, and opposing, interests, goals, and often identity constructs. The process that drives political polarization, the nature of the polarized political state, and the methods used to manage it are all essential to the study of factionalism. Indeed, the particular concept of factionalism explored here has sometimes, especially in European political science scholarship, been called “polarization” and found to be a critical predictor of political violence. The effect of political polarization on economic growth has been a particularly productive research topic in recent years. Timothy Frye (2002) addresses the effects of extreme polarization between communist and postcommunist groups that he calls “factions” on economic growth in Central and Eastern European countries. Frye borrows significantly from Stephen Haggard and Robert Kaufman, who use antisystem parties, which they describe as “left and populist parties that have historically mobilized around anti-capitalist or anti-oligarchic protests,” to measure political polarization (1995: 167). Frye uses the term faction to depict these groups, which he identifies as polarized over sociopolitical and economic ideology, although Frye does not address systemic conditions created by the contention among his factions, nor the etymology of the word “faction.” Although Frye does not stop to consider the meaning of the term faction that he applies to these groups, he does identify an important correlate of intense polarization, which is also a key part of our definition of factionalism: conflation of policy debates with identity politics associated with merging social, economic, and political cleavages. Originally identified by Alberto Alesina and Allen Drazen (1991), refined by Alberto Alesina and Howard Rosenthal (1995), and further examined by Morris Fiorina (1996), this conflation results in a war of attrition between factions whereby no policy consensus can emerge until one group “wins” by marginalizing its opponent. Only when one faction

6

Democratization and the Mischief of Faction

wins the political struggle (e.g., after an election allows the victor to subjugate the loser) should we expect coherent government policy, a productive response by the private sector, and improved economic performance. The consolidation of political forces around a roughly similar policy ends the war of attrition and allows the winners to shift the costs of transformation onto the losers (Frye 2002). This result of polarization, a war of attrition over policy, has been identified in many other studies of polarization, across different types of cleavages. William Easterly and Ross Levine (1997), for example, developed indexes of ethnic fractionalization to measure the effect of polarization along ethnic cleavages on economic growth in Africa. Philip Keefer and Stephen Knack, of the World Bank, conducted a similar study of ethnic polarization, also incorporating economic inequality, finding similar results (2000, 2002). In situations where political systems are highly polarized, regardless of the nature of the social cleavage (e.g., religious, ethnic, ideological), political stalemate between the two groups develops and, unless a force emerges that encourages cooperation, will prevent the development of coherent policy until one group emerges victorious or until the divisions shake the polity apart. This tendency is one of the reasons why factionalism, which incorporates this combination of polarization along parallel and reinforcing cleavages (regardless of cleavage type), is so dangerous. The emergence of polarization is particularly common in democratizing states and young democracies, which tend to be characterized by political and social divisions between supporters and opponents of the preceding autocratic regime who often benefited from and suffered under that regime, respectively. Adrienne Lebas (2006), for example, argues that polarization in new democracies is caused not by preexisting ethnic, religious, or ideological cleavages themselves, but as the result of mobilization strategies of social movements and political parties against the autocrat, which often reflect patterns of social and political activism along those cleavages. Her case study of Zimbabwean politics demonstrates that political groups (generally, parties) purposefully polarized society to build and mobilize particular constituencies. Once set in motion, however, the polarization took on a life of its own, exceeding its instigators’ realm of control. Lebas’s work identifies an important element in our conceptualization of factionalism: elites seeking greater political power will use polarization and mass mobilization as strategies to build their organizations and mobilize their constituents against the opposing faction. A similar dynamic is identified in the cases of Egypt and Thailand in Chapter 6 of this book.

Democratization and the Mischief of Faction

7

Sometimes, however, factionalism seems to occur endogenously, without an instrumental cause. Studies of such natural causes of polarization are numerous. Murat Somer (2001) addresses a phenomenon in Yugoslavia similar to that studied by Lebas in Zimbabwe, using cascade theory to examine how polarization became self-propagating along ethnic lines as divisive political questions were associated with ethnic identity (Somer 2001). Why ethnicity becomes salient, however, is left unclear. Several studies on polarization in Venezuela have highlighted the combined effects of natural polarization and strategic polarization, especially as used by Hugo Chávez in his rise to power (Corrales 2005; Garcia-Guadilla 2003; Lopez Maya 2004). Sook-Jong Lee (2005) argues that a combination of economic forces and divisive elite groups in South Korea, including political leaders, the media, and leaders of civic movement organizations, has caused polarization along socioeconomic cleavages. The model that I employ builds on this work, assuming that autocratic governments inherently create divisive conditions among economic and political elites who, upon liberalization, frequently then mobilize the population along existing identity cleavages to garner their support in a bid to seize control of the polity and right the collective wrongs inflicted or proposed by the opposition. This also means that overt factionalism, in theory, could be avoided through strategies that minimize elite incentives to mobilize the population such as through a guided transition that incorporates opposition elites into the polity, or by co-optation of opposition elites by the autocratic regime. Avoiding the polarization associated with factionalism has been rare, and managing polarization and stopping polarization from leading to strife is particularly difficult for young democracies. Such states are particularly vulnerable because they are obligated to allow manifestations of factionalism (e.g., mass protests and electoral boycotts) to proceed, but lack the conflict resolution institutions (e.g., independent judiciary, legislative checks on executive authority, participatory local and regional governments, rule of law) needed to manage and depolarize an intensely polarized society. Polarization (and its consequent political attrition) will continue to worsen until such institutions are implemented or, as in so many cases, political violence erupts to resolve the polarization. In proposing solutions for young democracies some, especially those writing on economic consequences of political polarization, have argued in favor of improving systems of property rights (Alesina and Drazen 1991; Alesina and Rosenthal 1995; Frye 2002; Keefer and Knack 2000, 2002) while others have focused on the role a strong judiciary can play in managing polarization-related conflict (Garcia-Guadilla 2003; Corrales

8

Democratization and the Mischief of Faction

2005). However, given the weak judiciaries, strong executives, and nascent legal systems in many young democracies, it is difficult to see either solution having a significant impact on the reduction of polarization. These strategies may be more effective, however, among consolidated democracies, as suggested for example by studies of Malta (Cini 2002) and the United States (Layman, Carsey, and Menasce Horowitz 2006). In this book, I look to such institutional and structural forces as possible factors for mitigating factionalism in Chapters 3 through 6. The example case study of Belgium in Chapter 2 confirms the findings of Michelle Cini and Geoffrey Layman, Thomas Carsey, and Juliana Menasce Horowitz that polarization in mature democracies can indeed be intensified by structural rules or institutional reforms that inhibit self-correcting mechanisms in the democratic system. In general, polarization studies make a significant contribution to our understanding of how the advanced state of polarization characteristic of factionalism occurs, and offer suggestions into where we should look to determine how polarization might be managed to avoid strife. Studies of sectarian and ethnic conflict offer analyses of a specific type of societal polarization, with factions divided by religious and ethnic identity, or both. As factionalism is, by definition, a condition characterized by highly polarized political participation that has taken on parallel and reinforcing cleavages along identity lines (in addition to other factors), these two areas of research are quite relevant to a complete understanding of the factionalism problem. Polarization matters, and identity cleavages matter, but they are individual pieces of the broader and more fundamental problem of factionalism. The term sect, like faction, has developed a pejorative meaning in its recent usage. Studies of sectarian conflict usually refer to conflict between religious groups, including groups within a larger religious orientation (e.g., Catholic and Protestant Christians in Northern Ireland, or Sunni and Shiite Muslims in Iraq) or between different religions (e.g., Hindus, Muslims, and Sikhs in India). Ethnic conflict studies, in contrast, analyze conflict between competing groups of people who identify with each other on the basis of a presumed common genealogy or ancestry, or because of recognition by others as a separate ethnic group. These identity groups are often intersectional, as in the case of Sri Lanka with the Tamil/Sinhalese ethnic cleavage overlapping a religious Hindu/Buddhist cleavage, or in Myanmar where the Burmese/Rohingya cleavage overlaps a religious Buddhist/Muslim cleavage; both cases also feature distinct geographic, economic, and linguistic cleavages parallel to these identity lines. In cases of factionalism, such intersectional cleavages are

Democratization and the Mischief of Faction

9

parallel and reinforcing, and often are made salient by elites through political rhetoric and mass mobilization. Understanding how groups form their identities has been the subject of considerable theoretical development, particularly but not exclusively in social psychology. Ted Robert Gurr has described groups as “psychological communities whose core members share a distinctive and collective identity based on cultural traits and life ways that matter to them or to others with which they interact” (Gurr 1993: 3). While Muzafer Sherif and Carolyn Sherif have identified tendencies of the group to pressure members to conform to group norms and goals to achieve cohesion, individuals also have a natural incentive to associate with an identity group (Sherif and Sherif 1953; Kawachi and Berkman 2000). Being part of a group offers security and a sense of meaning, and provides an individual a cohort with which to associate (J. Ross 1993). The mutually beneficial relationship between identity group and member, documented in social science as early as 1906, leads to in-group identification and intergroup differentiation (Summer 1906). When an individual’s group is the center of his or her frame of reference, fellow group members are perceived as being intrinsically closer than out-group individuals who are perceived as outsiders (Levine and Campbell 1972). Many scholars have applied identity group theory to both sectarian and ethnic conflicts. Simon Haddad (2000) analyzed sectarian polarization in Lebanon, measuring each faction’s strength of group consciousness in terms of group cohesion, group solidarity, and satisfaction with group membership. He found high degrees of group consciousness and was able to conclude that, because of a lack of a strong national identity, individuals were more likely to associate themselves with their sect (faction) than with the Lebanese state. Caroline Nagel (2002) also applied group consciousness theory to government attempts to develop a common national group consciousness that would supersede that of its religious factions, and came to conclusions that were similar to Haddad’s despite his focus on spatial identity effects. Many similar studies linking ethnic, sectarian, and ethnosectarian conflict to identity mobilization and group consciousness theory have surfaced recently. Peter Shirlow (2003), John Alderdice (2007), and Sheena McGrellis (2005) have produced substantial work on group consciousness and identity formation in Ireland. Vali R. Nasr, Mariam Abdou Zahab, and Muhammed Makki, Saleem Ali, and Kitty Van Vurren have done similar work in Pakistan, and Manochehr Dorraj has applied identity mobilization theory to Iranian politics (Dorraj 2006; Nasr 2000, 2002; Zahab 2002; Makki, Ali, and Van Vuuren 2015). The

10

Democratization and the Mischief of Faction

Arab Spring revolutions have produced significant bodies of literature on the development of polarization around identity groups across the Middle East and North Africa (MENA). Like the work of Haddad and Nagel, these studies of polarization in diverse countries have found that high group consciousness within polarized groups is correlated with the lack of a strong national identity and, perhaps not surprisingly, a strong tendency toward conflict and instability. These studies, from multiple disciplines and subfields, point to several key findings that underlie a common pattern associated with democratization. Factionalism is latent in most, if not all, societies under repressive regimes because repression effectively divides society into supporters and opponents of the repressors. And it often distributes resources down those lines as well, encouraging group formation and creating parallel and reinforcing cleavages along, frequently, identity lines. Authoritarian governance strategies—suppression and repression, in practice variations on martial law—keep factionalism from manifesting in its overt form. However, when an autocratic state initiates political liberalization, whether due to international pressure, domestic pressure, or, more rarely, out of the benevolence of the autocrat, the latent factionalism is able to manifest. Factions emerge along the lines of one or more major existing cleavages, which become parallel with and reinforce other social, economic, political, or identity cleavages (due in part to instrumental efforts by elites), and display different characteristics depending on institutional, historical, and cultural settings. These factions (or, more accurately, the elites leading them) employ antisystem rhetoric and utilize disruptive political tactics, often in the form of organizing electoral boycotts, mass demonstrations, or political riots, while the faction in power responds coercively but typically within certain bounds short of violence or repression (e.g., electoral manipulation rather than forced disappearances). Factions claim that control over the polity, marginalization of the opposite faction, and often wholesale reform of the political system constitute the only remedy for contention and, sometimes, survival. As these large groups become institutionalized and gain power, they often act to limit the ability of the opposite contender to act and to co-opt moderates into their groups (or, in the worst cases, kill them). The polarization of these groups thus becomes sustained; leaders cultivate in-group status by focusing on identity rather than policy cleavages, in part to minimize the often quite significant policy divisions among their component organizations, and thereby form a more cohesive unit. This in turn aggravates political contention since any political disagreements become associated with

Democratization and the Mischief of Faction

11

symbolic group identity rather than questions of policy. As the system polarizes, the number of main factions approaches two, and political issues become increasingly identity based, which intensifies group identity coherence and reinforces incentives for group formation. Liberalizing governments are then faced with a dilemma: factionalism is dangerous and damages political institutions, but cracking down on the increasingly raucous political participation means reverting to a more authoritarian governing strategy and effectively forces factionalism back underground. To stay the course means to take what might seem like an enormous risk and allow the factional political dynamic to play out on the national stage, with potentially destabilizing effects. Historically, factionalism in and following democratic transitions was dealt with by one of three strategies: (1) gradual, although often violent, resolution of polarizing questions within the polity; (2) territorial expansion delaying resolution of polarizing questions or allowing, encouraging, or forcing relocation of polarized identity groups; and (3) by marginalizing or disenfranchising politically significant factions. Today, however, political elites are much less patient, are generally unable to rely on territorial expansion to delay the resolution of factionalism, and are much more closely watched than the American founders. The modern world anticipates that developing democracies will undergo immediate elections, guarantee universal franchise, and protect civil liberties and political rights, expectations that can be difficult and even dangerous to meet while the factionalism condition is manifest. Japan and Germany each waited years before holding their first national elections for a chief executive after World War II, despite boasting nearly every positive precondition of successful democratization known to social science—strong economies; experience with democracy in the past; large middle classes; homogenous religious, ethnic, and national identities. Yet the Western world expected Iraq, with a political society and neighborhood much less conducive to liberalization, to hold free and fair elections with universal suffrage in a much faster time while simultaneously warding off civil war. Perhaps not surprisingly, of those societies that have become outwardly factional, which include most democratizing states, more than half have experienced a failed democratic transition or a major armed conflict. It is no coincidence that these crises tend to come during, or immediately following, a national election, as elections explicitly ask members of the population to mobilize in support of opposing elites. While autocratic systems can maintain stability by coercively preventing opposition organization and political discourse in general—this

12

Democratization and the Mischief of Faction

is the autocratic modus operandi—young democracies are particularly vulnerable to factionalism. That is because, in addition to inheriting a factional legacy left by autocratic predecessors, young or incomplete democracies are expected to manage the factional condition peacefully, but lack the institutions and experience required to effectively manage polarization within conventional deliberative norms. Thus, while only one autocracy (Nepal, 2002−2005) has been coded as openly factional during the contemporary period, factionalism in anocracies is nearly endemic and occurs even in relatively institutionalized democracies. The concept of factionalism offers a conceptual reimagining of the preeminent problem facing young democracies, but is not new. Employed here, factionalism reflects the central problem underlying studies of political polarization (predominantly sociological studies of group formation, and studies of ethnic and sectarian conflict), all of which have identified independent aspects of the problem of factionalism, as defined here, as driving forces behind political instability, autocratic backsliding, and failed transitions. Operationalizing Factionalism This book operationalizes factionalism using the definition provided in the Polity IV users’ manual (Marshall, Jaggers, and Gurr 2010: 25).3 In the Polity scheme, factional is the middling (code 3) category, of five categories on the ordinal competitiveness of participation (PARCOMP) indicator, which categorizes political competitiveness and, to a degree, regulation of competitiveness by the states, and ranges from 1 (repressed) to 5 (competitive). PARCOMP is one of six indicators used to characterize a regime’s pattern of authority in the Polity IV scheme, and one of two variables (with PARREG [regulation of participation]) used to characterize the political competition (POLCOMP) component variable. PARCOMP measures the extent to which “alternative preferences for policy and leadership can be pursued in the political arena” (Marshall et al. 2004: 25). In factional polities, measured as PARCOMP = 3, parochial “factions compete for political influence in order to promote particularistic agendas and favor group members to the detriment of common, secular, or cross-cutting agendas” (Marshall et al. 2004: 25). In practice, Polity IV coders identify and code the factional condition based on its manifestations in political society in ways that are systemic and sustained, and coding occurs with a high degree of intercoder reliability due in part to the extensive training required of Polity coders.

Democratization and the Mischief of Faction

13

Factionalism onsets are typically dated to a major act of liberalization that allowed the factional condition to manifest, such as the lifting of martial law or promulgation of a new constitution. Alternatively, factionalism onsets may be dated to an event or process that is associated with major manifestations of factionalism, such as a national election. Coders are trained to identify indicators of polarized politics, including the emergence of antisystem rhetoric, the formation of unnatural alliances among groups with conflicting interests, incidents of mass mobilization (e.g., mass demonstrations, protests), coercive behavior by government (e.g., extrajudicial detentions, election manipulation), and disruptive political tactics (e.g., electoral boycotts, widespread strikes), any of which may indicate polarized politics and the potential for the existence of factionalism. Factionalism, however, is typically coded only when most or all of these indicators are present on a sustained and systemic basis. Polity IV is coded annually, with news events archived throughout the year and analyzed at year’s end for evidence of changes to any of the indicators that comprise the Polity framework, including PARCOMP (and, thus, factionalism). Any of the indicators of factionalism, but particularly disruptive tactics with evidence of mass mobilization (e.g., election boycotts, mass demonstrations), coercion by the state (e.g., illegal detentions, election postponement), and emergence of antisystem rhetoric or action, cue coders to research a case more thoroughly. To do so, they must investigate whether additional indicators of factionalism exist and, if so, whether there is evidence of sustained factionalism, or whether a case would be more accurately coded using a different value (e.g., PARCOMP = 2 for suppression; PARCOMP = 5 for competitive).4 It is particularly difficult for the Polity coder to differentiate between mass mobilization and disruptive political tactics that occur in the context of conventional politics (e.g., strikes are normal in French political behavior) and such tactics that are truly indicative of the factional condition, which features similar forms of apparently disruptive political behavior. Although student strikes that shut down all universities and public transportation are normal in Paris or Dakar, the same would be an indicator of significant political polarization in London or Accra (and, thus, a cue to look for additional factionalism indicators). The difference is often in the distinct nature of rhetoric, in political tactics and their apparent purposes, and in the nature of factional alliances because under factionalism the political system polarizes into two distinct political poles vying for control of the polity and marginalization of the opposing faction, and frequently displays state coercion and antisystem rhetoric and tactics. Because factionalism is a middling

14

Democratization and the Mischief of Faction

condition on an ordinal variable, coders carefully consider borderline cases and generally prefer to maintain the code applied the previous year in the event of significant uncertainty. Coders also discuss and debate such cases at length, looking for evidence in the historical record to inform a coding decision. Coders keep a log of the events that caused them to assign a coding change and these logs, as well as the logs from the original Polity coding archive, are available from the Center for Systemic Peace on request. I strongly recommend that readers interested in the details of Polity’s structure or coding practices consult the Polity IV user’s manual (Marshall, Jaggers, and Gurr 2010). After the Political Instability Task Force published its Phase IV findings highlighting the importance of factionalism in predictive models of democratic backsliding and in political instability, I worked with Monty G. Marshall, director of the Polity project and president of the Center for Systemic Peace, to conduct a comprehensive review of the Polity data series. Our emphasis was on review of factionalism onsets and periods, partly to investigate criticism by Vreeland (2008) that the Polity scheme conflated political violence with the coding of factionalism. We reviewed every change in governance in every country in the world back to 1946 (or in some cases independence), verifying that Polity code changes were valid, accurately dated, and detailed in the logs, and we altered the dataset in those few cases where codes were demonstrated to be erroneous (primarily among mature democracies whose complex histories were often neglected by the original coders). This review, which took 6 years to complete and analyze, resulted in the production of more than 100 country reports noting every change in governance in every country ever coded as factional in the Polity scheme. This body of reports—the source of much of the data for this book—describe, for every episode of factionalism in the world between 1955 and the completion of the reports in 2010, the identities of the core factions, the background information on, and the nature of, the onset event, and the results of the factionalism experience. I then led a review and coding exercise of those country reports with a team of graduate and undergraduate research assistants at Simmons College in 2013 and 2015, coding these country reports and, where country reports were missing or needed updating, the historical record as maintained by the Keesing’s World News Archive (Keesing’s 2017) and the BBC Monitoring: International Reports archive available through Lexis-Nexis (BBC Monitoring 2017). Summaries of these results are presented in Chapter 2. While factionalism certainly has the potential to be destructive, this review project and our subsequent analysis demonstrated that fac-

Democratization and the Mischief of Faction

15

tionalism is distinct from, and not conflated with, political violence or instability. Factionalism is nearly universal among democratizing states, and many factional states avoid political violence and major episodes of political instability altogether. Furthermore, violence by the state, or widespread political violence in the population, in most cases actually disqualifies a case from being coded factional, as political violence typically indicates suppression (2) or repression (1) to be more appropriate PARCOMP values. State repression precludes the occurrence of factionalism whereas effective resistance to state authority (e.g., revolution or civil war) also precludes factionalism by fragmenting the polity and establishing a rival separatist polity, which is distinctly coded in the Polity scheme. This review also dramatically strengthened our confidence in the specific details of factionalism episodes between 1946 and 2015 (and the Polity data series more generally) by confirming the accuracy of those coding decisions (and changing the codes when they were inaccurate), tying every coding change to a specific event, and documenting patterns of state and societal behavior during every factionalism episode. This inductive enterprise to verify validity within the Polity scheme was accompanied by a deductive effort, primarily by Marshall, to situate factionalism (and political competition generally) in a theoretical framework, briefly discussed here and explained in substantially more detail in the Appendix for those interested in a longer theoretical treatment. Marshall’s societal-systems process (SSP) model conceives of democratic and autocratic regimes as independent from, but existing at natural equilibrium points along, a continuum characterized by different types of political participation and contention within a polity. These types range from conventional politics to open warfare. Factionalism is a distinct middling condition on this continuum, as well as the equilibrium point at which autocracies are most common, primarily due to the polarizing nature of autocratic governance (Marshall and Cole 2008, 2012; Marshall 2014). Factionalism under autocratic governance is latent due to the repressive tactics employed by autocratic regimes to maintain order. Liberalization (in terms of deregulation and increasing density of complexity, defined here as social and political participation in the polity) and militancy both undermine autocratic authority. Democratic systems, in contrast, are at an equilibrium point in conventional politics. And democratic institutions weaken as political energy in the system increases, causing political tensions to escalate and driving the political process toward contentious politics and issue factionalism modes, which with further polarization and cleavage alignment can lead

16

Democratization and the Mischief of Faction

to the condition of factionalism being studied here, called “polar factionalism” in the SSP model. In the worst cases, intensification can lead to probing militancy and even open warfare. In this model established democratic systems are generally self-correcting, so migration along the continuum past contentious politics is typically rare, while autocratic systems can rely on force to restore equilibrium. Anocracies and young democracies are particularly vulnerable to migration rightward along the SSP continuum since they lack the institutions and experience necessary to self-correct and are inhibited from using force to resolve contention and stop escalating tensions. So too, established democratic systems might cease to be self-correcting if democratic institutions are altered or undermined, as I show in this book has occurred in Belgium, and potentially the United States, in recent years. Figure 1.1 presents Marshall’s general SSP model (Marshall and Cole 2012), which takes the form of a schematic including six conditions from conventional politics through open warfare. These conditions are distinguished by the nature of political participation and regulation of political affairs in the polity, which manifest in the degree of polarization, the rhetoric used in political discourse (especially identitybased symbolism and rhetoric), the nature and degree of mass mobiliza-

Figure 1.1 The Societal-Systems Process Model

Note: Monty G. Marshall, in Marshall and Cole (2008) first developed the societalsystems process (SSP) model and the diagram on which this revised figure is based. It is also discussed at length in the Appendix, and in Marshall’s various works on societalsystems analysis (Marshall 2011, 2014).

Democratization and the Mischief of Faction

17

tion and disruptive political tactics, the nature of political alliances and coalitions (particularly the emergence of unnatural alliances and parallel social cleavages), and the scope and organization of extreme or antisystem rhetoric and behavior. Note that the SSP model is not a model of regime type—many of these already exist in the political science literature—but rather the nature of state-society interaction, which helps to contextualize patterns of authority in different regime types and to explain common trajectories of regime transition. A more thorough discussion of this model and descriptions of each of these conditions are provided in the Appendix. Although any given state-society system may exhibit characteristics of any of the conditions on the continuum in Figure 1.1, most mature democracies exhibit political competition characteristic of the conventional politics condition, with a tendency to occasionally migrate rightward as problems emerge that escalate tensions, particularly when political cleavages align (either naturally or through instrumental behavior by elites) with identity, wealth, power, discrimination, or status. In general, though, for mature democracies, migration into nonconventional political conditions should be rare and typically temporary, as selfcorrecting institutions and political culture should disincentivize disruptive tactics, encourage cross-cutting cleavages, and create incentive structures that favor returning to conventional political dynamics. Furthermore, the degree to which politically salient cleavages are parallel and reinforcing in a factionalism episode should be inversely correlated to the probability of peacefully resolving a factionalism episode. Formally, I hypothesize the following: Hypothesis 1: As the number of parallel and reinforcing cleavages increases, the probability of achieving a peaceful resolution of the factionalism episode decreases.

Autocracies, in contrast, should cluster in the polar factionalism condition since such regimes generate the polarization underlying factionalism as they award spoils to their supporters and repress their opponents, although factionalism in such cases is latent due to repression of mass political participation. Unlike democratic systems, autocracies are generally not self-correcting, but rather rely on instrumental coercion, repression, and military force to return to their equilibrium point. For those autocracies that do not utilize force to restore order and repress manifestations of factionalism, liberalization allows open factionalism that will drive some state-societal systems leftward, with factions typically

18

Democratization and the Mischief of Faction

demanding more open liberal political participation. Many such states do respond with force, and either open factionalism is successfully repressed, or the state-societal system migrates rightward, degenerating into militancy and warfare. This can lead the most unfortunate cases, such as Central African Republic (explored in Chapter 6), to experience near-total collapse of central authority (state failure). For a state-societal system to move along the SSP model, there must be a change in its energy state. Conventional politics in a democracy is a complex system, requiring substantial ongoing energy inputs to maintain. Self-correcting mechanisms such as institutions and political culture in mature democracies, however, pose an obstacle to rightward migration in the process model that requires additional energy to overcome. Democracies thus exhibit characteristics of metastability:5 they exist at equilibrium under conventional politics but, with sufficient additional energy, they can move to a temporarily higher state of energy. Under the right circumstances, and depending on central government choices, state-societal systems can migrate to higher or lower energy states. A state-societal system migrating rightward reaches an alternative equilibrium point at factionalism where, if it has adopted coercive tactics associated with autocratic states, it can repress interaction at a relatively low energy cost that minimizes complexity in the system. At this point, which for an autocratic state is its natural point of equilibrium, a state-societal system is again metastable—maintaining equilibrium through minor systemic perturbations. We might make an analogy of the state-societal system to a ball on a curving slope, with democratic and autocratic equilibrium points analogous to valleys in the slope. Generally, it requires less energy to move rightward than it requires to move leftward in the model since complexity, which we might operationalize as the density of politically significant interaction among people and organizations, increases moving leftward on the model and decreases moving rightward. This is modeled in Figure 1.1. In practice, of course, democratization and autocratization are not linear processes, and some state-societal systems may effectively jump from one condition to another. For both autocratization and democratization, deviation from the equilibrium points in Figure 1.1. should generally be self-limiting, and thus migration should be rare. For autocracies, such migration is generally associated with transitions in regime or loss of the monopoly on the use of force (enabling militancy and open warfare). The societal, economic, and political disturbance associated with migration consumes tremendous energy and resources and, particularly in movement

Democratization and the Mischief of Faction

19

toward militancy, damages the social and economic networks that are essential for society to function. Those polities that transition through militancy toward open warfare risk fragmentation of the polity as the system collapses or fragments. The concept of state failure from the literature is also compatible with this model, with state failure or state collapse reflecting a transition into militancy, open warfare, and, typically between the two conditions, polity fragmentation.6 The PITF Phase IV findings indicate a strong statistical relationship without a strong a priori theoretical causal mechanism, making the factionalism research endeavor initially a largely inductive one. In the absence of a complete theoretical framework, and despite our detailed 6-year review of the data and coding decisions, some readers might still be concerned that the relationships between factionalism and democratization, and factionalism and political instability events, are tautological. Thus, while perhaps tedious for many readers, some may desire a more thorough treatment of the SSP model we use to conceptualize factionalism. The Appendix describes in full Marshall’s societal-systems process model (Marshall and Cole 2008, 2012; Marshall 2014), as well as my modifications, to situate factionalism as one among several modes of political behavior in complex societal-systems ranging from conventional politics to open warfare, and describes each of these modes (e.g., issue factionalism, probing militancy) in detail. Again, note that in theory and in coding practice, regime type is separate from this model of political behavior, although coding results demonstrate that regime types cluster around certain types of political behavior. I have slightly revised the SSP model by explicitly including concepts that were implicit in previous iterations of the model, specifically concepts related to social complexity such as metastability of societal-systems at the model’s equilibrium points. The SSP model allows us to situate factionalism theoretically to construct deductive hypotheses about factionalism and factionalism management strategies. In all, two-thirds of independent countries in the world—107 out of 167 in the Polity dataset—experienced factionalism in at least 1 year between 1946 and 2015, and some experienced multiple episodes of factionalism, for a total of 172 discrete onsets of factionalism.7 Of these episodes, approximately half ended in either a major episode of political instability (as defined by the PITF) or an incident of autocratization. The other half includes cases that moved from factional to more conventional forms of participation in a process of democratic consolidation as well as cases that continue to endure factionalism. A few other states managed to transition to democratic governance while avoiding

20

Democratization and the Mischief of Faction

factionalism. Determining how states can successfully manage factionalism short of political violence or autocratization or, better, avoid factionalism altogether, is the central goal of this book, as we respond to Madison’s call for a “proper cure” to the mischief of faction (1787: 10). In this book, I focus on explaining how to manage and mitigate the worst impacts of factionalism, which is a major impediment to democratization, but I do not set out to thoroughly explain democratic transition. Nonetheless, as an analysis and reimagining of the central problem facing democratizing states, this book builds on and engages with the work of multiple generations of scholars who have dedicated their careers to explaining divergent democratization outcomes. Democratization scholarship has focused on a wide array of causes of transition outcomes, including among others identity (e.g., culture, ethnicity, religion),8 economic development,9 resource rents,10 elite behavior,11 popular values related to democracy,12 neighborhood effects,13 international pressure and support,14 institutions and the timing of their development,15 and path dependence (especially of a pretransition regime type).16 This body of work is theoretically rich, featuring enduring debates and a substantial body of (often contradictory) empirical findings. It would require an entire volume to review thoroughly, although for readers unfamiliar with this body of work Barbara Geddes (2007) offers an excellent article-length meta-analysis of democratization to that date. My analysis in this book supports the importance of some of these factors in explaining democratization outcomes (e.g., elite behavior, international pressure, institutional development and timing), not as individual competing explanations of democratization as they have often emerged in the literature, but as smart practices that prevent or mitigate the excesses of a common problem of factionalism. As I discuss in more detail in the chapters that follow, factionalism thus explains multiple competing findings from the democratization literature, unified in a single holistic approach. In this book I identify these smart practices for managing or mitigating factionalism, and reach points of engagement with the existing literature on democratization, through a combination of deductive reasoning and inductive analysis. My methods include a hypothesis-testing macrocomparative analysis of the country reports and data collected over the course of this project, and an inductive analysis utilizing a panel of four most-different systems analyses of factional outcomes. In Chapter 2, I tell the story of factionalism uncovered by our review of the historical record, including data for all factionalism episodes from 1946 to 2015 presented in multiple tables organized by region, with information gleaned from the Polity IV dataset, more than 100 country

Democratization and the Mischief of Faction

21

reports, and news archives. This includes dates of factionalism onsets and endings, changes in governance structure during factionalism episodes, as well as information about the nature of factionalism onset events; primary, secondary, and tertiary factionalism cleavage types; and factionalism episode ending events. The utility of this information is illustrated with brief vignettes: descriptions of political history in Burundi, Peru, and Belgium, which help readers contextualize the tables of data, put the story back into the numbers, and allow them to find the stories of their own cases, regions, and events in the data tables, and then compare those stories to other cases systematically. I conclude the chapter with a macrocomparative analysis of factionalism onset events and factionalism cleavage types. Chapters 3 through 6 offer a panel of four most-different systems comparative case study analyses, organized by factionalism outcome, to investigate any practices or circumstances that could explain their common outcomes. In Chapter 3, I examine cases that avoided factionalism in their democratic transitions (to date), including Senegal, Taiwan, and Uruguay. In Chapter 4, I analyze cases that experienced persistent factionalism (episodes of long duration that were still ongoing at the end of 2015), including Bangladesh, Bolivia, and Zimbabwe. In Chapter 5, I investigate cases where factionalism episodes ended in democratization, including Chile, Comoros, Estonia, and Tunisia. However, it should be noted that successful factionalism management does not necessarily mean that these cases are fully consolidated, and backsliding or a return to factional politics is always a threat to democratic polities. Tunisia, for example, has experienced democratic governance for a much shorter period than Chile had prior to the 1973 coup, but is included here not because we assume its democratic transition is complete, but rather because it is one of the only states in its region to have survived a factionalism episode and transitioned to democracy (even if that democracy, against hope, turns out to be fleeting). Finally, in Chapter 6, I investigate cases where factionalism episodes ended in autocratization or the collapse of central government, including Belarus, Central African Republic, Egypt, and Thailand. Conducting multiple MDS analyses allows us to explore the factionalism phenomenon in great breadth at the cost of depth, which may disappoint regional experts accustomed to case studies or small-n comparative case study analyses. It has an added benefit, however, of building into the analysis sets of cases across chapters that control for region and, in some cases, economic, ethnic, and social indicators as well. Thus, Tunisia and Estonia pair with Egypt and Belarus, in Chapters 5

22

Democratization and the Mischief of Faction

and 6, respectively, offering a most-similar systems (MSS) controlled comparison built into the research design across those chapters. Indeed, Chapters 5 and 6 include four case studies instead of three to make these comparisons possible, and to include cases from the Latin American and MENA regions. Similarly, Bolivia (Chapter 4) and Chile (Chapter 5) offer an MSS analysis inside Andean South America, and the more methodologically adventurous might include Uruguay (Chapter 3) in that MSS comparison as well. The sub-Saharan African and Asian cases are so distinct from one another that an MSS comparison is implausible. Central African Republic is included specifically to include a case where factionalism management resulted in autocratization, which then gave way to state collapse. While uncommon, this factionalism outcome is one of the most dangerous and destructive and, as such, deserves special attention. Table 1.1 describes this panel of analyses, organized by chapter and region, and includes in parentheses the Polity institutionalized democracy (DEMOC) score for each case at the end of 2015. DEMOC is an ordinal 1–10 scale, where 1 denotes no substantial democratic institutions and 10 denotes substantial consolidated democratic institutions. Reflecting the MDS design, cases have similar DEMOC scores within chapters (organized by common outcomes): factionalism avoidance and democratiza-

Table 1.1 Most-Different Systems Design Matrix Chapter 4 Avoidance

Chapter 5 Persistence

Chapter 6 Democratization

Uruguay (10) Taiwan (10) —

Bolivia (7) Bangladesh (3) —

Chile (10) —

Sub-Saharan Africa Senegal (7)

Latin America Asia

Middle East and North Africa Europe

—

Zimbabwe (5)

—

Comoros (9)

Tunisia (7) Estonia (9)

Chapter 7 Autocratization

Central African Republic (−77: collapse) — Thailand (0) Egypt (0) Belarus (0)

Note: Numbers in parentheses are the Polity institutionalized democracy (DEMOC) score for each case at the end of 2015; DEMOC is an ordinal 1–10 scale, where 1 denotes no substantial democratic institutions and 10 denotes substantial consolidated democratic institutions.

Democratization and the Mischief of Faction

23

tion cases were democratic at the end of 2015, as indicated by DEMOC scores between 7 and 10; persistence cases were generally anocratic with DEMOC scores between 3 and 7; and autocratization cases had minimal democratic institutions at the end of 2015, tending to be anocratic or autocratic with DEMOC scores of 0, or in the case of Central African Republic −77 indicating the collapse of central authority in that polity. These case studies are largely qualitative, detailing the political history of these state-societal systems from 1946 to 2015, paying particular attention to periods of factionalism and accounting for all events recorded as governance changes in the Polity IV dataset and in the set of country reports prepared during our extensive review of Polity IV. Most of these country reports were completed by 2010, and thus are missing developments between then and 2015, and some country reports missed events that were later captured in changes to the Polity IV record (sometimes as a result of the review project). I thus have updated the record for these cases through 2015 and where required in the historical record. In the case of Thailand, its country report was prepared as part of the review project’s pilot study before review and report-writing procedures were standardized, particularly with regard to the method of identifying factional identities and onset events; I substantially overhauled this and similar reports. I prepared these updates and overhauls using the same methodology as the original reports, relying primarily on media coverage of news events as compiled and recorded in news archives, in particular Keesing’s World News Archive and the BBC Monitoring: International Reports news archive available through LexisNexis, and for election results Adam Carr’s Psephos archive (Carr 2017; Keesing’s 2017; BBC Monitoring 2017).17 This methodology is described in more detail in Chapter 2, and the Polity IV review and country report−writing process is described in detail in Monty G. Marshall and Benjamin R. Cole (2012). I rereviewed all country reports for the 14 cases included as part of this analysis. In Chapter 7, I synthesize results from Chapters 3 through 6 and consider the MSS comparisons across those chapters, looking for evidence of smart practices for countries embarking on democratic transition that try to avoid or mitigate the effects of factionalism. Of the many circumstances and practices associated with factionalism episodes, I identified five factors from the comparative analyses as particularly important in explaining divergent factionalism outcomes: 1. Gradual or guided democratization experiences may allow states with favorable conditions to avoid factionalism altogether.

24

Democratization and the Mischief of Faction

2. Guaranteeing a substantial role for opposition elites inside the new political system incentivizes conventional political participation and disincentivizes disruptive and antisystem behavior, undermining factionalism. 3. Individual leaders and their choices matter. 4. Pressure from the international community can change the incentive structures underlying policy choices by leadership, encouraging more liberal and inclusive policies that expand participation and undermine factionalism. 5. And states that enter factionalism episodes with professional, rather than politicized, militaries are much more likely to survive factionalism without political violence and democratize. In Chapter 7, I compare these findings to the existing democratization literature and the macrocomparative analysis of Chapter 2, which points toward directions for future quantitative research of factionalism episodes. As the case of Belgium (described in a brief vignette in Chapter 2) illustrates, these findings are also relevant for the “old” democracies in the world. Chapter 7 concludes by considering what these findings mean for countries like the United States, which was coded factional in November 2016 for only the second time since the American Civil War.

Notes 1. This team was led by Monty G. Marshall, director of Polity IV and the Center for Systemic Peace, and in addition to me included Center for Systemic Peace researchers Min Zaw Oo, Eliot Elzinga, and Gabrielle Elzinga-Marshall. This research (and much of the work for this book) was funded by Societal-Systems Research Inc., with support from the US Political Instability Task Force. 2. These include analyses of interactions between factionalism and democratic transition, authoritarian (pretransition) regime type, youth bulge and demographic dynamics, homicide rates, inequality, and state failure. See Marshall and Cole (2008, 2011, 2012, 2014, 2017), Cole (2013, 2014, 2016), Cole and Blume (2014a, 2014b), as well as a piece in the Harvard International Review by Marshall (2011) and an innovative and ambitious videobook project by Marshall and the Center for Systemic Peace (Marshall 2014). 3. Polity was developed by Ted Robert Gurr and Monty G. Marshall, and its current version, Polity IV, has become one of the most commonly used datasets for differentiating systems of governance in the political science literature, covering all countries with more than 500,000 people back to 1800 or independence. The dataset has been reviewed extensively, and is generally praised for its intercoder reliability, parsimonious design, and high degree of cross-national reliability (Munck and

Democratization and the Mischief of Faction

25

Verkuilen 2002). Like most large-n datasets describing governance patterns, however, Polity IV preferences reliability over validity, and as such has been subject to criticism by country experts, frequently in cases of borderline coding decisions. Polity is also relatively poor at distinguishing among levels and styles of democratic governance among mature democracies, in part due to its creation in 1973 when autocratic governments were the norm and distinguishing among democratic governments was a low priority for researchers. See the Polity IV handbook (Marshall, Jaggers, and Gurr 2010) for more detail on the Polity IV coding scheme and practices, and Munck and Verkuilen (2002) and Marshall and Jaggers (2002) for a review of Polity in the context of alternative measurement schemes. 4. Countries coded factional (PARCOMP = 3) are nearly always also coded on the regulation of participation (PARREG) variable as either PARREG = 2 (multiple identity politics) or PARREG = 3 (sectarian politics). The PARREG = 2 multiple identity coding describes competition characterized by “relatively stable and enduring political groups which compete for political influence at the national level . . . but there are few recognized overlapping interests” (Marshall, Jaggers, and Gurr 2010: 30). The PARREG = 3 sectarian politics coding describes a landscape where “political demands are characterized by incompatible interests and intransigent posturing . . . when one identity group secures central power it favors group members in central allocations and restricts competing groups’ political activities . . . [or where] political groups are based on restricted membership and significant portions of the population historically have been excluded” (Marshall, Jaggers, and Gurr 2010: 30). PARCOMP and PARREG combine to form the component variable political competition (POLCOMP) variable, a 10-point ordinal scale of political competition ranging from 1 (suppressed competition) to 10 (institutional electoral competition). States coded with PARCOMP = 3 (factional) and PARREG = 2 (multiple identity) are described as POLCOMP = 6 (factional/restricted), and states coded with PARCOMP = 3 (factional) and PARREG = 2 (sectarian) are described as POLCOMP = 7 (factional). Because these sets of codes nearly always occur together, and these combinations are unique, beginning Polity coders often code PARCOMP and PARREG simultaneously using the POLCOMP variable, which was introduced in 2004 with the rollout of Polity IV, rather than coding its indicators independently. 5. For a brief overview of metastability in complex social systems, see CioffiRevilla (2014: 167−168), Moffat (2003), or Winder (2007). 6. See Marshall and Cole (2017) for a comprehensive review of state failure and state fragility research, including an extensive discussion of changes in preferred terminology and conceptual definitions in the literature over time. 7. These figures are accurate as of the 2015 edition of Polity IV. Note that historical revisions by the Polity coding team may result in fluctuations to these numbers, as Polity IV is a living dataset. 8. Banfield’s Moral Basis of a Backward Society (1958) was one of the first to pose a link between culture or values and democratic governance and economic development, and his work in Italy was followed by many others. Following in his line of reasoning, scholars have pointed to Roman Catholicism (Wiarda 1972; Lipset 1991), Islam (Inglehart 2003; Lafoff 2004; Norris and Inglehart 2006), Arab ethnicity (Stepan and Robertson 2004), and Confucianism (Fukuyama 1995) as being incompatible with democracy, often in contrast to Protestantism. These arguments hold that because the population of a state is predominantly Catholic, Muslim, or Confucian, and because these religions are structured hierarchically (or, in the case of Confucianism, espousing patriarchal values), its people will be more likely to accept similar authoritarian rule by government. Protestantism in Europe,

26

Democratization and the Mischief of Faction

it is argued, places value on individualism and secular government, and features a more democratic structure with an equitable relationship between pastor and congregation. Furthermore, the assumption that people will submit to authoritarianism because they were raised in patriarchal religions is unwarranted and generally left undefended. The logic that a person, let alone an entire population, would be willing to accept repression and human rights violations because their religion is governed hierarchically is weak at best. Moreover, while it is possible that Latin American democracies have struggled because of their Roman Catholic heritage, it seems more likely that they have struggled because of repeated US interventions during the Cold War, a history of tension between indigenous peoples and European imperialism, economies heavy on resource extraction and drug production, and extreme sustained income inequality. Indeed, it was not uncommon for Catholic religious leaders in Latin America to lead the fight against repressive governments during the 1980s. Similarly, the link between Islam and autocracy has been difficult to establish empirically because of confounding factors, including economic dependence on oil exports, historical tensions between religious and ethnic groups in the Middle East (often initiated and maintained by European powers), the overlap of Arab and Muslim populations, and the relatively successful transitions of Muslim democracies in Southeast Asia. This critique of cultural determinism does not mean that culture is unimportant and, indeed, some aspects of culture are critical to democratic development such as gender roles. For example, if a society is traditionally patriarchal and gender biased, implementing universal suffrage may disrupt cultural norms significantly. 9. The relationship between development and democratization was one of the first recognized in the literature, and remains one of the most contested. Lipset (1959) was one of the first to articulate a theoretical explanation for the strong correlation between wealth and democratization, which has been repeatedly confirmed in the literature (see, e.g., Lipset 1994; Burkhart and Lewis-Beck 1994; Moore 1995; Przeworski and Limongi 1997). This phenomenon has many different explanations, but most focus on the role of economic development in changing the values of the population, increasing the size of the middle class, and encouraging urban migration. Recent works have refined Lipset’s analysis, discovering that the correlation is not linear but rather there exists a transition zone, within which transitions are the most likely and above which increases in wealth do not translate into likelihood of transition. Some, such as Przeworski and Limongi, have argued that this is because the increased wealth acts as a stabilizing force once the transition zone is passed, allowing authoritarian regimes that survived to endure (Przeworski and Limongi 1997). That economic development leads to changes in class structure and consciousness, demographics, and even popular and elite values, is uncontested. These changes are critical for a society, and particularly societal elites, to accept democratic institutions but it has been extremely rare for successful democratic transitions to have been caused by concerted class action. Importantly, many middle-income countries during the third wave experienced the changes in social structure that modernization theorists predict (indeed, China is doing so now), without concurrent transitions to democratic governance. The populations of South Korea and Taiwan, for example, certainly became more accepting of democratic norms during their economic development periods in the 1970s and 1980s, but it took some other agency to cause leaders to pursue democratic transition. Changes in social structure cannot themselves be responsible for the decision to implement democratic institutions. 10. States with economies based on resource extraction, and oil in particular, struggle with democratization and with maintaining democratic institutions, although

Democratization and the Mischief of Faction

27

a few highly developed mature democracies pose exceptions to this rule such as Norway, Canada, and the United Kingdom. See, for example, M. L. Ross (2001). 11. Various models have been proposed focusing on behavior by political and economic elites to explain divergent democratization outcomes. Among the more important of these have focused on elite pact formation, most notably the seminal Transitions from Authoritarian Rule anthology (O’Donnell, Schmitter, and Whitehead 1986), which was refined and extended in a rich body of more recent scholarship (see, e.g., Hagopian 1990; Karl 1990; Threlfall 2008; Hinnebusch 2015). 12. Recent work using public opinion surveys has been particularly insightful in this regard, finding distinct cross-cultural conceptualizations of terms like democracy or liberal, which in turn may affect democratization potential. See, for example, a special issue (no. 4) in the 2010 volume of Journal of Democracy for a strong set of case studies in this regard (Braizat 2010; Bratton 2010; Chu and Huang 2010). 13. Some have found regional effects to be quite profound on democratic transition probability, as well as instability, war, and other outcomes, although a weak theoretical basis and concerns about spuriousness have led to these findings being contested (see, e.g., Cederman, Hug, and Wegner 2008; Tolstrup 2009; Freyburg et al. 2011). 14. International pressure on autocratic governments to democratize and international support of democratizing states are both known to correlate with democratic transition probability, although major actors that promote democratization abroad, such as the United States, also have a long history of undermining democracy, particularly during the Cold War era. Scholars disagree about what types of aid, pressure, or both work better than others; however, distinguishing between nongovernmental organization−based aid (Landolt 2007; Kamstra and Knippenberg 2014), official development or democratization aid from states (Bader 2010), negative reinforcement (Hovdenak 2009; Wobig 2015), coerced or internationally managed transitions (Nenadovic 2010; Lemay-Hebert 2012), and international democracy promotion regimes (Ulfeder 2008; Legler and Tieku 2010). The journal Democratization has also dedicated several special issues to these topics; see its 2010 volume, issue 6, and 2012 volume, issue 3, for example. Studies have also found regionally distinct patterns with regard to democracy promotion and democratization outcomes. See, for example, McCoy (2006) on Latin America, Bader (2010) on Eastern Europe, and Cavatorta (2005) and Abbott (2018) on the Middle East and North Africa. 15. The key debate in this area has been between supporters of sequencing, the idea that democratic institutions must be established, or at least taking root, before mass participation and electoral competition is encouraged, and supporters of gradualism who argue that necessary institutions can be developed at the same time that people are learning to participate in elections. Leading the sequencing argument in recent years have been Edward D. Mansfield and Jack Snyder, who built on the work of Dankwart A. Rustow (Rustow 1970; Mansfield and Snyder 1995, 2005, 2007). They argue that a strong legal system, representative government institutions, and a free press are critical preconditions for democratic transition, often but not necessarily established by liberal-minded autocrats, sometimes through pact formation with moderate opposition groups, and with international support. Once institutions are either established or taking root, depending on one’s interpretation of the sequencing argument, the authoritarian government will use these institutions to liberalize, by opening up some areas for public debate and guaranteeing certain rights (e.g., habeas corpus, press freedom, limited civil liberties), consistent with the ideas of O’Donnell, Schmitter, and Whitehead (1986) and Huntington (1991). Importantly, sequencing scholars argue that premature transitions (i.e., those initiated without established impartial institutions) result in “birth defects” in

28

Democratization and the Mischief of Faction

institutions that haunt the democracy for the rest of its life, and which actually decrease the likelihood of another transition to occur successfully. However, even within Western Europe smooth sequencing−based transitions have been rare, as Berman (2007b) and Carothers (2007b) note. Many Western European democracies achieved stability only after significant conflict, false starts, and temporary periods of autocratic backsliding. The United States, Great Britain, and Spain all suffered bloody civil wars after attempts at democratic transition, before they achieved stability (and universal suffrage many years later). Similarly, France’s democracy did not stabilize until the establishment of the Fifth Republic in 1958, 160 years after the French Revolution. Even South Korea, an oft-cited example of the sequencing story, had a failed democratic experiment in the Second Republic, leading to Park Chung-hee’s dictatorship, apparently without damning the now-democratic Sixth Republic with birth defects. In contrast, gradualists, led by Sheri Berman and Thomas Carothers, argue that the vulnerability associated with democratic transition is constant and universal, regardless of the sequence in which the transition occurs (Berman 2007a, 2007b; Carothers 2007a, 2007b). Institution building should occur simultaneously with encouragement of mass participation and competitive elections. Preconditions that are supportive of democracy (i.e., economic development, rule of law) are great, but they are not necessary for the transition to be successful. The greatest flaw in the gradualist argument is a tendency to ignore Snyder and Mansfield’s arguments about the destabilizing effects of competitive elections. Without agreement on the rules of the game, elections can encourage violence and create incentives for government manipulation of the system. Without preexisting institutions, respected by all parties, to oversee and judge the election, there can be no guarantee of fairness and thus no incentive to participate fairly. The victors in a democracy must also be willing to respect the rights of the minority and govern the entire population, a norm that seems unlikely to be developed immediately, especially in the tumultuous and often divisive atmosphere surrounding the disintegration of an autocratic regime and the formation of new institutions. Gradualism and sequencing claim some empirical support, but both theories also have serious flaws and the arguments of each seem to capture only a small part of the democratic transition story. Importantly, both generally ignore civil society, focusing instead on institutional development (i.e., rule of law, free press, representative institutions) and competitive elections as critical factors. While sequencing relies on the role of the liberal autocrat and a moderate opposition, and gradualism the radical democratic reformer, neither of these groups can consistently be relied on to bring about transition. Philosopher kings are rare, opposition to a repressive regime is rarely moderate if it exists openly, and radical democrats are just as likely to bring about their own destruction as they are a successful revolution. While Frederik Willem de Klerk and the moderate leaders of the Africa National Congress were busy negotiating a democratic transition in South Africa, radical reformers in the Inkatha Freedom Party, Pan-African Congress, and reactionaries in the Conservative Party nearly brought the country into civil war. 16. For example, O’Donnell, Schmitter, and Whitehead (1986), Haggard and Kaufman (1995), Linz and Stepan (1996), and Geddes (1999) have developed and refined theories of democratization based on the nature of the autocratic regime type prior to the onset of transition, finding that party-based autocratic regimes are more likely to endure. Building on this work, Marshall and Cole (2008) found that partybased autocratic regimes are also more likely to withstand factionalism during democratic transition without falling into civil war.

Democratization and the Mischief of Faction

29

17. The search term for Keesing’s World News Archive is simply the country name, restricted by the study period plus 2 years before and after the study period, due to compilations in Keesing’s sometimes appearing months after (and sometimes before) an event occurs; hits for relevant years are then reviewed based on headline relevance. The search term set for LexisNexis, once in the BBC Monitoring: International Reports search engine, is the following: {(country_name) and [(#STX000160#) OR (#N921190MM#) OR (#STX000563#) OR (#STX001917#) OR (#STX000239#) OR (#ST000CTCO#) OR (#STX001129#) OR (#STX001371#) OR (#STX001377#) OR (#STX001510#) OR (#ST0009ZGJ#) OR (#ST00097CY#) OR (#STX001812#) OR (#ST000CQJE#) OR (#STX001345#) OR (#N813940MM#) OR (#ST000D33D#) OR (#ST000CTIY#) OR (#STX000459#) OR (#STX001136#) OR (#N922120QM#) OR (#ST0008X9C#) OR (#ST0009EOQ#) OR (#ST0009ZVD#) OR (#N921120CC#) OR (#ST0009G89#) OR (#N813311MM#) OR (#STX001218#) OR (#STX001220#) OR (#N813940GM#) OR (#ST0009FAF#) OR (#STX000271#) OR (#ST000CUW5#) OR (#STX000302#) OR (#STX001149#) OR (#ST000CYCH#) OR (#ST000D170#) OR (#STX000827#) OR (#ST000CUTM#) OR (#STX001498#) OR (#STX001566#) OR (#STX000119#) OR (#ST000CNTG#) OR (#STX000420#) OR (#STX000430#) OR (#ST0008WX0#) OR (#STX001304#) OR (#STX001456#) OR (#STX001551#)] and Date[geq(XX/XX/XXXX)]}.

2 Factionalism in the Post–World War II World

There were 172 discrete episodes of factionalism from 1946 to 2015, in 107 states, meaning that factionalism was not only present, but openly manifest as opposed to latent, in those cases. In most cases, the factionalism condition emerged immediately following a major act of liberalization such as ending martial law, legalizing political parties, or promulgating a new (liberal) constitution. In this chapter I look at factionalism throughout the world during the postwar period, searching for common types of factionalism cleavages, factionalism onset types, and factionalism ending events. Thus, I describe all 172 episodes in broad strokes in this chapter, primarily in tables organized by region, which include the events, faction types, and beginning and ending dates of factionalism episodes. This allows a macrocomparative comparison among factional cases, and a global view of the factionalism condition. The research underlying this chapter and, indeed, the book, was driven by the need for greater context for understanding the factionalism phenomenon, which necessitated a comprehensive review of the Polity data series. Every case was reviewed, resulting either in alteration or confirmation of beginning and ending date events of every change in governance, paying particular attention to the factionalism episodes and governance changes within factionalism episodes. Note, however, that the notion of any fixed beginning date or ending date of factionalism reflects an artifact of coding rather than any claim of a “true” beginning or ending point. Factions often form and become active before the inauguration, election, or constitutional reforms identified in the Polity scheme as the beginning of an openly factional 31

32

Democratization and the Mischief of Faction

period, for example, and factions may remain active for a time after martial law is imposed or after consolidation causes factionalism to be resolved, which is usually a long-term process. Moreover, these events are often representative of other changes happening at about the same time, and thus are to a degree arbitrarily, although rigorously and systematically, identified. For example, in the example of Belgium (see the Factional Episodes in Europe section below), Polity identifies a 1971 constitutional reform package, which institutionalized linguistic identity groups into the political system, as the beginning date of a major change in the Belgian political system. In practice, however, that constitutional reform was the culmination of changes to political competition driven by a series of structural changes regarding linguistic identity groups that began with legislative reforms 3 years earlier. The review project produced a report for every country experiencing factionalism in the study period (and many countries that did not) that described the events in full context, often beginning long before the study period and following events through to the present day. Thus, I have high confidence that the beginning date and ending date events, and their resulting Polity codes, are valid and representative of structural or societal changes driving changes in political competition or regime type at the time, even while admitting that any beginning date or ending date of a regime or factionalism episode is necessarily artificial. These reports also produced descriptions of the factions themselves that I describe in the Analysis of Faction and Event Types section of this chapter. Their descriptions required navigating the identity traits of groups and combinations, alignments, and alliances among discrete groups, which in the historical record are often difficult to positively identify, tend to be intersectional, and are variable in persistence. These 172 episodes of factionalism were nearly all isolated to the developing world during this period, although Polity records factionalism at some point in the histories of nearly every country older than 50 years. Regional differences are notable. During the 1946−2015 study period, factionalism was the most common in Latin America, affecting all but three countries (Costa Rica, Jamaica, Uruguay). SubSaharan Africa had the highest rate of ongoing factionalism episodes at the end of 2015, and its factions were more likely to be personalistic or clientelistic than in other regions. Factionalism episodes in Asia and Latin America tended to be consistently more complex than in other regions, with more discrete episodes of open factionalism and more changes in regime during often protracted factional episodes, than in other regions. During the study period, one “old” industrial-

Factionalism in the Post–World War II World

33

ized democracy (Belgium) experienced factionalism, demonstrating that such states are not immune from factionalism.1 Beyond regional differences, factionalism is most common among anocratic regimes (i.e., regimes with Polity scores between −5 and 5), consistent with the social-systems process (SSP) model of factionalism as being most likely to manifest following liberalization by an autocratic regime. Such liberalization often results in a mixture of autocratic and democratic practices and institutions, characterized in the Polity scheme as anocratic governance. Two-thirds of states that were factional during the 1946−2015 period were anocratic during at least some portion of their factionalism episode(s). Consolidated autocracies on the Polity scale (i.e., regimes with Polity scores between −8 and −10) are by definition unable to be (openly) factional, as the same repressive practices that allow a state to be autocratic also repress the open polarization that characterizes factionalism. Interestingly, the condition is rare even among weak autocracies (i.e., regimes with Polity scores of −6 to −7) with only two factionalism episodes (North Yemen, 1948−1962; Nepal, 2002−2006) occurring under such regimes. The conditions that allow factionalism to manifest are uncommon in autocratic states, largely because such conditions reflect a degree of liberalization inconsistent with autocratic governance and with Polity’s coding scheme for autocracies. Among democracies, however, the factionalism condition is not uncommon: 23 percent of factionalism episodes are in countries that meet narrow definitions of democracy but are not consolidated (i.e., weak democracies, or regimes with Polity scores of 6 to 7), and 10 percent of cases are in nearly consolidated democracies with Polity scores of 8. These factional democracies sometimes exhibit rightward migration along the SSP model, as in cases such as Belgium, where a cyclical pattern can set in: opposition and government behaviors increase polarization and damage democratic institutions, which in turn are less capable of managing polarization and unconventional political behavior. More often, however, factional democracies are formerly autocratic or anocratic states that are still liberalizing, with electoral systems for recruiting executives but the executives face few or no constraints. Other states in this condition, particularly among those factional democracies closer to 10 on the Polity scale, have adopted and implemented electoral systems for executive recruitment and do implement substantial constraints on the chief executive, but political competition and regulation continue to utilize the disruptive and coercive tactics consistent with factionalism. Tables 2.2−2.6 describe all factionalism episodes during the study period, by region. Each unique onset is consecutively numbered using

34

Democratization and the Mischief of Faction

integers (e.g., 1, 2), with changes in governance within a factionalism episode shaded gray and consecutively lettered under the numbered episode (e.g., 1a, 1b). While recorded in the data as discrete episodes, subsequent factionalism episodes after the first sometimes feature similar factional identities, reflecting the tendency of factions to endure and remain latent when repressed by the state. When liberalization reoccurs, the same factions, sometimes reinforced and strengthened by their period in the metaphorical wilderness, reemerge. In other cases, however, new episodes feature new factions. In addition to the beginning and ending dates of these episodes, Tables 2.2–2.6 also include factional identity characteristics (primary, secondary, tertiary), which I identified in each episode based on the methodology described in the Analysis of Faction and Event Types section of this chapter. These types are described in Tables 2.2–2.6 using letters, such that S stands for social identity-based factions (e.g., ethnic, religious, linguistic); E represents economic factions; P represents personalistic or clientelistic factions; I represents issue- or policy-based factions; and G represents geographic or regional factional cleavages. Tables 2.2–2.6 also include onset event types (e.g., elections, constitutional reform), and ending event types (e.g., coup d’état, martial law); Table 2.1 provides a key for reading these values. Tables 2.2−2.6 also offer the Polity IV combined Polity score (POLITY), which measures regime type characteristics on a scale ranging from −10 to 10, where −10 to −6 are generally considered autocracies, −5 to 5 are anocracies, and 6 to 10 are democracies. While the

Table 2.1 Factionalism Onset and Ending Event Code Descriptions Onset Event Type

1. Elections 2. Party legalization 3. Constitutional reform or new constitution 4. Independence 5. New transitional government 6. New government 7. Mass political action 8. General liberalization 9. Collapse of central authority 10. Withdrawal of occupation forces 11. Other

Ending Event Type

1. Elections 2. Armed conflict 3. Constitutional reform or new constitution 4. Coup 5. New transitional government 6. New government 7. General autocratization 8. General liberalization 9. Ban on political parties 10. Secession, fragmentation, or state change 11. Peace/reconciliation agreement 99. Ongoing

Factionalism in the Post–World War II World

35

POLITY score is commonly used as a continuous variable in the literature, this is inappropriate because of the nature of Polity’s components and additive structure, which create inconsistent conceptual gaps between values on the scale. The only difference between a −9 and a −10, for example, is whether the executive is selected through selfselection or ascription, which few would argue has any substantive bearing on measuring levels of democracy or autocracy in the statesocietal system. The difference between 9 and 10, on the other hand, is substantial, as a democracy with a score of 9 may have problems in the electoral system (e.g., widespread fraud), weaker constraints on the executive, or more coercive or restrictive regulation of political competition. Also, the POLITY variable obfuscates one of Polity IV’s most innovative conceptual developments, which is that it assumes that democratic and autocratic features occur simultaneously in most states, and that each deserve careful study. The POLITY variable is included here despite these weaknesses because of its widespread familiarity in the discipline and ease of use. Polity component (executive constraints [EXCONST], political competition [POLCOMP], and executive recruitment [EXREC]) scores are also included to allow finer-grained comparison of institutions and structures than POLITY alone offers. While all three are categorical variables, EXCONST and POLCOMP have ordinal characteristics. EXCONST ranges from 1 (unlimited executive authority) to 7 (executive parity or subordination) while POLCOMP ranges from 1 (suppressed competition) to 10 (institutional electoral competition). EXREC is strictly categorical, using the following codes: (1) ascription; (2) dual executive: designation + ascription; (3) designation; (4) self-selection; (5) gradual transition from self-selection; (6) dual executive: ascription and election; (7) transitional or restricted election; and (8) competitive election. In all four Polity indicators, −66, −77, and −88 codes indicate periods of interregnum: −66, foreign occupation; −77, collapse of central authority; and −88, transitional government. Factionalism Episodes in Sub-Saharan Africa Table 2.2 describes all factionalism episodes in sub-Saharan Africa since 1946, including unique factionalism onsets and changes in governance within a factionalism episode. Of the five continental “regions,” sub-Saharan Africa had the highest rate of ongoing factionalism episodes at the end of 2015, with more than one-third of states that have experienced factionalism since 1946 still

36

Angola Benin

Burkina Faso Burundi

1 1 2 1 1 1a 2 3 1

Cameroon Central African Republic 1 1a 1b 1c Chad 1 Comoros 1 1a 1b 1c 1d Congo 1 2 Djibouti 1 Equatorial Guinea 1 Ethiopia 1 1a Gabon 1 Ghana 1 2 Guinea 1 2 2a Guinea-Bissau 1 2 2a 2b 3 Côte d’Ivoire 1 2 Kenya 1 1a Liberia 1 Lesotho 1 2

4 8 5 5 7 6 6 7 3

8 3 5 3 6 3 9 3 5 8 8 3 4

8 8 9 10

1 11 7 12 9 7 1 9 10 4 10 4 10 4 8 3 6

15 1 2 30 1 16 5 15 2

23 15 9 25 3 21 16 15 1 15 15 16 10

12 15 5 16

2 4 13 21 28 4 23 14 1 13 26 11 23 18 2 28 5

1997 1960 1970 1978 1962 1963 1998 2015 1992

1993 2003 2005 2013 1996 1990 1995 1996 1999 2000 1960 1992 1999

1968 1995 2005 2009

6 4 5 1 4

5 1 1

1

1 1

4 3 1

4 1 6

1979 2 1992 1 1995 1 2010 6 2013 1994 2 2000 1 2003 2005 2012 11 2000 6 2011 1 2007 1 2008 1997 1 1993 1 2002 1

8 10 10 11 6 10 10 99 99

3 5 3 99 99 9 3 4 8 12 12 10 2

2 9 99 99

12 12 2 9 99 6 9 9 9 6 9 99 4 8 8 5 5

22 28 26 25 15 18 31 99 99

14 8 24 99 99 15 14 30 14 25 8 15 21

28 4 99 99

31 8 14 27 99 7 13 30 7 22 19 99 17 3 11 15 25

2002 8 1963 7 1972 4 1980 7 1963 7 1965 2001 5 9999 7 9999 99

2003 99 2005 2013 9999 9999 99 1995 4 1996 1999 2000 2001 1963 4 1997 4 2013 6

1969 4 2005 1 9999 9999 99

1981 4 1996 1 2010 5 2013 99 9999 1998 1 2003 4 2005 2009 2014 1 2002 2 9999 99 2008 3 2010 2003 5 1998 5 2012 8

S P P S S S P P

S

P

S

P P

S S

P

S

S P P

I S E

S S

E

S P P P

E S

P

P P

P S S S

P E P

G G S P I

I

POLCOMP

EXREC

POLITY

Tertiary Type

G S

EXCONST

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.2 Factionalism Cases in Sub-Saharan Africa, 1946–2015

−3 2 −2 5 0 −3 −1 −1 −4

5 −1 −1 −77 −2 4 −77 4 −2 −1 E 4 5 2 2 1 −3 3

6 −1 −1 1 4 5 5 −1 6 1 4 4 7 7 0 8 8

3 8 3 8 2 2 4 3 3

3 3 3 5 5 3 3 4 2

6 6 7 6 7 7 6 6 6

7 7 3 7

3 3 3 4

7 6 6 6

7 5 7 4 2 7 5 3 6 −77 −77 −77 5 2 6 7 5 6 −77 −77 −77 7 5 6 4 1 7 4 2 7 8 4 6 8 5 6 7 3 7

8 5 5 7 7 8 7 4 7 5 7 7 8 7 7 8 8

5 3 3 3 4 4 5 2 6 4 5 5 6 7 2 7 7

7 6 6 6 7 7 7 7 7 6 6 6 7 7 6 7 7

continued

37

Madagascar Malawi Mali Niger Nigeria Rwanda Sierra Leone Sudan Swaziland Tanzania Togo Uganda Zambia

Zimbabwe

1 2 3 1 1 1 1 1a 2 1 1 2 3 1 2 3 4 1 1 1 1a 1b 1 2 1 1a 2 2a 2b 1 2 2a 2b 2c

5 3 2 2 5 12 10 12 10 7 4 3 7 1 4 4 4 9 10 8 2 5 12 7 10 8 11 11 12 4 9 7 2 5

19 18 18 6 16 27 1 9 2 1 27 17 12 1 22 2 11 6 25 26 6 28 16 28 24 15 1 19 28 19 11 27 11 22

1972 7 2009 6 2014 6 2005 11 1997 1 1992 3 1960 6 1964 1979 3 1961 6 1961 6 1996 1 2002 6 1956 4 1965 1 1986 1 2010 1 1968 6 2015 6 1993 1 2005 2010 1980 1 2005 3 1964 6 1968 1991 1 1996 2001 1980 4 1999 2 2001 2009 2013

2 11 99 99 6 1 12 1 1 7 3 5 9 11 5 6 7 4 99 2 5 99 7 99 8 12 11 12 10 3 7 2 5 99

4 20 99 99 7 28 8 15 1 5 23 25 16 17 25 30 8 12 99 5 27 99 29 99 14 8 18 27 29 8 26 10 21 99

1975 2011 9999 9999 2002 1996 1964 1966 1984 1973 1967 1997 2007 1958 1969 1989 2011 1973 9999 2005 2010 9999 1985 9999 1968 1972 1996 2001 2008 1983 2001 2009 2013 9999

7 6 99 99 1 4 1

S P P P E S S

4 4 4 4 6 4 4 4 10 4 99 99

S S P G G G S S G P P P

4 99 9

P P S

2 99

S P

1

P

S G G S S S S

S

POLCOMP

EXREC

POLITY

Tertiary Type

I

EXCONST

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.2 Continued

−3 0 6 6 6 8 8 7 7 −5 6 4 5 8 7 7 −2 0 3 −2 −4 −2 3 −1 2 0 6 1 5 4 −3 −4 1 4

3 3 7 7 8 8 8 8 8 3 8 7 7 8 8 8 4 6 7 5 3 3 7 5 8 7 8 3 7 7 3 3 7 7

3 5 6 6 5 7 7 7 7 1 5 5 5 7 7 7 2 2 4 2 2 3 4 3 3 2 5 5 5 5 3 2 3 5

6 6 7 7 7 7 7 6 6 6 7 6 7 7 6 6 6 6 6 6 6 7 6 6 6 6 7 7 7 6 6 6 6 6

Note: POLITY, combined Polity score; EXREC, executive recruitment; EXCONST, executive constraints; POLCOMP, political competition. Under “Faction Type” columns, “S” stands for social identitybased factions, “E” represents economic factions; “P” represents personalistic or clientelistic factions; “I” represents issue- or policy-based factions; “G” represents geographic or regional factional cleavages. Bolded rows indicate changes in governance affecting the Polity scheme within factionalism episodes.

38

Democratization and the Mischief of Faction

experiencing it on an ongoing basis. At the end of 2015, this included Burundi, Cameroon, Central African Republic (in a condition of state collapse), Ethiopia, Gabon, Côte d’Ivoire, Madagascar, Malawi, Tanzania, Togo, Uganda, and Zimbabwe, all states that have struggled to liberalize from autocratic regimes. This includes one-fifth of all states in the region, and reflects the large number of anocracies on the continent as well as the stalled liberalization that has become characteristic of sub-Saharan African regime transitions. While most of the factionalism episodes that are ongoing began fairly recently, some of these states (e.g., Ethiopia and Zimbabwe) have been (openly) factional for 20 years or more and seem to be stuck under an anocratic regime. Their leaders and institutions have been unwilling or unable to implement additional liberalization needed to dissipate factionalism, but also unwilling or unable to reinstitute martial law and repress the open factional behavior that is devastating those countries’ political, economic, and social institutions. Burundi’s case is illustrative of how to read Tables 2.2–2.6, as it includes unique onsets of factionalism and changes in governance inside these episodes and shares the experiences of many state-societal systems in sub-Saharan Africa. Burundi has experienced three discrete onsets of (open) factionalism since independence—in 1962, 1998, and 2015. Burundi’s first factionalism episode (episode 1) began at independence, with the country intensely polarized along economic and ethnic lines as the ruling (minority) Tutsi elites backing the Tutsi monarchy relaxed restrictions on the majority, generally poor, Hutu ethnic group. As shown in the shaded line in Table 2.2, on June 16, 1963 the state experienced a significant change in governance while remaining factional (episode 1a). In that case, the Tutsi monarch appointed Prime Minister Pierre Ngendeandumwe (an ethnic Hutu), who in turn appointed a cabinet dominated by Hutus, a development that effectively eliminated legislative constraints on executive authority but did not change executive recruitment methods or political competition. The factionalism episode that began in 1962 ended on October 18, 1965, as a Hutu-led coup attempt on October 19 resulted in the deposition of the government, seizure of power by the Tutsi-dominated military, and creation of a one-party state under the Tutsi nationalist party, Unity for National Progress (UPRONA), whose repressive practices drove factionalism underground. Between the end of this factional episode and the next, the state would undergo severe political instability, genocide, ethnic war, a new constitution (1992), and multiple coups and coup attempts. The state’s second discrete episode of (open) factionalism began on June 5, 1998, when President Pierre Buyoya, an ethnic Tutsi who had

Factionalism in the Post–World War II World

39

seized power in a coup in 1996 (and, earlier, in 1987), signed the Transitional Constitutional Act in cooperation with Leonce Ngendakumana, speaker of the National Assembly and a Hutu. This act partially restored the country’s 1992 constitution that Buyoya had abrogated on seizing power in 1996, beginning a liberalization process that allowed factionalism to manifest. This factionalism episode concluded in 2001 with the inauguration of a new transitional government featuring a power-sharing agreement that put an ethnic Hutu, Domitien Ndayizeye, into the presidency. The country’s third discrete episode began in mid-2015, when President Pierre Nkurrunziza, who had been elected in free, fair, and peaceful elections in 2005 and reelected in 2010, announced that he planned to seek an unconstitutional third term in office. Nkurrunziza’s announcement led to widespread protest, a May coup attempt, election delays, and a boycott of the July 2015 election by all major opposition parties. This return to disruptive and coercive behavior—a form of rightward migration in the SSP model—by both the state and opposition political parties reflected a new open manifestation of factionalism concomitant with Nkurrunziza’s autogolpe, characterized by changes in executive recruitment and executive constraints. This factionalism episode (episode 3) was ongoing at the end of 2015, indicated in Table 2.2 by 99 under ending event and the ending date listed as 99/99/9999 for that factional episode. Burundi’s case is also representative of the many countries that have experienced multiple factionalism episodes, in that factional identities tend to remain consistent over time, with the potential to change depending on government policies and bevhavior. In Burundi’s first two episodes of factionalism, the primary cleavage separating government from opposition was ethnicity, specifically along Hutu/Tutsi lines. This reflected the historical power differential between the two groups. When factionalism emerged in the 1990s after being dormant for decades (episode 2 in Table 2.2), factions split largely along this historical division of power with a Hutu majority opposing the minority Tutsi government, although in that case two major Hutu groups, Forces for the Defense of Democracy (FDD) and National Liberation Forces (FNL) remained outside of the polity engaged in armed conflict against the government until 2005. While factional identities had been consistent until this point, conflict between the FDD and FNL indicated brewing salient divisions within the groups comprising the Hutu political majority. Interestingly, the third episode of factionalism in Burundi did not split along the Hutu/Tutsi lines that dominated prior episodes, but instead was intensely personalistic; it was oriented around President Nkurunziza’s decision to mount a third presidential campaign, largely

40

Democratization and the Mischief of Faction

seen as unconstitutional. This is reflected most clearly in the alliance between Hutu and Tutsi opposition parties that emerged in the midst of the May and June 2015 unrest in the country, as the salient question dividing factions became less about ethnicity than about the selfaggrandizing actions of the country’s leadership. Factionalism Episodes in the Middle East and North Africa Table 2.3 describes the factionalism episodes that have occurred in the Middle East and North Africa (MENA) since 1946, including 14 countries and 20 discrete onsets of factionalism, which covers a substantial majority of the countries in the region. Indeed, the only countries that did not experience a factionalism episode during this period were Kuwait, Libya, Oman, Qatar, Saudi Arabia, and the United Arab Emirates, which have had in common consistently repressive practices over the past 70 years. Five episodes were ongoing at the end of 2015, in Algeria, Iraq, Israel, Lebanon, and Turkey. Factionalism Episodes in Asia Table 2.4 describes factionalism episodes in Asia from 1946 to 2015. While the cases of the Middle East and North Africa were fairly simple, with few changes in governance during factionalism episodes and, with the sole exception of Turkey, less than three episodes of factionalism per case, Asia includes some of the most complex factionalism cases in the world. Twenty-three countries in Asia experienced factionalism sometime between 1946 and 2015, including 44 discrete onsets of factionalism. Of these, seven countries experienced ongoing episodes of factionalism at the end of 2015, including Bangladesh, Cambodia, Kyrgyzstan, Malaysia, Nepal, Sri Lanka, and Tajikistan. Pakistan and Thailand are notable for experiencing five and six discrete factionalism episodes, respectively. Other countries have had fewer discrete episodes of factionalism, but have experienced factionalism over long periods with frequent changes in governance during those factionalism episodes. Malaysia, for example, has been factional since 1969, with many structural changes in governance over that period. Similarly, Sri Lanka has been dealing with open factionalism since 1970, with only a brief hiatus during a 4-year period between 2006 and 2010.

41

Factionalism in the Post–World War II World

Algeria

Bahrain Egypt Iran Iraq

Israel Jordan Lebanon

Morocco Syria Tunisia Turkey

North Yemen Yemen

1 2 2a 1 1 1 2 1 1a 1 1 1a 2 1 2 1 1a 1b 1 1 1a 1 2 3 1 1 1a

2 11 4 8 2 1 8 12 9 11 7 1 11 8 4 3 6 11 9 9 6 1 6 8 2 4 2

24 17 9 24 27 12 4 21 8 1 1 9 9 18 28 2 2 13 6 18 2 28 16 28 18 28 22

1989 2 1995 1 2004 2010 7 2005 3 1947 1 1997 1 2010 6 2014 1981 11 1951 11 1952 1989 1 1970 1 2005 6 1956 4 1961 1963 1950 6 1993 8 2002 1954 11 1997 6 2014 6 1948 6 1993 1 2012

1 4 99 3 2 8 1 9 99 99 1 3 7 5 99 6 11 6 11 6 1 5 7 99 9 2 9

14 8 99 14 11 19 24 7 99 99 8 13 5 25 99 1 12 6 28 1 14 27 28 99 19 21 21

1992 2004 9999 2011 2011 1953 2004 2014 9999 9999 1952 1957 1992 1975 9999 1961 1963 1965 1951 2002 2011 1960 2011 9999 1962 2012 2014

7 99 7 5 4 4 6

99 3

8 2 99 4 6 6

4 1 99 4 1

I P I S S

S S

S S S P S P

S S S S S

S

E G P P I S

−2 −3 2 −5 E −3 −1 P 3 3 6 6 −4 −1 −4 5 6 −5 −88 −3 2 E −3 −4 4 7 3 −6 −2 3

POLCOMP

EXCONST

POLITY

Tertiary Type

S S

EXREC

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.3 Factionalism Cases in the Middle East and North Africa, 1946–2015

3 3 7 3 3 6 5 5 6 1 2 6 3 3 6 2 5 6 6 4 6 7 4 6 7 6 7 7 7 10 2 3 6 2 5 6 2 3 6 8 4 7 7 6 7 1 2 6 −88 −88 −88 2 3 7 3 7 6 3 3 6 3 2 6 7 5 6 8 7 6 7 4 6 1 1 6 4 2 6 5 5 7

Note: POLITY, combined Polity score; EXREC, executive recruitment; EXCONST, executive constraints; POLCOMP, political competition. Under “Faction Type” columns, “S” stands for social identitybased factions, “E” represents economic factions; “P” represents personalistic or clientelistic factions; “I” represents issue- or policy-based factions; “G” represents geographic or regional factional cleavages. Bolded rows indicate changes in governance affecting the Polity scheme within factionalism episodes.

Factionalism Episodes in Latin America Table 2.5 describes factionalism episodes in Latin America between 1946 and 2015, including 20 countries that have experienced factionalism at some point in these years. Of the countries of mainland South and Central America, only Uruguay escaped factionalism during this period; it had experienced a brutally repressive dictatorship from 1973 to 1985, followed by an extraordinarily rapid and successful transition

42

Azerbaijan Bangladesh Cambodia

China East Timor Fiji India

Indonesia

Kazakhstan

Kyrgyzstan Laos Malaysia

Myanmar Nepal Pakistan

1 1a 1b 1 2 3 3a 1 2 1 1 1 2 1 1a 1 1a 1b 1 1a 1 1 1 1a 1b 1c 1d 1 1 2 3 3a 1 1a 1b 1c 1d 2 3 4 4a 5 5a 5b 5c

8 6 7 4 9 11 1 5 7 12 3 7 11 6 7 7 8 3 12 8 12 1 5 2 6 3 3 1 2 10 5 5 8 9 3 10 3 6 8 11 2 3 11 8 4

30 9 1 18 26 18 6 24 29 26 17 26 2 27 1 1 16 15 16 16 1 2 16 20 16 8 7 4 16 5 5 11 14 12 8 17 1 9 14 17 16 10 28 19 19

1991 4 1992 1993 1972 4 1991 6 2013 7 2014 1993 6 2013 1 1946 1 2006 11 1990 3 2001 6 1975 7 1977 1948 6 1950 1957 1991 4 1995 2011 4 1960 5 1969 1 1971 1995 2008 2014 1948 4 1959 11 2002 6 2009 6 2013 1947 3 1948 1949 1951 1956 1962 1 1973 7 1988 6 1997 2007 3 2007 2008 2010

6 6 11 12 1 1 99 7 99 1 7 5 11 6 5 8 3 7 8 6 99 5 2 6 3 3 99 3 12 4 5 99 9 3 10 2 10 11 7 2 10 11 8 4 5

8 30 11 16 10 5 99 5 99 10 14 16 23 30 14 15 14 5 15 25 99 3 19 15 7 6 99 2 15 24 10 99 11 7 16 29 7 28 5 15 12 27 18 18 10

1992 1993 1995 1974 2007 2014 9999 1997 9999 1949 2012 1999 2004 1977 1995 1950 1957 1959 1995 2002 9999 1961 1971 1995 2008 2014 9999 1962 1960 2006 2014 9999 1948 1949 1951 1956 1958 1969 1977 1997 1999 2007 2008 2010 2013

S

E

7 4 7

P P P

I S

4 99 10 1 1 8 6

P P P G S S S

4

S

3

G

P

4 2 99

S S S

4 3 6 99

S P G G

E

E

P

1 6 8

E P E

P E P

4

4

E

I I

P

−3 1 −3 S 8 6 4 1 1 2 −5 7 5 5 E 7 8 3 0 −1 −3 −4 7 −1 1 4 3 6 5 8 2 −6 6 7 −4 2 4 5 8 1 8 8 7 −2 2 5 6

POLCOMP

EXCONST

POLITY

Tertiary Type

9

EXREC

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.4 Factionalism Cases in Asia, 1946–2015

3 7 3 8 8 7 7 7 5 3 7 7 7 8 8 3 3 3 3 3 7 4 7 7 7 8 8 8 8 2 7 8 3 7 7 7 8 7 8 8 8 5 5 7 7

3 3 3 7 5 5 3 3 5 1 7 6 6 6 7 7 5 4 3 2 7 3 3 5 4 5 5 7 3 1 6 6 1 3 4 5 7 3 7 7 6 2 4 5 6

6 6 6 7 7 6 6 6 6 6 7 6 6 7 7 7 6 6 6 6 7 6 6 6 6 7 6 7 6 6 7 7 7 7 7 7 7 6 7 7 7 6 7 7 7

continued

43

Philippines Singapore Solomon Islands South Korea

1 7 1a 11 2 2 1 6

1 1 1a 1b 2 South Vietnam 1 1a 1b 1c Sri Lanka 1 1a 1b 1c 1d 2 2a 2b Tajikistan 1 2 2a Thailand 1 2 3 4 4a 4b 5 6 6a 6b

10 8 8 7 12 10 11 6 3 5 8 12 12 11 9 1 4 9 11 6 7 2 10 12 7 2 2 6 1 7

1 12 3 5

9 15 6 16 18 26 2 12 30 28 17 23 7 5 8 9 28 9 13 26 1 11 8 19 25 24 25 5 23 11

1950 1969 1987 1959

1990 1948 1952 1960 1963 1955 1963 1965 1973 1970 1978 1982 2001 2003 2010 2015 2015 1991 1998 2003 1955 1969 1974 1978 1988 1991 2006 2007 2008 2011

8

3 6

6 6 1 4 1

3 6 2

8 1 3 3 7 3

11 9 12 9

6 8 4 5 10 11 6 3 12 8 12 12 11 10 1 4 99 11 6 99 10 11 10 7 2 9 9 1 7 5

11 23 31 16

4 5 15 16 17 1 11 29 31 16 22 6 4 22 8 27 99 10 25 99 20 17 6 24 23 13 18 22 10 22

1969 1 1972 1991 8 1963 10

2000 2 1952 3 1960 1961 1972 4 1963 4 1965 1973 1975 1978 3 1982 2001 2003 2006 2015 1 2015 9999 1992 6 2003 99 9999 1958 4 1971 4 1976 4 1988 1 1991 1992 2006 4 2008 6 2011 2014

E P

G

E E S

G

S

G

S S

G G

S S

G G

S S S S

G G G G

S

POLCOMP

EXREC

POLITY

Tertiary Type

P I I

EXCONST

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.4 Continued

5 2 8 7

8 −3 −4 8 3 −3 −3 −66 −3 8 6 5 6 5 3 4 6 −2 −1 −3 −3 2 3 2 3 −1 8 −1 4 7

8 8 8 8

5 3 7 7

6 6 7 6

8 7 7 3 3 6 3 2 6 8 7 7 8 3 7 3 3 6 4 1 6 −66 −66 −66 3 3 6 8 7 7 8 5 7 8 5 6 8 6 6 8 5 6 7 4 6 8 4 6 8 6 6 3 3 7 3 4 6 3 3 6 3 3 6 8 3 6 8 3 7 7 3 7 8 3 7 4 3 6 8 7 7 3 4 6 7 5 6 8 6 7

Note: POLITY, combined Polity score; EXREC, executive recruitment; EXCONST, executive constraints; POLCOMP, political competition. Under “Faction Type” columns, “S” stands for social identitybased factions, “E” represents economic factions; “P” represents personalistic or clientelistic factions; “I” represents issue- or policy-based factions; “G” represents geographic or regional factional cleavages. Bolded rows indicate changes in governance affecting the Polity scheme within factionalism episodes.

44

Argentina Bolivia Brazil Chile

Colombia

Dominican Republic

Ecuador

El Salvador Guatemala

Guyana Haiti

1 3 2 10 2a 5 1 6 2 10 3 2 3a 2 1 5 1a 7 1b 9 1c 1 1 5 1a 9 1 4 2 11

1 2 2a 2b 1 1a 1b 1c 2 2a 2b 1 2 3 1 2 2a 2b 3 1 1a 1b 2 1 2 2a 3 4

12 6 8 8 6 11 6 6 1 11 11 10 2 6 12 3 3 3 1 5 12 4 10 12 10 1 11 5

12 31 15 18 11 12 7 2 1 3 23 22 5 9 5

20 2 17 15 7 9 3 23 21 26 29 27 21 2 11 7 2 4 15 26 17 11 6 16 16 1 26 14

1973 1 1983 1 1989 1956 1 1982 6 2003 7 2009 1947 — 1958 1961 1963 1955 8 1964 1948 7 1995 7

1962 1966 1978 1994 1948 1961 1968 1970 2000 2006 2007 1960 1972 1984 1950 1966 1970 1974 1986 1966 1968 1978 1992 1990 1994 1996 2000 2006

1 1

1 7 5 1 6 1 1 6 4 1 1 6 1 1

3 5 12 11 8 2 99 6 9 1 4 9 9 7 99

9 8 8 8 11 6 6 2 11 11 2 3 2 9 6 3 3 3 1 12 4 10 99 9 12 1 3 1

24 14 31 4 6 6 99 30 2 22 9 4 11 20 99

25 16 14 16 8 2 22 15 25 28 16 15 20 25 27 1 3 5 15 16 10 6 99 30 31 10 8 12

1976 7 1989 8 1989 1964 4 1985 8 2009 99 9999 1958 — 1961 1963 1964 1964 1 1973 1957 1 9999 99

1963 4 1978 1 1994 1996 1961 4 1968 1970 1972 2006 99 2007 2013 1964 7 1977 4 1991 1 1954 4 1970 1 1974 1978 1996 1 1968 3 1978 1980 9999 99 1991 4 1995 1 1999 2004 4 2010 6

E I S P S

E I

E I I

P S E G

S

E E E E E

P

S

E S

P

S P P

P E E

P

E

POLCOMP

EXREC

POLITY

Tertiary Type

E

EXCONST

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.5 Factionalism Cases in Latin America, 1946–2015

6 8 7 −3 8 8 7 5 6 5 3 5 6 −5 7

8 −3 6 5 2 −1 5 0 E 6 7 5 −3 −1 6 2 3 1 −3 3 2 1 0 6 7 7 7 −2 5

8 8 8 3 8 8 8 8 8 8 8 8 8 3 8

8 3 8 7 8 4 8 8 7 8 8 3 3 8 8 8 7 3 8 3 3 3 8 8 8 7 3 7

5 7 6 3 7 7 6 5 5 4 3 4 5 1 6

7 3 5 5 3 3 5 1 6 6 4 3 4 5 3 3 3 3 3 7 6 5 5 6 6 7 3 6

7 7 7 6 7 7 7 6 7 7 7 7 7 6 7

7 6 7 7 6 6 6 6 7 7 7 6 6 7 6 7 6 6 7 6 6 6 7 7 7 7 7 6

continued

45

Honduras

Mexico Nicaragua Panama Paraguay

Peru

Suriname Trinidad Venezuela

1 1a 1b 2 2a 1 1a 1b 2 1 1a 1b 1 1 2 2a 2b 2c 2d 1 1a 2 3 3a 4 1 1a 1b 1 1 1a 1b 1c

10 3 12 1 11 9 7 8 1 2 2 7 1 7 5 6 8 3 4 7 7 7 7 7 12 11 2 8 8 11 12 2 4

22 29 5 28 25 2 14 16 1 22 27 6 3 29 2 23 15 29 27 29 29 29 29 29 30 25 25 15 31 13 18 11 19

1956 6 1971 1972 1982 3 1985 1977 3 1988 1994 1999 1 1984 8 1990 1995 1955 6 1947 — 1989 1 1992 1998 1999 2003 1950 2 1956 1963 6 1980 1 1990 1993 3 1975 4 1980 1980 1962 6 2001 7 2006 2009 2013

3 12 7 11 11 7 8 7 11 2 7 1 10 7 6 8 3 4 8 7 7 10 7 4 11 2 8 3 11 12 2 4 99

28 4 25 24 26 13 15 5 30 26 5 9 12 11 22 14 28 26 14 28 18 3 28 5 21 24 14 16 26 17 10 18 99

1971 4 1972 1980 1985 1 1989 1988 1 1994 1997 2000 8 1990 1 1995 2007 1968 4 1954 — 1992 3 1998 1999 2003 2008 1956 1 1962 1968 4 1990 4 1992 2000 5 1980 7 1980 1982 1984 8 2006 99 2009 2013 9999

I I E

P

E I I

E E E P S

S E

I I

POLCOMP

EXREC

POLITY

Tertiary Type

E

EXCONST

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.5 Continued

−1 0 −1 6 5 −3 0 4 6 −1 6 8 4 −5 2 7 6 7 8 −2 5 5 7 8 1 5 0 −1 8 6 5 −3 4

4 5 4 7 7 3 3 7 7 5 8 8 7 3 7 7 7 7 8 5 7 8 8 8 7 7 3 4 8 8 8 3 7

3 3 3 6 5 3 4 4 6 3 5 7 5 1 3 7 6 7 7 2 5 5 6 7 3 5 4 3 7 5 4 3 4

6 7 6 7 7 6 7 7 7 6 7 7 6 6 7 7 7 7 7 6 7 6 7 7 6 7 7 6 7 7 7 6 7

Note: POLITY, combined Polity score; EXREC, executive recruitment; EXCONST, executive constraints; POLCOMP, political competition. Under “Faction Type” columns, “S” stands for social identitybased factions, “E” represents economic factions; “P” represents personalistic or clientelistic factions; “I” represents issue- or policy-based factions; “G” represents geographic or regional factional cleavages. Bolded rows indicate changes in governance affecting the Polity scheme within factionalism episodes.

46

Democratization and the Mischief of Faction

to democratic governance that did not include open manifestations of factionalism. Cuba and Jamaica also did not experience factionalism during the study period. All but three countries in Latin America (Brazil, Chile, Suriname) experienced multiple discrete onsets of factionalism, with some (e.g., Peru) being quite complex with recurring and unique factional identities emerging over time. Honduras is notable for being openly factional from independence in 1839 to 1989, with only brief interludes during periods of interregnum. Four cases— Bolivia, Colombia, Guyana, and Venezuela—were experiencing ongoing episodes of factionalism at the end of 2015. Peru offers an illustrative example of the nonlinear path that democratic transition often takes and the ways in which factionalism affects, and is affected by, liberalization and autocratization processes. Peru has experienced four discrete episodes of factionalism: (1) 1950−1956; (2) 1963−1968; (3) 1980−1990; and (4) 1993−2000. Within episodes 1 and 3 there were also substantial changes in governance, described in Table 2.4 as intraepisode events 1a and 3a. Peru has struggled with factionalism for most of its history, including a century-long factional stretch from independence during 1821 to 1919. Politics in this period, and into the 1950s, was characterized by frequent transitions in power between civilian (often leftist) political parties and the military, which ruled with the support of Peru’s (typically conservative) economic elites. In 1948 General Manuel A. Odria toppled Peru’s civilian government in a military coup and, 2 years later, resigned his military post and successfully ran for the presidency as a civilian, unopposed due to an election boycott by the opposition. The return to nominally civilian rule reflected a degree of liberalization, especially in political competition and executive recruitment, which allowed factionalism to manifest itself. Odria cracked down quite coercively on the faction led by the leftist opposition party, American Popular Revolutionary Alliance (APRA), but the government generally allowed dissent in political discourse. In 1956 Odria made significant moves toward further liberalization by legalizing some opposition parties and calling for general elections, in which he did not participate. Manuel Prado, former coup leader (1914) and president (1939−1945) won the election, legalized APRA, and allowed the return of APRA leadership from exile. These developments were characteristic of ongoing liberalization in the formerly autocratic regime, in terms of executive recruitment, executive constraints, and political competition, described in episode 1a in Table 2.5. Factionalism, with polarization along the same economic cleavage that had characterized the country’s factionalism since independence,

Factionalism in the Post–World War II World

47

continued until 1962 when the country was poised to have its second peaceful transfer of power due to an election. In June 1962 three candidates, Victor Raúl Haya de la Torre (APRA), Fernando Belaunde (Popular Action Party), and former president Odria, contested the election, and none achieved the one-third plurality needed to win the presidency. This pushed the vote to the Peruvian Congress but, despite the possible crisis, APRA agreed to form a coalition with and to support Odria’s candidacy, in a remarkable display of conventional political behavior. The military responded by demanding annulment of the election results, citing claims of electoral irregularities that were disproven by the state’s Electoral Board. The factional episode ended on July 19, when a military junta seized control of the state, arrested President Prado, annulled the election results, banned APRA, dissolved the legislature, and appointed General Ricardo Pérez Godoy to the presidency. The repressive practices of the military dictatorship again drove factionalism underground, as the statesocietal system returned to its autocratic equilibrium point. Factionalism reemerged in Peru in July 1963 (episode 2 in Table 2.5), when a newly elected government was inaugurated. The military junta that came to power in 1962 oversaw new elections on June 9, 1963, and again legalized APRA, allowing opposition groups (except the Communist Party) to organize against the government and compete in the electoral process. This again allowed the repressed factionalism in the system to manifest itself, which it did, again along largely economic lines. Belaunde won the election representing a coalition of the center-right (middle-class) Population Action and Christian Democratic Parties, with 39 percent of the vote, and was sworn in on July 28, 1963. APRA joined with President Odria’s personal political party, National Union Odristas, to form a majority coalition in congress opposed to Belaunde and the military. The Belaunde regime was more liberal than its military predecessor, but was unable to transition past the open factionalism and mixed democratic institutions. Executive constraints were light even with the APRA coalition in the legislature, and political competition remained transitional, with limited suppression of political behavior, including a ban on the Communist Party. Belaunde was reelected in June 1968, but his attempts to liberalize the system further were stopped by the military on October 3, 1968, when a junta overthrew the government, dissolved Congress, and suspended constitutional protections, again driving factionalism underground. Interestingly, the new military government did not represent the economic elites with whom the military was traditionally allied; instead, General Juan Velasco’s presidency

48

Democratization and the Mischief of Faction

saw the implementation of many far-left policies, including land reform, education reform, nationalization of oil and other industries, and formation of political alliances with the Soviet bloc. While leftist intellectuals and academics received some freedom to speak and publish for the first time under the military regime, organized political competition would remain repressed for the next 12 years. The country’s next episode of factionalism (episode 3 in Table 2.5) began on July 29, 1980, following a gradual transition to civilian rule that had begun in 1977, including the lifting of martial law and resumption of political organizing (August 19, 1977), legislative elections (June 18, 1978), and creation of a new constitution (July 12, 1979). The new legislature announced general elections for May 18, 1980, in which Belaunde was again elected president for the Popular Action Party, and assumed office on July 28, 1980. The Popular Action Party (later becoming the Democratic Front [FREDEMO]) also won a narrow majority in both houses of Congress while APRA came in second in each chamber. The new government was characterized by more liberal policies toward political competition and greater constraints on the executive than the prior civilian regimes, but was unable to consolidate institutions sufficiently to normalize political competition out of a factional mode. Episode 3a (Table 2.5) denotes an increase in executive constraints associated with the June 1990 election of political outsider and antiestablishment candidate, Alberto Fujimori, who was inaugurated on July 28, 1990. His ability to act was constrained by the legislature, now dominated by FREDEMO. Factionalism remained rampant, and even intensified, despite the additional executive constraints, as antisystem rhetoric and perceptions of an ineffectual government intensified in the context of the increasingly successful Shining Path guerrilla insurgency and hyperinflation. Peru’s most democratic regime to date fell back into autocratic equilibrium and factionalism was again driven underground on April 6, 1992, when Fujimori executed an autogolpe, using security threats and hyperinflation as justification. With military support, Fujimori suspended the constitution and implemented martial law. Factionalism emerged again for a fourth discrete episode with the promulgation of a new constitution that dramatically weakened executive constraints, but also liberalized political competition. Fujimori allowed competition to resume, in part, because of his stranglehold on the institutions of political power in the state. Fujimori’s success in fighting the Shining Path in the 1990s undermined the justification for his strongman presidency, however, and in

Factionalism in the Post–World War II World

49

1999 and 2000 the previously fragmented opposition formed an unnatural alliance to challenge his regime. In April 2000 opposition parties boycotted the presidential election, which Fujimori won amid claims of widespread voter fraud. Mass protests began and, facing an internal power struggle with his own security chief, Fujimori exiled himself to Japan and announced his resignation, which the legislature rejected before removing him from office due to “moral incapacity.” He was replaced by an interim government led by Valentin Paniagua, leader of the AP and president of the legislature. This most recent factionalism episode (episode 4 in Table 2.5) concluded with Fujimori’s ouster, as political competition began to normalize and political behavior became more conventional and as the state shifted away from coercive practices. New elections in June 2001 saw APRA’s candidate, former president Alan Garcia lose to Alejandro Toledo of the center-left Peru Possible Party, the country’s first indigenous Andean Indian to take the presidency. Toledo pursued generally moderate policies but conducted massive reforms to depoliticize the military, including a purge of senior officers, that effectively removed the military from Peru’s political landscape. Structurally, bans on political parties were lifted, executive constraints were restored, and subsequent elections were free, fair, and peaceful. Indeed, Toledo lost reelection in 2006 to APRA’s Garcia, in a peaceful transition of power that consolidated the country’s nascent democratic institutions. While economic inequality is still a major political issue in Peru, political parties have become increasingly diverse in terms of policy preferences, leading to increasingly cross-cutting rather than reinforcing cleavages among them. Economic elites, the middle class, and the working class now spread loyalties across political organizations, movements, and candidates. As a result of state choices and these trends in political competition, Peru’s state-societal system has effectively topped the hill in its leftward migration in the SSP model, reaching a new equilibrium point as a consolidating democracy experiencing generally conventional politics. Peru’s case, where leftward and rightward migration moves a societal system in and out of factionalism, is common across the world, but particularly in Latin American and Asian cases where the condition has tended to be more complex and repetitious compared to other regions. This also challenges previous notions of democratic transition as a linear phenomenon, and teleological theories of democratic transition in general. Choices by government and opposition actors increase or decrease the density of complexity in state-societal systems and change structures and patterns of behavior. In addition to altering the degree to

50

Democratization and the Mischief of Faction

which these systems are autocratic or democratic, these changes fundamentally alter patterns of political competition, allowing the manifestation or repression of factionalism which in turn affects the institutions and behaviors of actors in the system. In Peru this tendency was cyclical for most of the country’s history, with the military stepping in whenever factional competition had occurred for more than one or two presidential terms. The professionalization of the military and pluralization of political organization stopped the cycle of factionalism and repression, and allowed the state to settle in a democratic equilibrium point. Also, Peru is a good example of the military repeatedly arbitrating factional strife, common in Latin American regimes in the twentieth century but also notable in Egypt and Thailand (see Chapter 6), Pakistan, and Turkey.

Factional Episodes in Europe Table 2.6 describes factionalism episodes in Europe from 1946 to 2015, including 16 countries and 21 discrete onsets of factionalism. Of these, all cases except for Belgium occurred in Central and Eastern Europe, most of which were former Soviet states. Only three cases in Europe are ongoing, in Armenia, Belgium, and Ukraine. Armenia, Cyprus, and Ukraine are the only states in Europe to experience multiple factionalism onsets, with Ukraine having the most complex factionalism episodes and regime changes within those episodes. Among the European cases Belgium is notable as the only state among the mature democracies, the North American and Western European countries typically thought of as industrialized and fully consolidated democracies, to experience open factionalism during the study period. Because of this, Belgium serves as a critically important case for exploring the utility of the SSP model as it applies to consolidated democracies at the equilibrium point in conventional politics. Belgium’s atypical nature as a factional Western democracy also applies to the nature of its transition into factionalism, which was unaccompanied by broader structural changes in the political system: formal executive recruitment processes and executive constraints have been unchanged. Instead, political competition has become increasingly polarized, which has in turn made political institutions increasingly ineffective. Belgium’s factions are primarily identified by linguistic and nationalist cleavages, with a Flemish (Dutch-speaking) alliance of political parties contesting a francophone Walloon faction. This political cleavage

51

Factionalism in the Post–World War II World

Albania

1 12 1a 7 1b 5 Armenia 1 7 2 2 Belarus 1 8 1a 3 Belgium 1 6 Bulgaria 1 3 Croatia 1 6 Cyprus 1 8 2 4 Czechoslovakia 1 6 Estonia 1 9 1a 3 Georgia 1 4 2 4 Moldova 1 8 1a 8 1b 4 Romania 1 6 Russia 1 1 Serbia and Montenegro 1 3 Slovak Rep. 1 1 Ukraine 1 12 1a 9 2 11 2a 1 2b 9 3 11

12 1 26 7 21 25 16 10 30 25 16 16 9 6 7 9 9 27 4 7 21 1

11 1 1 10 28 1 30 27

1990 1992 1996 1995 2003 1991 1994 2007 1990 1991 1960 1968 1990 1991 1999 1991 2000 1991 1993 2001 1990 1992

2003 1993 1991 1993 2000 2006 2010 2014

2 1 1 4

1 1 4 4 1 1 4 4 1 4

1 6

3 4 6 7 6

6 5 7 9 99 3 4 99 7 10 12 7 12 3 3 11 1 8 4 3 11 10

6 10 9 7 12 9 2 99

30 25 24 27 99 15 15 99 23 1 21 15 31 6 13 6 24 3 6 5 15 16

2 30 9 18 31 29 22 99

1992 1996 1997 1996 9999 1994 1995 9999 2001 1995 1963 1974 1992 1999 2000 1995 2004 1993 2001 2005 1996 1993

P

E

7 99 7

S P E

E

99 8 11 10 4 10 8

I I S S S G I P P S

E E E

1 7

I E

P

1 6 1

2006 10 1998 6 1993 6 1994 2005 3 2010 2014 9999

G P G G

S

E

E S S

S S

S P P

POLCOMP

EXREC

POLITY

Tertiary Type

8

EXCONST

Secondary Type

Episode Ending Date

Primary Type

Episode Beginning Date

Ending Event

Onset Event

Country

Episode No.

Table 2.6 Factionalism Cases in Europe, 1946–2015

1 5 0 3 5 7 7 8 8 −3 8 7 8 6 7 4 5 5 7 8 5 5 6 7 6 5 6 7 6 4

7 8 3 8 7 7 8 8 8 3 8 8 8 7 8 7 7 7 7 8 7 7

7 8 7 7 8 8 8 7

3 5 5 3 5 7 7 7 7 3 7 6 7 7 7 5 5 5 7 7 5 5

6 6 6 5 5 6 5 5

6 6 6 7 7 7 6 7 7 6 7 7 7 6 6 6 7 7 7 7 7 7

7 7 7 7 7 7 7 6

Note: POLITY, combined Polity score; EXREC, executive recruitment; EXCONST, executive constraints; POLCOMP, political competition. Under “Faction Type” columns, “S” stands for social identitybased factions, “E” represents economic factions; “P” represents personalistic or clientelistic factions; “I” represents issue- or policy-based factions; “G” represents geographic or regional factional cleavages. Bolded rows indicate changes in governance affecting the Polity scheme within factionalism episodes.

became particularly salient after the Marshall Plan’s (named for US secretary of state George C. Marshall) post−World War II economic reforms in Europe that favored the Dutch-speaking coastal region of Flanders, with access to coastal trade, over the isolated but heavily industrialized interior region of French-speaking Wallonia. This in turn increased

52

Democratization and the Mischief of Faction

demands that the state recognize Dutch as an official language of Belgium, which led to political debates about restructuring the state into a federation to ensure representation of both Flemish and francophone communities. The increasingly salient linguistic divide resulted in government reforms in 1968−1969 that included a consociational system of dividing government ministers, diplomats, and military posts according to linguistic identity group, and the requirement of a two-thirds majority to pass legislation in each house of the legislature, thus requiring support from both identity groups. The state further formalized the linguistic divide in 1971 by promulgating constitutional amendments that guaranteed equality across the identity groups, but which also required all political candidates to declare publicly their loyalty to one of the two linguistic groups. While this alleviated political angst about linguistic equality, it also effectively reorganized political life around linguistic identity, polarizing the population and undermining cross-cutting political cleavages. Each linguistic identity group has its own issue-based parties, who unite in unnatural alliances to support linguistic identity interests over their distinct policy preferences. For example, the Groen! and Ecolo parties both pursue green party policy preferences, but Groen! is active in only Flemish communities and Ecolo in francophone communities, and they form coalitions with parties that share their linguistic identity and oppose their policy preferences, rather than uniting on their common green party policy orientation. These decisions that increased the salience of the ethnolinguistic divide also encouraged the development of distinct national and cultural identities between the linguistic groups, which have aligned with economic interests of the regions where those groups predominate. Over time, Flemish political parties have tended to favor greater autonomy from the central government for the wealthier Dutch-speaking regions, and a smaller role for government in social life (which predominantly benefits less prosperous Wallonia). In contrast, francophone parties have come to favor a larger role for government and policies to redistribute wealth from Flanders to French-speaking Wallonia. Leftist parties in Flanders, for example, won only 25 percent of the Flemish vote in the 2007 election while leftist parties in Wallonia won a majority. In the context of this increasingly divided political landscape, elections in June 2007 resulted in a narrow plurality for Yves Leterme, leader of the Flemish nationalist Christian Democratic and Flemish party, who had campaigned on the promise of constitutional amendments to grant the Dutch-speaking communities of Flanders autonomy from the central government. The autonomy question built on the

Factionalism in the Post–World War II World

53

salience of the linguistic divide but also threatened the welfare of Wallonia, francophone communities in Flanders, and French-speaking Brussels, for the parochial benefit of Dutch-speaking Flanders. As a result of this polarization, for months Leterme was unable to secure sufficient support from other political parties to form a majority coalition. Consequently, in February 2008 the monarch asked outgoing prime minister Guy Verhofstadt to form an interim government until Leterme could find a coalition partner. Nearly a year after the election, on March 20, 2008, Leterme formed a grand coalition with Verhofstadt and three centrist francophone parties by excluding far-right Flemish nationalist parties. The resulting coalition government included both pro-autonomy and antiautonomy factions, however, and the deadlock between their positions (in the context of the 1969 requirement for a two-thirds majority to pass legislation) left the legislature and the executive unable to resolve the factional political crisis, and allowed far-right nationalist movements in Flanders to flourish, which intensified factionalism along identity lines. Elections in June 2010 produced an even more fragmented polity, and no coalition was able to form a government for nearly 2 years. After 541 days, a coalition of parties produced a series of constitutional reforms, including providing additional autonomy over economic and social policies to regional and local governments and changing the Senate from an elected body to an assembly of regional legislatures similar to the German Bundesrat. Under the auspices of this agreement, a coalition finally formed that governed until 2014, when elections again produced a long delay—more than 5 months—in forming a coalition. Notably, political rhetoric, especially among the more extreme parties in Wallonia, included calls for the state’s fragmentation: a partition of Belgium in which the Walloon minority would break away from Belgium to become a province of France. The pursuit of parochial interests by mainstream political parties at the expense of the national good and indeed the continued existence of the polity, the routine formation and persistence of unnatural alliances, the emergence of widespread demonstrations and political violence, and the repeated inability of government to even form all characterize the onset and persistence of a factionalism episode in this case, even as executive constraints have remained rigorous and executive recruitment has remained based on a system of free and fair elections. Belgium is thus an anomalous case: in terms of the state-society relationship, Belgium is distinctly factional, with all of the hallmarks of open factionalism. The factionalism is not maintained because the state is failing to liberalize its institutions, however, but because it has repeatedly (through

54

Democratization and the Mischief of Faction

democratic processes) changed its institutions in such a way as to encourage persistent polarization along identity lines, undermining institutions that in most democracies are self-correcting. Belgium is therefore at a point on the SSP model that, for an autocratic state, would be a natural equilibrium, but Belgium is institutionally unable to employ the authoritarian means necessary to suppress the factional strife and cannot simply liberalize its way out of the problem because its institutions are already democratic. Hence, Belgium is subject to the mischief of factionalism, including increasingly vocal calls for the state’s dissolution, without the tools available to either democratizing or autocratizing regimes. The SSP model would thus predict persistent factionalism with its tendency toward generating instability, as indeed has characterized Belgian politics since 2007. Because its focus is not on regime type or structural attributes, the SSP model can account for important changes in political competition in a case like Belgium that most mainstream theories of democratization and backsliding cannot, even though the case is unusual among factionalism cases. Analysis of Faction and Event Types The research underlying the data on factions included the development of reports that described every factionalism episode, and indeed every change in governance recorded in the Polity IV dataset, initially from 1955 to 2015 and expanded here back to 1946. In addition to descriptions of these changes, this research also produced, for every discrete factionalism onset, identification and descriptions of the polarized factions that characterize the factionalism episode, including information on political organizations or parties, social identity groups, economic class, personalistic leaders, and any coalitions or alliances among groups or factions. This includes descriptions of fracturing within groups or factions. Often, as in the examples of Burundi, Peru, and Belgium above, factions are consistent over time, indeed often institutionalizing themselves as state decisions and policies favor the group in power and marginalize the opposition faction(s). Fracturing within factions, though, is often indicative of changes in the relative salience of various cleavages on the factional landscape, as in the Peruvian case where economic cleavages lost their salience to more cross-cutting policy- and issue-based divides. Review of the factionalism case reports yields five common types of cleavages separating factions and, thus, defining them. Two of these cleavage types represent structural divides in privilege and access to

Factionalism in the Post–World War II World

55

power: social identity (e.g., ethnicity, religion, tribe, language) and economic class. Ad hoc factions, in contrast, are often based around a particular issue (internal, external, or complex) or are personalistic. Factional cleavages rarely fit only one of these common divides, however; they more frequently are intersectional, overlapping, and complementary, and can evolve over time. In Burundi’s case, factional cleavages were initially strictly along Hutu/Tutsi ethnic lines, parallel with class, but evolved to be more complex, featuring increasing intra-Hutu division along economic and personalistic more than ethnic lines. Peru’s factionalism episodes in the 1960s featured factions divided along lines of class and ideology (leftist APRA vs. elite-supported military and center-right parties) competing with General Odria’s personalistic faction. In the case of Belgium, linguistic and nationalist identity divides complemented and reinforced the divide around two particular policy issues—distribution of wealth to Walloons and autonomy for Flemings—to intensify factionalism in that state, and over time that cleavage has become parallel with a class cleavage as well. To better understand how frequently different types of factions emerge and to explore trends in factional identity, I oversaw a coding exercise of all country reports and some episodes that were coded factional late in the project that never received country reports, in 2013 and again in 2015. After identifying all factional episodes in the Polity IV data, graduate and undergraduate research assistants, who I supervised, worked with me to review each country report, the Keesing’s World News Archive, and LexisNexis (especially BBC Monitoring) for data on episode onset event type, episode ending event type, factionalism manifestations, factionalism management strategies by the state, faction type, and identities of governing and opposition factions. Among faction types, coders identified primary, secondary, and tertiary faction types, recognizing the tendency of factions to be intersectional, and initially identified seven types: ethnic, economic, religious, personalistic/ clientelistic, policy/issue based, tribal, and regional/geographic. The data was then recoded into five types of cleavages to reflect similarities among the social identity factions, combining ethnic, religious, and tribal cleavages into a single typology. This data is included on a case-by-case basis in Tables 2.2−2.6, describing factionalism episodes by region. While all cases were assigned a primary faction type, only about half of cases were assigned a secondary faction type and only about one-sixth of cases were assigned a tertiary type. Also, because factions often are repetitive in subsequent factionalism episodes after the first episode, potentially inflating faction type frequencies in cases with multiple

56

Democratization and the Mischief of Faction

onsets, data on factionalism identities are presented both per episode, for 172 episodes, and in the first onset of the 107 countries that experienced at least one onset. Table 2.7 describes these frequencies. As shown in Table 2.7, personalistic/clientelistic and social identity (primary type) factions occur in roughly the same frequency, and together account for about 70 percent of cases in both all episode and first onset sets. Personalistic/clientelistic and regional/geographic factions were the most likely to recur in similar forms in subsequent factionalism episodes, with only minor differences in the other categories between the all episode and first onset sets of cases. Primarily economic factions account for 15 percent of factionalism types, policy/issue factions about 10 percent, and regional/geographic factions between 5 and 8 percent. Not surprisingly, social identity factions are also likely to have secondary and tertiary characteristics related to geography and economics, as social identity groups are often geographically segregated within states, and because social identity groups often become salient because of discrimination or structural economic disadvantages. Among the first onset episodes, of the 42 cases with factions divided primarily on social identity cleavages, 9 (21 percent) had secondary or tertiary economic components, 6 (15 percent) had a secondary regional/geographic component, and 5 (12 percent) exhibited significant secondary or tertiary policy/issue cleavages. Factions divided primarily by personalistic or clientelistic (patronage) cleavages were even more likely to exhibit intersectional identities, with nearly two-thirds carrying strong secondary or tertiary identities along social, economic, issue, or geographic lines. Fifteen (47 percent) of the 32 first onset cases with primarily personalistic/clientelistic factions had secondary or tertiary social identity cleavages, 7 (22 percent) Table 2.7 Frequency Distribution of Primary Faction Types All Episodes

Primary Faction Type

Social identity (ethnic, religious, tribal) Economic Personalistic or clientelistic Policy/issue based Regional/geographic

n

%

60 26

35 15

57 16 13

34 9 8

First Onset Cases

Primary Faction Type

Social identity (ethnic, religious, tribal) Economic Personalistic or clientelistic Policy/issue based Regional/geographic

n

%

42 16

39 15

32 11 6

30 10 6

Factionalism in the Post–World War II World

57

had secondary or tertiary economic cleavages, 3 (9 percent) had secondary policy/issue cleavages, and 2 (6 percent) had secondary or tertiary regional/geographic cleavages. This may reflect the tendency of charismatic personalistic leaders to take advantage of existing cleavages to mobilize supporters within factions, or the instrumental ability of such leaders to make underlying divisions salient. It may also reflect an artifact of media coverage and coding, whereby charismatic leaders are more likely to be focused on by the media covering events and disputes, leaving structural roots of problems undiscussed and thus outside coders’ ability to identify from contemporary news reports. Table 2.8 describes regional trends in primary faction types among first onset cases. Of the 34 initial onsets of factionalism in sub-Saharan Africa, roughly half were characterized by primarily social identity cleavages and about half by personalistic or clientelistic cleavages. The other faction types were less common, with only two cases showing primarily economic cleavages and one case with a primarily geographic cleavage. Most cases that demonstrated personalistic or clientelistic cleavages among factions in sub-Saharan Africa also had secondary social identity or economic cleavages. Factions in the Middle East were most likely to be oriented around social identity, with social identity cleavages characterizing 8 of 14 episodes. Personalistic and policy based divides were also common. Secondary and tertiary characteristics in this region were roughly uniformly distributed across the faction types. Note, however, that the small population of factionalism cases in the region makes drawing conclusions from this data dubious. Asia’s distribution of faction types mirrors that of sub-Saharan Africa. Of the 23 first onsets of factionalism in the region, 9 episodes (39 percent) were characterized primarily by social identity cleavages and 8 episodes by personalistic cleavages (35 percent), representing nearly three-fourths of the region’s cases. Secondary and tertiary cleavages were common, with 27 such identifications across the 23 countries. While social identity was the most common secondary/tertiary divide, these were roughly uniformly distributed. Latin America offers the most distinct pattern of factional identities, with half of cases exhibiting primarily economic factional identities. This reflects the long history of class politics in the region, as factions have often formed in support of particular economic classes, often the economic elite allied with the military in government and opposed by lower or middle-class factions. Factions divided primarily along social identity lines were the second most common (5 of 20),

Table 2.8 Faction Type Distribution by Region (First Onset Cases)

Region

Sub-Saharan Africa

Primary Secondary/ tertiary MENA Primary Secondary/ tertiary Asia Primary Secondary/ tertiary Latin America Primary Secondary/ tertiary Europe Primary Secondary/ tertiary

N

15

13 8 3 9

5 5

0 5

5

Social Identity

Economic

%

N

%

38 57

4 0

12 0

44

21 39

2

2 2

22 25

4 10

31

6

0 31

Note: MENA, Middle East and North Africa.

1 2

Personal/ Clientelistic N

6

15

14 9

1 8

17 50

5 13

39

2 3

3 2

3 3

2

Policy/ Issue

Regional/ Geographic

%

N

%

N

%

6 21

3 3

9 21

2 0

6 0

2 3

9 15

47

7 35

13 10

15 19

13

0

1 1

2 4

0

0

7 4

10 25 0

1

3 3

4 0

0 2

0

3

21 13

17 0

0 13 0

Total 33 14 23 20 16

Factionalism in the Post–World War II World

59

with personalistic and issue factions also represented. Personalistic and issue cleavages were also common secondary and tertiary identity characteristics of Latin America, and this probably describes the tendency for charismatic leaders to link policy issues and economic class identity in the pursuit of political power. Europe’s factionalism episodes offer the most uniform distribution of primary factional identities, with five episodes characterized by social identity cleavages, four by policy/issue cleavages, three by personalistic cleavages, and two each by economic and regional cleavages. Secondary and tertiary characteristics, however, were bimodally distributed in the European cases, with five cases (31 percent) exhibiting secondary social identity cleavages, six cases (39 percent) exhibiting economic cleavages, and two cases (13 percent) exhibiting personalistic or clientelistic cleavages. This suggests that regardless of the gross social distinction most clearly characterizing a European case, social or economic cleavages often underlie factional strife. The coding exercise also generated data on the types of events most commonly identified as factionalism episode onset or ending events, which are described on a case-by-case basis in Tables 2.2−2.6. Of all 172 discrete episodes of factionalism during the study period, events that signal the beginning of a factionalism episode were most frequently elections (35 percent) or the inauguration of a new government, typically with a distinctly different liberal approach to political competition (39 percent), that was usually put into power due to elections. New governments, or the elections preceding them, thus signal the beginning of nearly three-fourths of all factional episodes. Note that these events also reflect changes in the way political competition is regulated by the state since allowing elections usually also means allowing political parties to organize and assemble, free speech, and other concomitant forms of liberalization. This demonstrates the potential danger of rushing into elections before other institutions are developed, much as the sequentialists have argued in the political science literature (Manfield and Snyder 1995, 2005, 2007) that factionalism should be expected with election reforms. Constitutional reform, party legalization, and general liberalization account for another 16 percent of onset events, and 12 cases (7 percent) became factional due to the rapid and apparently spontaneous onset of mass political action such as widespread demonstrations.2 Twenty-one cases (12 percent) were factional when they achieved independence. The events that signal the end of a factionalism episode are more uniformly distributed, but generally are associated with three possible (shortterm) outcomes: consolidation of democratic institutions, consolidation of

60

Democratization and the Mischief of Faction

autocratic institutions and a return to repressive factionalism management (typically by coup or autogolpe), or a descent into militancy and open warfare; 31 cases (18 percent) were ongoing at the end of 2015. Elections, new governments (often based on elections), constitutional reform, and general liberalization are most often (but not exclusively) associated with (often short-term) democratic consolidation outcomes; these ended a combined 40 percent of factionalism episodes. Elections are the most common of these events, at 14.5 percent. Coups are the most common single outcome of factionalism episodes, and have ended approximately 25 percent of factionalism episodes since 1946, while general acts of autocratization and bans on political parties account for another 9 percent. Only about 7 percent of episodes have been directly ended by open warfare or polity fragmentation, but many episodes— nearly half of the 172 discrete episodes during this study period—have taken place in states that were simultaneously experiencing civil, ethnic, revolutionary wars or genocides. Factionalism need not end in violence to precipitate or reinforce violence. As with factional cleavage type, the data show some regional variation among onset and ending event types. While coup events were the most common way for factional episodes to end in sub-Saharan Africa, MENA countries, and Asia, they were extremely rare in Europe, with only one case ending in a coup. Elections, new governments, and liberalization events were more common in Europe, and in that region secession/fragmentation also was not uncommon, reflecting the strategies used by post-Soviet states, especially in and around the Balkans, to deal with ethnic and religious polarization. Elections and similar liberalization ending event types were particularly uncommon in the Middle East and North African episodes, with only one case ending with elections. Latin America, on the other hand, exhibits equal numbers of coup and election ending events, at 10 (29 percent) each, with generally equal numbers of similar liberalization and autocratization outcome event types. Ending event types also vary by factional identity type. Factionalism episodes characterized by primarily social identity cleavages, for example, end roughly equally commonly in either coups (25 percent) or general liberalization (24 percent), and are most likely of all faction types to be ended by warfare (7 percent). Elections as ending events, however, are relatively uncommon among factionalism episodes driven by social identity−based factions. Episodes characterized by personalistic and clientelistic factions exhibit a nearly identical frequency distribution. In contrast, episodes characterized by economic factional cleavages are the most likely to end in elections (35 percent), although here

Factionalism in the Post–World War II World

61

too coups (39 percent) are the most common outcome, and general autocratization events are not rare (15 percent). This may mean that economic cleavages are more easily addressed by electoral competition and its associated factional management strategies, but it could also reflect multicollinearity from the three-dimensional correlation among election ending event types, economic faction frequency, and the Latin American region. Interestingly, nearly all factional episodes characterized by issue- or policy-based cleavages ended in elections (20 percent) or liberalization events (40 percent), with only one coup and one example of general autocratization, perhaps because policy cleavages reflect less inherently intractable contention than that generated by social identity, personalistic, or economic cleavages. Hypothesis 1 generated by the SSP model in Chapter 1, and reviewed in the Appendix, predicts that cases with multiple types of cleavages would be less likely to experience peaceful resolutions of factionalism. Formally, as the number of parallel and reinforcing cleavages increases, the probability of achieving a peaceful resolution of the factionalism episode decreases. Comparing the relative ratios of ending event types, however, does not support this hypothesis. Of the 172 factionalism episodes between 1946 and 2015, only 28 are coded as having three or more types of factional cleavages. The frequency distribution of ending events among these cases is roughly similar to that of factionalism onsets as a whole, with about one-third of cases ending in elections or general liberation, another third ending in autocratization, and the remainder split among ongoing cases (about 21 percent), fragmentation or state collapse, and war. The lack of support for this hypothesis may be due to artifacts in the coding scheme, as coders were asked to identify the most prominent factional cleavage described in the country reports or news record and to record additional secondary and tertiary cleavage types when they were present. This may have led coders to not record multiple factional types in cases where factions reflected parallel cleavages, instead recording the most prominent faction type exhibited by the most prominent faction, and recording multiple cleavages only where they were cross-cutting instead of parallel. This hypothesis is also explored inductively in Chapters 3 through 6, as each case study includes an examination of factional identities and cleavages. The SSP model also predicts that factional cases should be more fragile, overall, than nonfactional cases, as the theoretical discussion in the Appendix explains in some detail. Formally, Hypothesis 2 in the Appendix states that if a state is factional, then it should be more fragile than nonfactional polities, on average. The State Fragility Index (SFI)

62

Democratization and the Mischief of Faction

measures fragility on a 0−25 scale, where 0 indicates very low fragility and 25 indicates extreme fragility (Marshall and Cole 2011, 2014). This index combines two major components, effectiveness and legitimacy, across four areas of state performance that the SSP model predicts should be affected by factionalism: economic, political, security, and social. The condition of factionalism contributes 1 point toward political legitimacy in the SFI and thus, all other things being equal, a null hypothesis would predict factional countries should have an average SFI score 1 point higher than nonfactional countries. A cross-sectional analysis of cases in 2015, however, demonstrates that the mean SFI score for factional cases (not including cases of collapsed central authority) was 12.28, indicating moderate fragility, more than 5 points higher than the average SFI score of 7.12 for nonfactional countries. This difference is substantial, at nearly one-fifth of the SFI scale. An independent samples t test also demonstrates that this mean difference of 5.16 points is highly statistically significant at the .001 level in a two-tailed test. These increased levels of fragility also endure over time: cases that were factional in 2005 were slightly more than 5 points more fragile on the SFI scale in 2015 compared to their nonfactional counterparts, and cases that were factional in 1995 were still 2 points more fragile 20 years later in 2015. Table 2.9 describes these results, which I have confirmed in multivariate regression models including multiple economic-, political-, and security-related control variables: factionalism has systemic and lasting effects on state fragility.

Table 2.9 Comparison of Mean Fragility Levels

Factional in 1995

Factional in 2005

Factional in 2015

Nonfactional Mean Factional Mean of 2015 SFI (SD) of 2015 SFI (SD) 7.54 (6.28)

6.87 (5.96)

7.12 (6.16)

9.97 (5.74)

12.11 (4.76)

12.28 (4.60)

Notes: SFI, State Fragility Index. a. Two-tailed test; equal variances assumed.

Mean Diff.

SE

t

p